System

Abstract

A system is provided.SOLUTION: A system including means for collecting stock turnover, sales data, and sales data of competitors' products from a data source, means for collecting social network service trend data, means for pre-processing the collected data and performing missing value compensation, data cleansing, and normalization to divide the data into training data and test data, means for building a generative artificial intelligence model based on the pre-processed data and predicting demand for products, means for calculating future demand fluctuation based on the prediction result and proposing an appropriate inventory quantity and order quantity, and means for displaying the proposal on a user interface and notifying the user.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In inventory management, there is a problem of overstocking or shortages due to inaccurate supply and demand forecasts. Conventional methods often reduce the accuracy of supply and demand forecasts because they rely on simple past data or only consider limited factors. The present invention aims to utilize generative artificial intelligence to integrate diverse data from multiple data sources, achieve highly accurate supply and demand forecasts, and optimize inventory management. [Means for solving the problem]

[0005] To solve this problem, the present invention provides an inventory management system having the following configuration. It includes means for collecting inventory turnover, sales data, sales data on competitors' products, and trend data on social network services from data sources. It also includes means for preprocessing the collected data, completing missing values, cleansing the data, normalizing it, and dividing it into training data and test data. It also includes means for building a generative artificial intelligence model based on the preprocessed data and forecasting product demand. It also includes means for calculating future demand fluctuations based on the forecast results and proposing appropriate inventory and order quantities, and means for displaying the proposal on a user interface and notifying the user. This enables highly accurate supply and demand forecasts, prevents overstocking and shortages, and realizes more efficient inventory management.

[0006] "Data source" refers to a source of information used to collect information necessary for supply and demand forecasting, including a company's ERP system, POS system, social network service (SNS) API, etc.

[0007] "Inventory turnover" is an indicator that shows the frequency of replacement of inventory items during a specific period, and is used to evaluate the efficiency of inventory management.

[0008] "Sales data" refers to information about the quantity and value of merchandise sold within a period of time, and serves as basic data for forecasting product demand.

[0009] "Sales data of competitors' products" refers to sales information on products of other companies that compete in the same market, and is used to improve the accuracy of supply and demand forecasts for one's own products.

[0010] "Social network service trend data" refers to data that shows trends in user-generated content and engagement on social networking sites and is used to understand social interests.

[0011] A "generative artificial intelligence model" is a data analysis model built using machine learning algorithms, and has the ability to predict future demand based on past information.

[0012] "Missing value imputation" is the process of filling in missing values ​​in a dataset with appropriate values ​​to maintain data integrity.

[0013] "Data cleansing" is the process of removing inaccurate or unnecessary information from a dataset, which is necessary to improve data quality.

[0014] "Normalization" is the process of adjusting data to a certain scale, allowing different ranges of data to be treated on a uniform basis.

[0015] "Training data" is the data set used to train a machine learning model, from which the model learns patterns.

[0016] "Test data" is a dataset used to evaluate the performance of a machine learning model and to confirm the model's generalization ability.

[0017] "Prediction results" are information about future supply and demand calculated using a generative artificial intelligence model, and are important for determining optimal product inventory and order quantities.

[0018] The "user interface" is an interface that allows the system and the user to communicate with each other and visually display prediction results and suggestions. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2]1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0020] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0021] First, the terms used in the following description will be explained.

[0022] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0023] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0024] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0025] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0026] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0027] [First embodiment]

[0028] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0029] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0030] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0031] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0032] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0033] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0034] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0035] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0036] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0037] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0038] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0039] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0040] This invention is a system called "Perfect Inventory Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[0041] Data collection

[0042] Server Operation

[0043] The server collects product inventory turnover, sales data, competitor product sales data, and SNS trend data from various data sources (e.g., the company's ERP system, POS system, social networking services). The data is monitored to ensure it is updated regularly, and new information is automatically incorporated.

[0044] Device behavior

[0045] The terminal provides an interface for users to operate. Users input data such as information on new products and promotional plans into the terminal, which adds the information necessary for prediction to the database.

[0046] Data Preprocessing

[0047] Server Operation

[0048] The server stores the collected data in a database, imputes missing values ​​in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes the data, and splits the dataset into training data and test data to build a highly accurate model.

[0049] Building a demand forecast model

[0050] Server Operation

[0051] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[0052] Running a prediction

[0053] Server Operation

[0054] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[0055] Output of results

[0056] Server Operation

[0057] The server stores the prediction results in a database, generates reports if necessary, calculates appropriate inventory and order quantities based on the prediction results, and makes suggestions to the user.

[0058] Device behavior

[0059] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[0060] Specific examples

[0061] Example 1: Demand forecast due to rising sales of competitors' products

[0062] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[0063] Example 2: Increased demand due to price reduction

[0064] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[0065] Based on these specific examples, this system can realize highly accurate supply and demand forecasts, prevent overstocks and shortages, and improve the efficiency of inventory management.

[0066] The processing flow will be explained below.

[0067] Step 1: Data collection

[0068] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[0069] Step 2: User Data Entry

[0070] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[0071] Step 3: Data Preprocessing

[0072] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[0073] Step 4: Splitting the Dataset

[0074] The server splits the preprocessed data into training data and test data, storing 80% of the data in the database as training data and 20% as test data.

[0075] Step 5: Feature Engineering

[0076] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates of social media trends, and prepares these features for input into the generative AI model.

[0077] Step 6: Building a generative AI model

[0078] The server uses machine learning libraries to build generative artificial intelligence models such as LSTM and Transformer. First, the model is trained using training data to learn patterns. After training is complete, the model is evaluated using test data to check accuracy and generalization performance.

[0079] Step 7: Forecast future demand

[0080] The server inputs new data into the generative AI model to predict future demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[0081] Step 8: Save and output prediction results

[0082] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[0083] Step 9: Display in the User Interface

[0084] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[0085] Step 10: User Actions

[0086] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocking or shortages.

[0087] Example 1

[0088] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0089] Current inventory management systems are prone to inventory shortages and surpluses due to inaccurate demand forecasts. In addition, inaccuracies in collected data affect the accuracy of forecasts, so there is a need for more efficient data preprocessing. Furthermore, it is difficult to forecast demand while taking into account competitor trends and social media trends, making strategic inventory management difficult. There is a need for a system that can solve these issues and achieve highly accurate inventory management.

[0090] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0091] In this invention, the server includes: [means for collecting inventory turnover, sales data, competitor product sales data, and social network trend data from multiple data sources;] [means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing it, and dividing it into training data and test data; and] [means for constructing a generative artificial intelligence model based on the preprocessed data and predicting product demand.] This enables [highly accurate demand forecasting and efficient inventory management].

[0092] A "data source" is a system or resource for obtaining data from various sources.

[0093] "Inventory turnover" is a value that indicates the number of times an inventory product is sold within a certain period of time.

[0094] "Sales data" is data that includes numerical information about product sales during a specific period.

[0095] "Sales data of competitors' products" refers to data that includes numerical information on sales of competitors' products that exist in the same market.

[0096] "Social network trend data" is data that includes information about trends and topics on online social networking services.

[0097] A "missing value" is a value that indicates a portion of a data set where information is missing.

[0098] "Data cleansing" is the process of removing inaccurate data and noise from a dataset to improve its accuracy.

[0099] "Normalization" is a process that unifies the scale of data to make it easier to compare data with different units or scales.

[0100] "Training data" is a data set used to train an artificial intelligence model.

[0101] "Test data" is a data set used to evaluate the performance of a trained artificial intelligence model.

[0102] A "generative artificial intelligence model" is a machine learning model built on collected and preprocessed data to perform a specific task (e.g., supply and demand forecasting).

[0103] "Demand forecasting" is the prediction of the demand for a product for a specific period in the future.

[0104] "Inventory quantity" refers to the number of products in a warehouse or on a storefront at a particular point in time.

[0105] The "order quantity" is the quantity of the product that should be newly ordered based on the demand forecast.

[0106] The "user interface" refers to the part that includes the means of interaction and screen display that allows the user to interact with the system.

[0107] "Notification" refers to messages or alerts sent by the system to convey information to the user.

[0108] This invention is a system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[0109] System Configuration

[0110] The system mainly consists of the following components:

[0111] 1. Server: Collects data, preprocesses data, builds models, executes predictions, and outputs results.

[0112] 2. Terminal: Provides the interface that the user operates and displays prediction results and suggestions.

[0113] 3. User: Operates the system, inputs the necessary data, and makes decisions based on the prediction results provided.

[0114] Hardware and Software

[0115] The server consists of server equipment with high-performance computing resources (e.g., workstations equipped with GPUs) and uses the following software:

[0116] Operating System: Linux

[0117] Database: MySQL or PostgreSQL

[0118] Programming language: Python

[0119] Machine learning libraries: TensorFlow, PyTorch

[0120] Data processing libraries: pandas, numpy

[0121] The devices consist of common PCs, tablets, smartphones, and other devices, and use the following software:

[0122] Web browser: Google Chrome, Firefox

[0123] Web application framework: Django or Flask

[0124] Data collection

[0125] Server Operation

[0126] The server collects data from multiple data sources, including the company's ERP system, POS system, and social media. Specifically, it uses SQL queries to retrieve inventory turnover and sales data from the ERP system and pulls sales data from the POS system via API. Social media data is collected by using APIs from Twitter and Facebook. All collected data is integrated into a database.

[0127] Data Preprocessing

[0128] Server Operation

[0129] The server performs data preprocessing on the collected data. It imputes missing values ​​with the mean or median, detects and removes outliers, and performs normalization processes such as standardizing date formats and converting data types. It also splits the dataset into training data and test data.

[0130] Building a demand forecast model

[0131] Server Operation

[0132] The server uses machine learning libraries such as TensorFlow and PyTorch to build generative AI models. It performs feature engineering to define features suitable for model input. It trains the model using algorithms suitable for time-series data, such as LSTM and Transformer. After training, it evaluates the model's performance and tunes hyperparameters as necessary.

[0133] Running a prediction

[0134] Server Operation

[0135] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[0136] Output of results

[0137] Server Operation

[0138] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[0139] Device behavior

[0140] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[0141] Specific examples

[0142] Example 1: Demand forecast due to rising sales of competitors' products

[0143] The server collects sales data and social media trend data from other companies' similar products, performs data preprocessing, and then uses a generative AI model to predict that demand for Product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[0144] Example 2: Increased demand due to price reduction

[0145] The server collects promotion information from the POS system and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[0146] This allows users to manage their inventory appropriately and prevent shortages and surpluses.

[0147] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0148] Step 1: Data collection

[0149] Server Operation

[0150] The server collects the necessary data from multiple data sources (e.g., the company's ERP system, POS system, and social media). Specifically, it uses SQL queries to obtain inventory turnover and sales data from the ERP system, and sales data from the POS system via API. Social media data is collected through APIs from Twitter and Facebook.

[0151] Specific operations and inputs / outputs

[0152] Input: SQL query, API request

[0153] Data manipulation: Pulling data from ERP systems, getting data from POS systems, collecting data from social media APIs

[0154] Output: Consolidated dataset in a database

[0155] Step 2: Data Preprocessing

[0156] Server Operation

[0157] The server performs data preprocessing on the collected data. Specifically, it imputes missing values ​​with the mean or median, and detects and removes outliers. It also performs normalization processes such as standardizing date formats and converting data types. Next, it splits the dataset into training data and test data.

[0158] Specific operations and inputs / outputs

[0159] Input: Integrated dataset

[0160] Data manipulation: Imputing missing values ​​(using Python's pandas library), removing outliers, normalizing (scaling) the data, and splitting the data (using scikit-learn's train_test_split function).

[0161] Output: Preprocessed training and test data

[0162] Step 3: Build a demand forecast model

[0163] Server Operation

[0164] The server uses machine learning libraries (TensorFlow, PyTorch) to build a generative AI model. First, feature engineering is performed to define features suitable for the model. Then, the model is trained using algorithms specialized for time-series data, such as LSTM and Transformer, and the model's performance is evaluated. Hyperparameter tuning is also performed as necessary.

[0165] Specific operations and inputs / outputs

[0166] Input: Training data

[0167] Data manipulation: feature engineering (creating new features), model training (building an LSTM model using TensorFlow), model evaluation (calculating precision, recall, and F-measure)

[0168] Output: A trained generative AI model

[0169] Step 4: Run a prediction

[0170] Server Operation

[0171] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[0172] Specific operations and inputs / outputs

[0173] Input: Newly collected data, trained generative AI model

[0174] Data manipulation: Inputting data into the model, running demand forecasts (predicting future demand through the model)

[0175] Output: Forecast result (future demand curve)

[0176] Step 5: Output the results

[0177] Server Operation

[0178] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[0179] Specific operations and inputs / outputs

[0180] Input: Prediction result

[0181] Data manipulation: Prediction results saved in database, report generation (automatic PDF report generation)

[0182] Output: Report, proposals

[0183] Device behavior

[0184] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[0185] Specific operations and inputs / outputs

[0186] Input: Prediction results and suggestions sent from the server

[0187] Data manipulation: visualization (display in graphs and dashboards)

[0188] Output: Display on the user interface, notifications

[0189] The above processing steps enable highly accurate supply and demand forecasts and efficient inventory management.

[0190] (Application example 1)

[0191] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0192] Product demand forecasts are extremely important in modern inventory management, but achieving highly accurate forecasts remains difficult. In particular, it is difficult to integrate information from diverse data sources and perform detailed analysis of factors that cause demand fluctuations, and the uncertainty of forecasts is likely to have a negative impact on corporate management. Furthermore, while there is a demand for timely supply and demand forecasts that reflect promotions and price reduction information, there is a lack of technology to achieve this. Therefore, there is a need for a system that can solve these problems and achieve highly accurate supply and demand forecasts and inventory management.

[0193] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0194] In this invention, the server includes: means for collecting inventory turnover, sales data, sales data on competitors' products, and trend data on social network services from data sources; means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing the data, and dividing it into training data and test data; means for constructing a generative artificial intelligence model based on the preprocessed data and forecasting product demand; means for calculating future demand fluctuations based on the forecast results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; and means for providing the user with sales promotion information and making supply and demand forecasts based on that information. This enables companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[0195] "Data Source" refers to an information source that provides data necessary for supply and demand forecasting, including inventory turns, sales data, sales data of competitors' products, and trend data of social network services.

[0196] "Inventory turnover" is an index that indicates the frequency of product inventory replacement within a certain period of time.

[0197] "Sales data" refers to data relating to the number of products sold and the amount of sales within a certain period of time.

[0198] "Sales data of competitors' products" refers to data on the sales performance of competitors' products that are similar to or belong to the same market as our company's products.

[0199] "Social network service trend data" refers to data on popular topics and topics of interest extracted from user activities and posts on platforms such as social networking sites.

[0200] "Missing value imputation" is the process of filling in missing values ​​in a dataset in an appropriate way.

[0201] "Data cleansing" is the process of removing inaccurate or unnecessary information from a data set.

[0202] "Normalization" is a preprocessing process that improves the learning effect of an algorithm by constraining data within a certain range.

[0203] "Training data" is the dataset used to train a machine learning model.

[0204] "Test data" is a dataset used to evaluate the performance of a machine learning model.

[0205] A "generative artificial intelligence model" is a supply and demand forecasting model built using machine learning algorithms based on collected and preprocessed data.

[0206] "Prediction results" are data and information regarding future demand calculated by the generated artificial intelligence model.

[0207] "Future demand fluctuations" refers to trends that indicate future increases or decreases in demand based on forecast results.

[0208] "Inventory" is the quantity of a product in stock at a particular point in time.

[0209] "Order quantity" refers to the quantity of the product to be newly ordered to meet future demand.

[0210] A "user interface" is a screen or operating means through which a system interacts with a user.

[0211] "Sales promotion information" is information about measures to increase sales, such as product promotions and price reductions.

[0212] This invention is a "Perfect Inventory Manager" system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users. The details of the system's operation are described below.

[0213] Data collection

[0214] Server Operation

[0215] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from various data sources, such as the company's ERP system, POS system, and social networking API platforms. The data is monitored for regular updates, and new information is automatically incorporated.

[0216] User Actions

[0217] Users can input data such as sales promotion information, new product information, and promotion schedules through their devices, and the information provided by users is also used to improve the accuracy of the prediction model.

[0218] Data Preprocessing

[0219] Server Operation

[0220] The server stores the collected data in a database and imputes missing values ​​in an appropriate manner. It also performs data cleansing to remove inaccurate data and noise and normalizes the data. Furthermore, it splits the dataset into training data and test data to build a highly accurate predictive model. This process utilizes data integration tools and the Python pandas library.

[0221] Building a demand forecast model

[0222] Server Operation

[0223] The server builds a generative AI model using machine learning libraries such as TensorFlow and PyTorch. It performs feature engineering and inputs appropriate features into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[0224] Running a prediction

[0225] Server Operation

[0226] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are analyzed. At the same time, factors behind demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[0227] Output of results

[0228] Device behavior

[0229] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, prompting the user to take appropriate action. For example, the terminal notifies the user of suggested inventory levels and order quantities.

[0230] Specific examples

[0231] When a user inputs new promotional information into the device, the system predicts changes in demand based on that information. For example, if the user inputs information like "20% off all products in the Christmas sale starting December 1, 2023," the generative AI model predicts an increase in sales for the next month and alerts the user that "demand will increase next month, so you should increase your inventory."

[0232] Prompt Sentence Examples

[0233] New promotion information:

[0234] Event: Christmas Sale

[0235] Start date: December 1, 2023

[0236] Discount: 20%

[0237] Predict the percentage increase in sales that this promotion will bring

[0238] This allows companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[0239] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0240] Step 1: Data collection

[0241] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources (e.g., a company's ERP system, POS system, or SNS API). The collected data is sent to the server in JSON or CSV format and stored in a database. This provides a variety of data necessary for demand forecasting.

[0242] Input: ERP system, POS system, SNS API, etc.

[0243] Output: Inventory turnover, sales data, competitor product sales data, SNS trend data (stored in database)

[0244] Step 2: Data Preprocessing

[0245] The server imputes missing values ​​in the collected data in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes (scales) the data to the range of 0 to 1, and splits it into training data and test data. This process uses Python's pandas library and scikit-learn's MinMaxScaler.

[0246] Input: Collected raw data (taken from database)

[0247] Output: Preprocessed data (scaled data after missing value imputation and data cleansing)

[0248] Step 3: Building a generative AI model

[0249] The server uses the preprocessed data to build a generative AI model. It uses TensorFlow and PyTorch to train a time-series prediction model using algorithms such as LSTM and Transformer. It separates the training data from the test data to evaluate the model's accuracy and tune it accordingly.

[0250] Input: Preprocessed data (training data and test data)

[0251] Output: Highly accurate generative AI model (supply and demand forecast model)

[0252] Step 4: Run a prediction

[0253] The server inputs the newly collected data into the generative AI model to predict future product demand. The prediction results are generated as data showing future demand fluctuations, and also generate detailed analysis results of the factors behind demand fluctuations, taking into account sales data of competitors' products and social media trend data.

[0254] Input: New data (latest data from data collection step)

[0255] Output: Forecast results (future demand fluctuation data and analysis results of demand fluctuation factors)

[0256] Step 5: Output the results

[0257] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, allowing the user to take appropriate action based on the results. For example, the terminal may notify the user of suggested inventory levels or order quantities.

[0258] Input: Prediction results and suggestions (data sent from the server)

[0259] Output: Data displayed in a user interface in the form of graphs and dashboards

[0260] Step 6: Enter promotion information

[0261] Users input sales promotion information (e.g., promotions and price reductions) through their terminals. This information is sent to the server and stored in a database, which is then used to forecast supply and demand for future purchases.

[0262] Input: Sales promotion information (promotions and price reduction information)

[0263] Output: Saved promotion information (stored in database)

[0264] Step 7: Promotion-based forecasting

[0265] The server predicts supply and demand based on sales promotion information entered by the user. For example, it predicts an increase in demand after a Christmas sale based on information about the sale. This prediction result is also sent to the terminal and notified to the user.

[0266] Input: Sales promotion information (obtained from database)

[0267] Output: Prediction results and proposals that take into account sales promotion information (sent from the server to the device)

[0268] Through these steps, the system achieves highly accurate supply and demand forecasts and inventory management.

[0269] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0270] This invention is a "Perfect Inventory Manager" system that uses generative AI to perform highly accurate supply and demand forecasts in inventory management, and by combining it with an emotion engine that recognizes user emotions, it improves the effectiveness of proposals and user satisfaction. This system operates in cooperation with the server, terminals, and users.

[0271] Data collection

[0272] Server Operation

[0273] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social network service APIs. The data is updated regularly and is set to be automatically imported.

[0274] Device behavior

[0275] The device provides an interface for users to operate. Users input information about new products and upcoming promotions into the device. This information is also stored in the database and used to improve the accuracy of the prediction model.

[0276] Data Preprocessing

[0277] Server Operation

[0278] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[0279] Building a demand forecast model

[0280] Server Operation

[0281] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It then trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[0282] Future demand forecast

[0283] Server Operation

[0284] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[0285] Saving and outputting prediction results

[0286] Server Operation

[0287] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[0288] Emotion Recognition and Feedback

[0289] Device behavior

[0290] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and text message content when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server.

[0291] Server Operation

[0292] The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the content and timing of the suggestions, and reflects the user's emotional feedback in optimizing the system.

[0293] Display on the user interface

[0294] Device behavior

[0295] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[0296] Specific examples

[0297] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[0298] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the device of this result and issues an alert to the user that "demand will increase next month, so you should increase your inventory." The device uses an emotion engine to collect the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional feedback, evaluates whether the user is satisfied, and reflects this in the next proposal.

[0299] Example 2: Increased demand due to price reduction and emotional feedback

[0300] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased." The terminal collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[0301] Based on these specific examples, this system can achieve highly accurate supply and demand forecasts, and by taking user emotions into consideration, it can improve the effectiveness of its proposals. As a result, it is possible to prevent overstocks and shortages and improve the efficiency of inventory management.

[0302] The processing flow will be explained below.

[0303] Step 1: Data collection

[0304] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[0305] Step 2: User Data Entry

[0306] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[0307] Step 3: Data Preprocessing

[0308] The server preprocesses the collected data. First, it performs imputation processing using a method to impute missing values ​​using the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and arranges it into a unified format.

[0309] Step 4: Splitting the Dataset

[0310] The server splits the preprocessed data into training data and test data. Specifically, 80% of the data is stored in the database as training data and 20% as test data.

[0311] Step 5: Feature Engineering

[0312] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates for social media trends, and prepares the selected features for input into the generative AI model.

[0313] Step 6: Building a generative AI model

[0314] The server uses machine learning libraries to build generative AI models such as LSTM and Transformer. First, it trains the model using training data to learn patterns. After training is complete, it evaluates the model using test data to check its accuracy and generalization performance.

[0315] Step 7: Forecast future demand

[0316] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[0317] Step 8: Save and output prediction results

[0318] The server stores the forecast results and analysis reports in a database, and calculates appropriate inventory and order quantities based on the stored results and makes recommendations to the user.

[0319] Step 9: Emotion Recognition and Feedback

[0320] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and the content of text messages when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server. The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. The server adjusts the suggestion content and timing according to the level of satisfaction, and reflects the user's emotional feedback in optimizing the system.

[0321] Step 10: Display in the User Interface

[0322] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[0323] Step 11: User Actions

[0324] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocks and shortages.

[0325] Example 2

[0326] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0327] In inventory management, not only is it difficult to make accurate supply and demand forecasts, but the inability to properly consider user emotions and feedback reduces the effectiveness of proposals and user satisfaction.In addition, the inability to fully utilize external data such as the trends of competitors and social media trends prevents the accuracy of supply and demand forecasts from improving.

[0328] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for collecting inventory turnover, sales data, sales data of competitors, and trend information on social network services from information sources; means for preprocessing the collected data, filling in missing data, cleaning the data, standardizing the data, and dividing it into training data and test data; means for constructing a generative AI model based on the preprocessed data and predicting product demand; means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; means for recognizing emotions from user operations, voice, and text messages on the terminal and transmitting the data to the server; means for analyzing the emotion data transmitted to the server and determining user satisfaction; and means for optimizing the content and timing of proposals based on user satisfaction. This not only enables highly accurate supply and demand predictions that take into account the trends of competitors and social networking services trends, but also enables proposals that reflect user emotions, thereby improving inventory management efficiency and user satisfaction.

[0329] The "information source" is a data supplier that provides a variety of data such as inventory turnover, sales data, sales data of competitors, and trend information of social networking services.

[0330] "Inventory turnover" is an indicator of the number of times inventory items are turned over within a certain period of time. It is typically used to evaluate how quickly inventory is sold.

[0331] "Sales data" refers to data that includes information such as the quantity, price, and sales period of products and services sold.

[0332] "Competitor sales data" refers to data on the sales of competitors' products and services in the market. This data is used to understand the competitive landscape and to forecast supply and demand.

[0333] "Social network service trend information" refers to information about popular topics, products, and services on social media. This allows us to understand consumer trends and market trends.

[0334] "Missing data completion" is a process of completing missing values ​​in a dataset by filling in the average value or previous or next values.

[0335] "Data cleaning" is the process of removing outliers and noise from a dataset to improve data quality.

[0336] "Data standardization" is the process of converting data into a unified format to make it consistent, which makes it easier to analyze and train models.

[0337] A "generative artificial intelligence model" is a model that uses machine learning algorithms to predict future situations based on past data. It is also called a generative AI model.

[0338] "User interface" refers to the screens and operating methods that allow users to interact with the system, input information, and check results.

[0339] "Emotion recognition" is the process by which the device analyzes emotions from the user's voice or text messages and determines whether the emotion is positive or negative.

[0340] "Satisfaction" is an index that indicates the degree of satisfaction that a user feels with the services or proposals provided.

[0341] Optimization is the adjustment and improvement of systems and processes to make them operate most efficiently and effectively.

[0342] This invention is a system for implementing highly accurate inventory management for a company. This system operates in cooperation with a server, terminals, and users. The hardware and software used to build the system of this invention include the following elements:

[0343] Hardware and Software

[0344] server

[0345] The server collects data, preprocesses it, builds generative AI models, and stores the prediction results. The server uses the following software:

[0346] Data collection: HTTP client for API communication (e.g., requests library)

[0347] Data preprocessing: Python's pandas and scikit-learn libraries

[0348] Building generative AI models: TensorFlow or PyTorch

[0349] Terminal

[0350] The device provides a user interface and performs emotion recognition. Specifically, it uses the following software:

[0351] User Interface: React.js or Vue.js

[0352] Emotion Recognition: IBM Watson's Natural Language Understanding API

[0353] User

[0354] The user operates the terminal to input the data necessary for inventory management and receives suggestions from the system.

[0355] Specific processing of the system

[0356] Data collection

[0357] The server periodically retrieves inventory turnover, sales data, competitor sales data, and trend information from social networking services from information sources, thereby incorporating specific data that reflects the latest market trends into the system.

[0358] Data Preprocessing

[0359] The collected data undergoes preprocessing, including missing data imputation, data cleaning, and data standardization, to transform it into a consistent dataset, which is then ready for model training.

[0360] Building generative AI models

[0361] Based on the preprocessed data, the server builds a generative AI model using TensorFlow or PyTorch. It uses algorithms specialized for time series data, such as LSTM and Transformer, to predict product demand. After training is complete, the model is evaluated using test data.

[0362] Demand forecasting and proposals

[0363] The server inputs newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are calculated and appropriate inventory and order quantities are calculated. The prediction results are notified to the user via the user interface.

[0364] Emotion recognition and feedback collection

[0365] The device uses an emotion engine to analyze the tone of voice and text message content when the user confirms the proposal, and sends the user's emotional data to the server. The server analyzes the emotional data, determines the user's level of satisfaction, and optimizes the content and timing of the next proposal.

[0366] Specific examples

[0367] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[0368] The server collects sales data and social media trend data from other companies' products and performs data preprocessing. Using a generative AI model, it predicts that demand for the product will increase by 50% next month. This result is notified to the device, and an alert is issued stating, "Demand will increase next month, so you should increase your inventory." The device then uses an emotion engine to collect the user's emotional feedback on this suggestion and sends it to the server. The server analyzes the emotional feedback and reflects it in the next suggestion.

[0369] Example 2: Increased demand due to price reduction and emotional feedback

[0370] The server collects promotional information and obtains information that the product price will be reduced by 10%. Based on this data, it uses a generative AI model to predict that demand will increase by 30% after the price reduction. The server sends the prediction result to the device and provides a notification that "product demand will increase after the price reduction, so production should be increased." The device collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[0371] As described above, the system of the present invention is capable of highly accurate supply and demand forecasting and proposals that take into account the user's emotions, preventing overstocks and shortages and improving the efficiency of inventory management.

[0372] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0373] Step 1: Data collection

[0374] Server Operation

[0375] 1. The server collects data from information sources via API communication. Specifically, it obtains inventory turnover and sales data from the ERP system, real-time sales data from the POS system, and trend information from the API of social media.

[0376] 2. Data inputs include API endpoints and authentication information for each information source, while outputs include collected inventory turnover, sales data, competitor sales data, and social media trend information.

[0377] Specific actions

[0378] The server uses the Python requests library to collect data from each information source, for example by sending an API request like this:

[0379] python

[0380] import requests

[0381] response = requests.get('https: / / api.example.com / inventory')

[0382] data = response.json()

[0383] Step 2: Data Preprocessing

[0384] Server Operation

[0385] 1. Store the collected data in a database.

[0386] 2. The server uses SQL queries to extract missing values ​​from the data and impute them with the average or surrounding values.

[0387] 3. The server performs data cleansing to remove outliers and noise.

[0388] 4. Finally, standardize the data to create a consistent dataset.

[0389] Input and Output

[0390] The input includes the raw data collected in step 1. The output is the formatted data that has been imputed and cleaned.

[0391] Specific actions

[0392] The server uses the Python pandas library to impute and clean the data, for example imputing missing values ​​as follows:

[0393] python

[0394] import pandas as pd

[0395] df = pd.DataFrame(data)

[0396] df.fillna(df.mean(), inplace=True)

[0397] Step 3: Building a generative AI model

[0398] Server Operation

[0399] 1. Using the preprocessed data, we build a generative AI model. Specifically, we define features for the time series data we will use.

[0400] 2. The server uses TensorFlow or PyTorch to train the model using algorithms such as LSTM or Transformer.

[0401] 3. After training, the model is evaluated using test data to check accuracy and generalization performance.

[0402] Input and Output

[0403] The input is a pre-processed dataset, and the output is a trained generative AI model.

[0404] Specific actions

[0405] Perform feature engineering and train an AI model. For example, train an LSTM model with the following code:

[0406] python

[0407] import tensorflow as tf

[0408] from tensorflow.keras.models import Sequential

[0409] from tensorflow.keras.layers import LSTM, Dense

[0410] model = Sequential()

[0411] model.add(LSTM(50, return_sequences=True, input_shape=(n_timesteps, n_features)))

[0412] model.add(LSTM(50))

[0413] model.add(Dense(1))

[0414] model.compile(optimizer='adam', loss='mse')

[0415] model.fit(train_X, train_y, epochs=50, batch_size=72, validation_data=(test_X, test_y))

[0416] Step 4: Forecast future demand

[0417] Server Operation

[0418] 1. Newly collected data is fed into a generative AI model to predict future product demand.

[0419] 2. Based on the demand forecast, a future demand curve is generated and factors affecting demand are analyzed taking into account sales data of competitors and social media trend information.

[0420] Input and Output

[0421] The inputs include new data collected in real time, and the outputs are future demand forecasts and analysis of the factors that cause fluctuations.

[0422] Specific actions

[0423] The server inputs the new data into the model and uses the model's forecasting capabilities to calculate demand. The model forecasts as follows:

[0424] python

[0425] future_predictions = model.predict(new_data)

[0426] Step 5: Save and output prediction results

[0427] Server Operation

[0428] 1. Save the prediction results and analysis reports in a database.

[0429] 2. The server calculates the appropriate inventory and order quantities based on the prediction results and generates a notification to suggest to the user.

[0430] Input and Output

[0431] The input includes future demand forecast results. The output is a saved forecast and a proposed notification.

[0432] Specific actions

[0433] The server stores the prediction results in a MySQL database and generates notifications to send to the device, for example using the following SQL query:

[0434] sql

[0435] INSERT INTO predictions (date, product_id, predicted_demand) VALUES (%s, %s, %s)

[0436] Step 6: Emotion recognition and feedback collection

[0437] Device behavior

[0438] 1. The device analyzes the user's operations, voice, and text messages using an emotion engine and sends the emotion data to the server.

[0439] Server Operation

[0440] 1. The server analyzes the transmitted emotion data and determines the user's satisfaction level.

[0441] 2. Optimize the content and timing of proposals based on customer satisfaction.

[0442] Input and Output

[0443] The input includes user emotion data sent from the device, and the output is a satisfaction analysis result and optimized recommendations.

[0444] Specific actions

[0445] The device uses IBM Watson API to analyze emotions and send the results to the server. For example, call the API as follows:

[0446] python

[0447] from ibm_watson import NaturalLanguageUnderstandingV1

[0448] from ibm_watson.natural_language_understanding_v1 import Features, EmotionOptions

[0449] nlu = NaturalLanguageUnderstandingV1(version='2021-03-25')

[0450] response = nlu.analyze(text='"user input text"', features=Features(emotion=EmotionOptions())).get_result()

[0451] Step 7: Display in the user interface

[0452] Device behavior

[0453] 1. Visually display prediction results and suggestions to the user.

[0454] 2. Present information in an easy-to-understand manner using graphs and dashboards, encouraging users to take appropriate action.

[0455] Input and Output

[0456] The input includes prediction results and suggestion notifications received from the server, and the output is visualized information provided to the user.

[0457] Specific actions

[0458] The terminal uses the React.js or Vue.js framework to display the prediction results in a dashboard format. For example, the display process can be done with the following code:

[0459] javascript

[0460] import React from 'react';

[0461] import { LineChart, Line, CartesianGrid, XAxis, YAxis, Tooltip} from 'recharts';

[0462] const data = [...] / / received prediction data

[0463] const Dashboard = () => (

[0464] <linechart width="{600}" height="{300}" data="{data}">

[0465] <line type="monotone" dataKey="predicted_demand" stroke="8884d8" / >

[0466] <cartesiangrid stroke="ccc" / >

[0467] <xaxis dataKey="date" / >

[0468] <yaxis / >

[0469] <tooltip / >

[0470] < / linechart>

[0471] );

[0472] export default Dashboard;

[0473] Through the above processing steps, companies can realize highly accurate supply and demand forecasts and proposals that take user emotions into consideration.

[0474] (Application example 2)

[0475] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0476] The accuracy of supply and demand forecasts in inventory management is low, which leads to a high risk of overstocking and shortages. In addition, there is a lack of proposals that take user emotions into account, so a method to improve user satisfaction is also needed.

[0477] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: [means for collecting inventory turnover, sales data, sales data of competitors' products, and trend data of social network services from data sources;] [means for preprocessing the collected data, imputing missing values, data cleansing, normalizing, and dividing the collected data into training data and test data;] [means for constructing a generative algorithm model based on the preprocessed data and predicting product demand;] [means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities;] [means for displaying the proposal on a user interface and notifying the user;] [means for analyzing user emotions using an emotion recognition engine and collecting feedback; and [means for optimizing the prediction model based on the feedback and improving the effectiveness of the proposal content.] This enables highly accurate supply and demand prediction and improved user satisfaction.

[0478] "Data Source" refers to information sources that provide inventory turns, sales data, competitor product sales data, and SNS trend data, such as a company's ERP system, POS system, and social network service API.

[0479] "Preprocessing" refers to the process of filling in missing values, cleansing, and normalizing the collected data, and then dividing the data into training data and test data.

[0480] A "generative algorithm model" is a predictive model that is constructed using machine learning algorithms using pre-processed data.

[0481] "Supply and demand forecasting" is a technique that uses generative algorithm models to predict future demand for a product.

[0482] The "user interface" is an interface for visually displaying to the user the proposals and prediction results presented by the server.

[0483] An "emotion recognition engine" is a technology that recognizes and analyzes a user's emotions from the tone of their voice and the content of their text messages.

[0484] "Feedback" refers to data collected about users' reactions and emotions to the proposals, and is used to optimize the predictive model.

[0485] This invention is a system called "Inventory Perfect Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[0486] Data collection

[0487] Server Operation

[0488] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from data sources, utilizing the company's ERP system, POS system, and SNS API. The data is updated periodically and automatically imported.

[0489] Device behavior

[0490] The device provides a user interface, and users input information about new products and promotional plans into the device. The device is also equipped with an emotion recognition engine that recognizes emotions from the tone of voice and content of text messages generated during user operations. This information is also stored in a database and used to improve the accuracy of the predictive model.

[0491] Data Preprocessing

[0492] Server Operation

[0493] The server preprocesses the collected data: missing values ​​are filled with the average or surrounding values, and inaccurate data and noise are removed through data cleansing. The data is then normalized and organized into a unified format.

[0494] Building a supply and demand forecast model

[0495] Server Operation

[0496] The server uses a machine learning library (e.g., TensorFlow, PyTorch) to build a generative algorithm model. Feature engineering is performed to define appropriate features and input them into the model. The model is trained using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[0497] Future demand forecast

[0498] Server Operation

[0499] The server inputs the newly collected data into a generative algorithm model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and trend data from social networking services.

[0500] Saving and outputting prediction results

[0501] Server Operation

[0502] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[0503] Emotion Recognition and Feedback

[0504] Device behavior

[0505] The device is equipped with an emotion recognition engine that recognizes the user's emotions from the tone of voice when the user operates the device and the content of text messages. The emotion recognition engine collects the user's emotion data and sends it to the server.

[0506] Server Operation

[0507] The server analyzes the emotion data sent from the emotion recognition engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the suggestion content and timing, and reflects the user's emotional feedback in optimizing the system.

[0508] Display on the user interface

[0509] Device behavior

[0510] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[0511] Specific examples

[0512] For example, suppose a new product D is scheduled to be released next month. The server collects existing sales data, competitor data, and social media trend information, and uses a generative AI model to predict supply and demand. As a result, the predictive model predicts that demand for product D will increase by 40%. Using emotion recognition feedback, the store manager receives positive feedback that the prediction is "very reliable."

[0513] Prompt Sentence Examples

[0514] The supply and demand forecast for next month shows a 40% increase in demand for new product D. Please arrange the necessary inventory based on this forecast. You can provide feedback, which will be evaluated by the sentiment recognition engine. For example, you can enter feedback such as "The forecast is very reliable."

[0515] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0516] Step 1:

[0517] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources. The input is data from ERP systems, POS systems, SNS APIs, etc., which is collected periodically and stored in a database. The output is the various collected data.

[0518] Step 2:

[0519] The server preprocesses the collected data. Specifically, it imputes missing values ​​with the average or surrounding values, and removes inaccurate data and noise through data cleansing. Next, it normalizes the data, organizes it into a unified format, and splits it into training data and test data. The input is the data collected in step 1, and the output is the preprocessed training data and test data.

[0520] Step 3:

[0521] The server uses machine learning libraries to build a generative algorithm model. Specifically, it uses TensorFlow or PyTorch to extract features from the preprocessed data and trains the model using algorithms such as LSTM or Transformer. The input is the preprocessed data from step 2, and the output is the trained generative algorithm model.

[0522] Step 4:

[0523] The server inputs new data into the generative algorithm model to predict future product supply and demand. The inputs are the newly collected data and the trained model, and the output is the future demand forecast.

[0524] Step 5:

[0525] The server calculates future demand fluctuations based on the forecast results and proposes appropriate inventory and order quantities. The input is the supply and demand forecast results obtained in step 4, and the output is the proposed inventory and order quantities.

[0526] Step 6:

[0527] The terminal displays the prediction results and suggestions sent from the server on a user interface. The input is the prediction results and suggestions sent from the server, and the output is a visual display on the user interface.

[0528] Step 7:

[0529] The user operates the device to input information about new products and promotion plans. Emotional data is also collected from the tone of voice and the content of text messages while operating the device. The input is the user's new product information and emotional data, and the output is the collected data.

[0530] Step 8:

[0531] The emotion recognition engine analyzes the user's emotion data and sends it to the server. The input is the emotion data sent from the device, and the output is the analyzed emotion information.

[0532] Step 9:

[0533] The server analyzes the emotion data sent from the emotion recognition engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the suggestion content and timing and reflects this in the next prediction. The input is the analyzed emotion data, and the output is the adjusted suggestion content.

[0534] Step 10:

[0535] The server finally optimizes the prediction model based on the prediction results and user feedback to improve the effectiveness of the recommendations. The input is the feedback data and the prediction results, and the output is the optimized prediction model.

[0536] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0537] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0538] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0539] [Second embodiment]

[0540] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0541] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0542] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0543] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0544] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0545] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0546] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0547] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0548] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0549] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0550] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0551] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0552] This invention is a system called "Perfect Inventory Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[0553] Data collection

[0554] Server Operation

[0555] The server collects product inventory turnover, sales data, competitor product sales data, and SNS trend data from various data sources (e.g., the company's ERP system, POS system, social networking services). The data is monitored to ensure it is updated regularly, and new information is automatically incorporated.

[0556] Device behavior

[0557] The terminal provides an interface for users to operate. Users input data such as information on new products and promotional plans into the terminal, which adds the information necessary for prediction to the database.

[0558] Data Preprocessing

[0559] Server Operation

[0560] The server stores the collected data in a database, imputes missing values ​​in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes the data, and splits the dataset into training data and test data to build a highly accurate model.

[0561] Building a demand forecast model

[0562] Server Operation

[0563] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[0564] Running a prediction

[0565] Server Operation

[0566] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[0567] Output of results

[0568] Server Operation

[0569] The server stores the prediction results in a database, generates reports if necessary, calculates appropriate inventory and order quantities based on the prediction results, and makes suggestions to the user.

[0570] Device behavior

[0571] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[0572] Specific examples

[0573] Example 1: Demand forecast due to rising sales of competitors' products

[0574] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[0575] Example 2: Increased demand due to price reduction

[0576] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[0577] Based on these specific examples, this system can realize highly accurate supply and demand forecasts, prevent overstocks and shortages, and improve the efficiency of inventory management.

[0578] The processing flow will be explained below.

[0579] Step 1: Data collection

[0580] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[0581] Step 2: User Data Entry

[0582] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[0583] Step 3: Data Preprocessing

[0584] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[0585] Step 4: Splitting the Dataset

[0586] The server splits the preprocessed data into training data and test data, storing 80% of the data in the database as training data and 20% as test data.

[0587] Step 5: Feature Engineering

[0588] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates of social media trends, and prepares these features for input into the generative AI model.

[0589] Step 6: Building a generative AI model

[0590] The server uses machine learning libraries to build generative artificial intelligence models such as LSTM and Transformer. First, the model is trained using training data to learn patterns. After training is complete, the model is evaluated using test data to check accuracy and generalization performance.

[0591] Step 7: Forecast future demand

[0592] The server inputs new data into the generative AI model to predict future demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[0593] Step 8: Save and output prediction results

[0594] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[0595] Step 9: Display in the User Interface

[0596] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[0597] Step 10: User Actions

[0598] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocking or shortages.

[0599] Example 1

[0600] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0601] Current inventory management systems are prone to inventory shortages and surpluses due to inaccurate demand forecasts. In addition, inaccuracies in collected data affect the accuracy of forecasts, so there is a need for more efficient data preprocessing. Furthermore, it is difficult to forecast demand while taking into account competitor trends and social media trends, making strategic inventory management difficult. There is a need for a system that can solve these issues and achieve highly accurate inventory management.

[0602] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0603] In this invention, the server includes: [means for collecting inventory turnover, sales data, competitor product sales data, and social network trend data from multiple data sources;] [means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing it, and dividing it into training data and test data; and] [means for constructing a generative artificial intelligence model based on the preprocessed data and predicting product demand.] This enables [highly accurate demand forecasting and efficient inventory management].

[0604] A "data source" is a system or resource for obtaining data from various sources.

[0605] "Inventory turnover" is a value that indicates the number of times an inventory product is sold within a certain period of time.

[0606] "Sales data" is data that includes numerical information about product sales during a specific period.

[0607] "Sales data of competitors' products" refers to data that includes numerical information on sales of competitors' products that exist in the same market.

[0608] "Social network trend data" is data that includes information about trends and topics on online social networking services.

[0609] A "missing value" is a value that indicates a portion of a data set where information is missing.

[0610] "Data cleansing" is the process of removing inaccurate data and noise from a dataset to improve its accuracy.

[0611] "Normalization" is a process that unifies the scale of data to make it easier to compare data with different units or scales.

[0612] "Training data" is a data set used to train an artificial intelligence model.

[0613] "Test data" is a data set used to evaluate the performance of a trained artificial intelligence model.

[0614] A "generative artificial intelligence model" is a machine learning model built on collected and preprocessed data to perform a specific task (e.g., supply and demand forecasting).

[0615] "Demand forecasting" is the prediction of the demand for a product for a specific period in the future.

[0616] "Inventory quantity" refers to the number of products in a warehouse or on a storefront at a particular point in time.

[0617] The "order quantity" is the quantity of the product that should be newly ordered based on the demand forecast.

[0618] The "user interface" refers to the part that includes the means of interaction and screen display that allows the user to interact with the system.

[0619] "Notification" refers to messages or alerts sent by the system to convey information to the user.

[0620] This invention is a system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[0621] System Configuration

[0622] The system mainly consists of the following components:

[0623] 1. Server: Collects data, preprocesses data, builds models, executes predictions, and outputs results.

[0624] 2. Terminal: Provides the interface that the user operates and displays prediction results and suggestions.

[0625] 3. User: Operates the system, inputs the necessary data, and makes decisions based on the prediction results provided.

[0626] Hardware and Software

[0627] The server consists of server equipment with high-performance computing resources (e.g., workstations equipped with GPUs) and uses the following software:

[0628] Operating System: Linux

[0629] Database: MySQL or PostgreSQL

[0630] Programming language: Python

[0631] Machine learning libraries: TensorFlow, PyTorch

[0632] Data processing libraries: pandas, numpy

[0633] The devices consist of common PCs, tablets, smartphones, and other devices, and use the following software:

[0634] Web browser: Google Chrome, Firefox

[0635] Web application framework: Django or Flask

[0636] Data collection

[0637] Server Operation

[0638] The server collects data from multiple data sources, including the company's ERP system, POS system, and social media. Specifically, it uses SQL queries to retrieve inventory turnover and sales data from the ERP system and pulls sales data from the POS system via API. Social media data is collected by using APIs from Twitter and Facebook. All collected data is integrated into a database.

[0639] Data Preprocessing

[0640] Server Operation

[0641] The server performs data preprocessing on the collected data. It imputes missing values ​​with the mean or median, detects and removes outliers, and performs normalization processes such as standardizing date formats and converting data types. It also splits the dataset into training data and test data.

[0642] Building a demand forecast model

[0643] Server Operation

[0644] The server uses machine learning libraries such as TensorFlow and PyTorch to build generative AI models. It performs feature engineering to define features suitable for model input. It trains the model using algorithms suitable for time-series data, such as LSTM and Transformer. After training, it evaluates the model's performance and tunes hyperparameters as necessary.

[0645] Running a prediction

[0646] Server Operation

[0647] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[0648] Output of results

[0649] Server Operation

[0650] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[0651] Device behavior

[0652] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[0653] Specific examples

[0654] Example 1: Demand forecast due to rising sales of competitors' products

[0655] The server collects sales data and social media trend data from other companies' similar products, performs data preprocessing, and then uses a generative AI model to predict that demand for Product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[0656] Example 2: Increased demand due to price reduction

[0657] The server collects promotion information from the POS system and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[0658] This allows users to manage their inventory appropriately and prevent shortages and surpluses.

[0659] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0660] Step 1: Data collection

[0661] Server Operation

[0662] The server collects the necessary data from multiple data sources (e.g., the company's ERP system, POS system, and social media). Specifically, it uses SQL queries to obtain inventory turnover and sales data from the ERP system, and sales data from the POS system via API. Social media data is collected through APIs from Twitter and Facebook.

[0663] Specific operations and inputs / outputs

[0664] Input: SQL query, API request

[0665] Data manipulation: Pulling data from ERP systems, getting data from POS systems, collecting data from social media APIs

[0666] Output: Consolidated dataset in a database

[0667] Step 2: Data Preprocessing

[0668] Server Operation

[0669] The server performs data preprocessing on the collected data. Specifically, it imputes missing values ​​with the mean or median, and detects and removes outliers. It also performs normalization processes such as standardizing date formats and converting data types. Next, it splits the dataset into training data and test data.

[0670] Specific operations and inputs / outputs

[0671] Input: Integrated dataset

[0672] Data manipulation: Imputing missing values ​​(using Python's pandas library), removing outliers, normalizing (scaling) the data, and splitting the data (using scikit-learn's train_test_split function).

[0673] Output: Preprocessed training and test data

[0674] Step 3: Build a demand forecast model

[0675] Server Operation

[0676] The server uses machine learning libraries (TensorFlow, PyTorch) to build a generative AI model. First, feature engineering is performed to define features suitable for the model. Then, the model is trained using algorithms specialized for time-series data, such as LSTM and Transformer, and the model's performance is evaluated. Hyperparameter tuning is also performed as necessary.

[0677] Specific operations and inputs / outputs

[0678] Input: Training data

[0679] Data manipulation: feature engineering (creating new features), model training (building an LSTM model using TensorFlow), model evaluation (calculating precision, recall, and F-measure)

[0680] Output: A trained generative AI model

[0681] Step 4: Run a prediction

[0682] Server Operation

[0683] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[0684] Specific operations and inputs / outputs

[0685] Input: Newly collected data, trained generative AI model

[0686] Data manipulation: Inputting data into the model, running demand forecasts (predicting future demand through the model)

[0687] Output: Forecast result (future demand curve)

[0688] Step 5: Output the results

[0689] Server Operation

[0690] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[0691] Specific operations and inputs / outputs

[0692] Input: Prediction result

[0693] Data manipulation: Prediction results saved in database, report generation (automatic PDF report generation)

[0694] Output: Report, proposals

[0695] Device behavior

[0696] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[0697] Specific operations and inputs / outputs

[0698] Input: Prediction results and suggestions sent from the server

[0699] Data manipulation: visualization (display in graphs and dashboards)

[0700] Output: Display on the user interface, notifications

[0701] The above processing steps enable highly accurate supply and demand forecasts and efficient inventory management.

[0702] (Application example 1)

[0703] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0704] Product demand forecasts are extremely important in modern inventory management, but achieving highly accurate forecasts remains difficult. In particular, it is difficult to integrate information from diverse data sources and perform detailed analysis of factors that cause demand fluctuations, and the uncertainty of forecasts is likely to have a negative impact on corporate management. Furthermore, while there is a demand for timely supply and demand forecasts that reflect promotions and price reduction information, there is a lack of technology to achieve this. Therefore, there is a need for a system that can solve these problems and achieve highly accurate supply and demand forecasts and inventory management.

[0705] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0706] In this invention, the server includes: means for collecting inventory turnover, sales data, sales data on competitors' products, and trend data on social network services from data sources; means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing the data, and dividing it into training data and test data; means for constructing a generative artificial intelligence model based on the preprocessed data and forecasting product demand; means for calculating future demand fluctuations based on the forecast results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; and means for providing the user with sales promotion information and making supply and demand forecasts based on that information. This enables companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[0707] "Data Source" refers to an information source that provides data necessary for supply and demand forecasting, including inventory turns, sales data, sales data of competitors' products, and trend data of social network services.

[0708] "Inventory turnover" is an index that indicates the frequency of product inventory replacement within a certain period of time.

[0709] "Sales data" refers to data relating to the number of products sold and the amount of sales within a certain period of time.

[0710] "Sales data of competitors' products" refers to data on the sales performance of competitors' products that are similar to or belong to the same market as our company's products.

[0711] "Social network service trend data" refers to data on popular topics and topics of interest extracted from user activities and posts on platforms such as social networking sites.

[0712] "Missing value imputation" is the process of filling in missing values ​​in a dataset in an appropriate way.

[0713] "Data cleansing" is the process of removing inaccurate or unnecessary information from a data set.

[0714] "Normalization" is a preprocessing process that improves the learning effect of an algorithm by constraining data within a certain range.

[0715] "Training data" is the dataset used to train a machine learning model.

[0716] "Test data" is a dataset used to evaluate the performance of a machine learning model.

[0717] A "generative artificial intelligence model" is a supply and demand forecasting model built using machine learning algorithms based on collected and preprocessed data.

[0718] "Prediction results" are data and information regarding future demand calculated by the generated artificial intelligence model.

[0719] "Future demand fluctuations" refers to trends that indicate future increases or decreases in demand based on forecast results.

[0720] "Inventory" is the quantity of a product in stock at a particular point in time.

[0721] "Order quantity" refers to the quantity of the product to be newly ordered to meet future demand.

[0722] A "user interface" is a screen or operating means through which a system interacts with a user.

[0723] "Sales promotion information" is information about measures to increase sales, such as product promotions and price reductions.

[0724] This invention is a "Perfect Inventory Manager" system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users. The details of the system's operation are described below.

[0725] Data collection

[0726] Server Operation

[0727] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from various data sources, such as the company's ERP system, POS system, and social networking API platforms. The data is monitored for regular updates, and new information is automatically incorporated.

[0728] User Actions

[0729] Users can input data such as sales promotion information, new product information, and promotion schedules through their devices, and the information provided by users is also used to improve the accuracy of the prediction model.

[0730] Data Preprocessing

[0731] Server Operation

[0732] The server stores the collected data in a database and imputes missing values ​​in an appropriate manner. It also performs data cleansing to remove inaccurate data and noise and normalizes the data. Furthermore, it splits the dataset into training data and test data to build a highly accurate predictive model. This process utilizes data integration tools and the Python pandas library.

[0733] Building a demand forecast model

[0734] Server Operation

[0735] The server builds a generative AI model using machine learning libraries such as TensorFlow and PyTorch. It performs feature engineering and inputs appropriate features into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[0736] Running a prediction

[0737] Server Operation

[0738] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are analyzed. At the same time, factors behind demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[0739] Output of results

[0740] Device behavior

[0741] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, prompting the user to take appropriate action. For example, the terminal notifies the user of suggested inventory levels and order quantities.

[0742] Specific examples

[0743] When a user inputs new promotional information into the device, the system predicts changes in demand based on that information. For example, if the user inputs information like "20% off all products in the Christmas sale starting December 1, 2023," the generative AI model predicts an increase in sales for the next month and alerts the user that "demand will increase next month, so you should increase your inventory."

[0744] Prompt Sentence Examples

[0745] New promotion information:

[0746] Event: Christmas Sale

[0747] Start date: December 1, 2023

[0748] Discount: 20%

[0749] Predict the percentage increase in sales that this promotion will bring

[0750] This allows companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[0751] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0752] Step 1: Data collection

[0753] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources (e.g., a company's ERP system, POS system, or SNS API). The collected data is sent to the server in JSON or CSV format and stored in a database. This provides a variety of data necessary for demand forecasting.

[0754] Input: ERP system, POS system, SNS API, etc.

[0755] Output: Inventory turnover, sales data, competitor product sales data, SNS trend data (stored in database)

[0756] Step 2: Data Preprocessing

[0757] The server imputes missing values ​​in the collected data in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes (scales) the data to the range of 0 to 1, and splits it into training data and test data. This process uses Python's pandas library and scikit-learn's MinMaxScaler.

[0758] Input: Collected raw data (taken from database)

[0759] Output: Preprocessed data (scaled data after missing value imputation and data cleansing)

[0760] Step 3: Building a generative AI model

[0761] The server uses the preprocessed data to build a generative AI model. It uses TensorFlow and PyTorch to train a time-series prediction model using algorithms such as LSTM and Transformer. It separates the training data from the test data to evaluate the model's accuracy and tune it accordingly.

[0762] Input: Preprocessed data (training data and test data)

[0763] Output: Highly accurate generative AI model (supply and demand forecast model)

[0764] Step 4: Run a prediction

[0765] The server inputs the newly collected data into the generative AI model to predict future product demand. The prediction results are generated as data showing future demand fluctuations, and also generate detailed analysis results of the factors behind demand fluctuations, taking into account sales data of competitors' products and social media trend data.

[0766] Input: New data (latest data from data collection step)

[0767] Output: Forecast results (future demand fluctuation data and analysis results of demand fluctuation factors)

[0768] Step 5: Output the results

[0769] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, allowing the user to take appropriate action based on the results. For example, the terminal may notify the user of suggested inventory levels or order quantities.

[0770] Input: Prediction results and suggestions (data sent from the server)

[0771] Output: Data displayed in a user interface in the form of graphs and dashboards

[0772] Step 6: Enter promotion information

[0773] Users input sales promotion information (e.g., promotions and price reductions) through their terminals. This information is sent to the server and stored in a database, which is then used to forecast supply and demand for future purchases.

[0774] Input: Sales promotion information (promotions and price reduction information)

[0775] Output: Saved promotion information (stored in database)

[0776] Step 7: Promotion-based forecasting

[0777] The server predicts supply and demand based on sales promotion information entered by the user. For example, it predicts an increase in demand after a Christmas sale based on information about the sale. This prediction result is also sent to the terminal and notified to the user.

[0778] Input: Sales promotion information (obtained from database)

[0779] Output: Prediction results and proposals that take into account sales promotion information (sent from the server to the device)

[0780] Through these steps, the system achieves highly accurate supply and demand forecasts and inventory management.

[0781] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0782] This invention is a "Perfect Inventory Manager" system that uses generative AI to perform highly accurate supply and demand forecasts in inventory management, and by combining it with an emotion engine that recognizes user emotions, it improves the effectiveness of proposals and user satisfaction. This system operates in cooperation with the server, terminals, and users.

[0783] Data collection

[0784] Server Operation

[0785] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social network service APIs. The data is updated regularly and is set to be automatically imported.

[0786] Device behavior

[0787] The device provides an interface for users to operate. Users input information about new products and upcoming promotions into the device. This information is also stored in the database and used to improve the accuracy of the prediction model.

[0788] Data Preprocessing

[0789] Server Operation

[0790] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[0791] Building a demand forecast model

[0792] Server Operation

[0793] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It then trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[0794] Future demand forecast

[0795] Server Operation

[0796] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[0797] Saving and outputting prediction results

[0798] Server Operation

[0799] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[0800] Emotion Recognition and Feedback

[0801] Device behavior

[0802] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and text message content when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server.

[0803] Server Operation

[0804] The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the content and timing of the suggestions, and reflects the user's emotional feedback in optimizing the system.

[0805] Display on the user interface

[0806] Device behavior

[0807] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[0808] Specific examples

[0809] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[0810] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the device of this result and issues an alert to the user that "demand will increase next month, so you should increase your inventory." The device uses an emotion engine to collect the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional feedback, evaluates whether the user is satisfied, and reflects this in the next proposal.

[0811] Example 2: Increased demand due to price reduction and emotional feedback

[0812] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased." The terminal collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[0813] Based on these specific examples, this system can achieve highly accurate supply and demand forecasts, and by taking user emotions into consideration, it can improve the effectiveness of its proposals. As a result, it is possible to prevent overstocks and shortages and improve the efficiency of inventory management.

[0814] The processing flow will be explained below.

[0815] Step 1: Data collection

[0816] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[0817] Step 2: User Data Entry

[0818] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[0819] Step 3: Data Preprocessing

[0820] The server preprocesses the collected data. First, it performs imputation processing using a method to impute missing values ​​using the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and arranges it into a unified format.

[0821] Step 4: Splitting the Dataset

[0822] The server splits the preprocessed data into training data and test data. Specifically, 80% of the data is stored in the database as training data and 20% as test data.

[0823] Step 5: Feature Engineering

[0824] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates for social media trends, and prepares the selected features for input into the generative AI model.

[0825] Step 6: Building a generative AI model

[0826] The server uses machine learning libraries to build generative AI models such as LSTM and Transformer. First, it trains the model using training data to learn patterns. After training is complete, it evaluates the model using test data to check its accuracy and generalization performance.

[0827] Step 7: Forecast future demand

[0828] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[0829] Step 8: Save and output prediction results

[0830] The server stores the forecast results and analysis reports in a database, and calculates appropriate inventory and order quantities based on the stored results and makes recommendations to the user.

[0831] Step 9: Emotion Recognition and Feedback

[0832] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and the content of text messages when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server. The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. The server adjusts the suggestion content and timing according to the level of satisfaction, and reflects the user's emotional feedback in optimizing the system.

[0833] Step 10: Display in the User Interface

[0834] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[0835] Step 11: User Actions

[0836] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocks and shortages.

[0837] Example 2

[0838] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0839] In inventory management, not only is it difficult to make accurate supply and demand forecasts, but the inability to properly consider user emotions and feedback reduces the effectiveness of proposals and user satisfaction.In addition, the inability to fully utilize external data such as the trends of competitors and social media trends prevents the accuracy of supply and demand forecasts from improving.

[0840] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for collecting inventory turnover, sales data, sales data of competitors, and trend information on social network services from information sources; means for preprocessing the collected data, filling in missing data, cleaning the data, standardizing the data, and dividing it into training data and test data; means for constructing a generative AI model based on the preprocessed data and predicting product demand; means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; means for recognizing emotions from user operations, voice, and text messages on the terminal and transmitting the data to the server; means for analyzing the emotion data transmitted to the server and determining user satisfaction; and means for optimizing the content and timing of proposals based on user satisfaction. This not only enables highly accurate supply and demand predictions that take into account the trends of competitors and social networking services trends, but also enables proposals that reflect user emotions, thereby improving inventory management efficiency and user satisfaction.

[0841] The "information source" is a data supplier that provides a variety of data such as inventory turnover, sales data, sales data of competitors, and trend information of social networking services.

[0842] "Inventory turnover" is an indicator of the number of times inventory items are turned over within a certain period of time. It is typically used to evaluate how quickly inventory is sold.

[0843] "Sales data" refers to data that includes information such as the quantity, price, and sales period of products and services sold.

[0844] "Competitor sales data" refers to data on the sales of competitors' products and services in the market. This data is used to understand the competitive landscape and to forecast supply and demand.

[0845] "Social network service trend information" refers to information about popular topics, products, and services on social media. This allows us to understand consumer trends and market trends.

[0846] "Missing data completion" is a process of completing missing values ​​in a dataset by filling in the average value or previous or next values.

[0847] "Data cleaning" is the process of removing outliers and noise from a dataset to improve data quality.

[0848] "Data standardization" is the process of converting data into a unified format to make it consistent, which makes it easier to analyze and train models.

[0849] A "generative artificial intelligence model" is a model that uses machine learning algorithms to predict future situations based on past data. It is also called a generative AI model.

[0850] "User interface" refers to the screens and operating methods that allow users to interact with the system, input information, and check results.

[0851] "Emotion recognition" is the process by which the device analyzes emotions from the user's voice or text messages and determines whether the emotion is positive or negative.

[0852] "Satisfaction" is an index that indicates the degree of satisfaction that a user feels with the services or proposals provided.

[0853] Optimization is the adjustment and improvement of systems and processes to make them operate most efficiently and effectively.

[0854] This invention is a system for implementing highly accurate inventory management for a company. This system operates in cooperation with a server, terminals, and users. The hardware and software used to build the system of this invention include the following elements:

[0855] Hardware and Software

[0856] server

[0857] The server collects data, preprocesses it, builds generative AI models, and stores the prediction results. The server uses the following software:

[0858] Data collection: HTTP client for API communication (e.g., requests library)

[0859] Data preprocessing: Python's pandas and scikit-learn libraries

[0860] Building generative AI models: TensorFlow or PyTorch

[0861] Terminal

[0862] The device provides a user interface and performs emotion recognition. Specifically, it uses the following software:

[0863] User Interface: React.js or Vue.js

[0864] Emotion Recognition: IBM Watson's Natural Language Understanding API

[0865] User

[0866] The user operates the terminal to input the data necessary for inventory management and receives suggestions from the system.

[0867] Specific processing of the system

[0868] Data collection

[0869] The server periodically retrieves inventory turnover, sales data, competitor sales data, and trend information from social networking services from information sources, thereby incorporating specific data that reflects the latest market trends into the system.

[0870] Data Preprocessing

[0871] The collected data undergoes preprocessing, including missing data imputation, data cleaning, and data standardization, to transform it into a consistent dataset, which is then ready for model training.

[0872] Building generative AI models

[0873] Based on the preprocessed data, the server builds a generative AI model using TensorFlow or PyTorch. It uses algorithms specialized for time series data, such as LSTM and Transformer, to predict product demand. After training is complete, the model is evaluated using test data.

[0874] Demand forecasting and proposals

[0875] The server inputs newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are calculated and appropriate inventory and order quantities are calculated. The prediction results are notified to the user via the user interface.

[0876] Emotion recognition and feedback collection

[0877] The device uses an emotion engine to analyze the tone of voice and text message content when the user confirms the proposal, and sends the user's emotional data to the server. The server analyzes the emotional data, determines the user's level of satisfaction, and optimizes the content and timing of the next proposal.

[0878] Specific examples

[0879] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[0880] The server collects sales data and social media trend data from other companies' products and performs data preprocessing. Using a generative AI model, it predicts that demand for the product will increase by 50% next month. This result is notified to the device, and an alert is issued stating, "Demand will increase next month, so you should increase your inventory." The device then uses an emotion engine to collect the user's emotional feedback on this suggestion and sends it to the server. The server analyzes the emotional feedback and reflects it in the next suggestion.

[0881] Example 2: Increased demand due to price reduction and emotional feedback

[0882] The server collects promotional information and obtains information that the product price will be reduced by 10%. Based on this data, it uses a generative AI model to predict that demand will increase by 30% after the price reduction. The server sends the prediction result to the device and provides a notification that "product demand will increase after the price reduction, so production should be increased." The device collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[0883] As described above, the system of the present invention is capable of highly accurate supply and demand forecasting and proposals that take into account the user's emotions, preventing overstocks and shortages and improving the efficiency of inventory management.

[0884] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0885] Step 1: Data collection

[0886] Server Operation

[0887] 1. The server collects data from information sources via API communication. Specifically, it obtains inventory turnover and sales data from the ERP system, real-time sales data from the POS system, and trend information from the API of social media.

[0888] 2. Data inputs include API endpoints and authentication information for each information source, while outputs include collected inventory turnover, sales data, competitor sales data, and social media trend information.

[0889] Specific actions

[0890] The server uses the Python requests library to collect data from each information source, for example by sending an API request like this:

[0891] python

[0892] import requests

[0893] response = requests.get('https: / / api.example.com / inventory')

[0894] data = response.json()

[0895] Step 2: Data Preprocessing

[0896] Server Operation

[0897] 1. Store the collected data in a database.

[0898] 2. The server uses SQL queries to extract missing values ​​from the data and impute them with the average or surrounding values.

[0899] 3. The server performs data cleansing to remove outliers and noise.

[0900] 4. Finally, standardize the data to create a consistent dataset.

[0901] Input and Output

[0902] The input includes the raw data collected in step 1. The output is the formatted data that has been imputed and cleaned.

[0903] Specific actions

[0904] The server uses the Python pandas library to impute and clean the data, for example imputing missing values ​​as follows:

[0905] python

[0906] import pandas as pd

[0907] df = pd.DataFrame(data)

[0908] df.fillna(df.mean(), inplace=True)

[0909] Step 3: Building a generative AI model

[0910] Server Operation

[0911] 1. Using the preprocessed data, we build a generative AI model. Specifically, we define features for the time series data we will use.

[0912] 2. The server uses TensorFlow or PyTorch to train the model using algorithms such as LSTM or Transformer.

[0913] 3. After training, the model is evaluated using test data to check accuracy and generalization performance.

[0914] Input and Output

[0915] The input is a pre-processed dataset, and the output is a trained generative AI model.

[0916] Specific actions

[0917] Perform feature engineering and train an AI model. For example, train an LSTM model with the following code:

[0918] python

[0919] import tensorflow as tf

[0920] from tensorflow.keras.models import Sequential

[0921] from tensorflow.keras.layers import LSTM, Dense

[0922] model = Sequential()

[0923] model.add(LSTM(50, return_sequences=True, input_shape=(n_timesteps, n_features)))

[0924] model.add(LSTM(50))

[0925] model.add(Dense(1))

[0926] model.compile(optimizer='adam', loss='mse')

[0927] model.fit(train_X, train_y, epochs=50, batch_size=72, validation_data=(test_X, test_y))

[0928] Step 4: Forecast future demand

[0929] Server Operation

[0930] 1. Newly collected data is fed into a generative AI model to predict future product demand.

[0931] 2. Based on the demand forecast, a future demand curve is generated and factors affecting demand are analyzed taking into account sales data of competitors and social media trend information.

[0932] Input and Output

[0933] The inputs include new data collected in real time, and the outputs are future demand forecasts and analysis of the factors that cause fluctuations.

[0934] Specific actions

[0935] The server inputs the new data into the model and uses the model's forecasting capabilities to calculate demand. The model forecasts as follows:

[0936] python

[0937] future_predictions = model.predict(new_data)

[0938] Step 5: Save and output prediction results

[0939] Server Operation

[0940] 1. Save the prediction results and analysis reports in a database.

[0941] 2. The server calculates the appropriate inventory and order quantities based on the prediction results and generates a notification to suggest to the user.

[0942] Input and Output

[0943] The input includes future demand forecast results. The output is a saved forecast and a proposed notification.

[0944] Specific actions

[0945] The server stores the prediction results in a MySQL database and generates notifications to send to the device, for example using the following SQL query:

[0946] sql

[0947] INSERT INTO predictions (date, product_id, predicted_demand) VALUES (%s, %s, %s)

[0948] Step 6: Emotion recognition and feedback collection

[0949] Device behavior

[0950] 1. The device analyzes the user's operations, voice, and text messages using an emotion engine and sends the emotion data to the server.

[0951] Server Operation

[0952] 1. The server analyzes the transmitted emotion data and determines the user's satisfaction level.

[0953] 2. Optimize the content and timing of proposals based on customer satisfaction.

[0954] Input and Output

[0955] The input includes user emotion data sent from the device, and the output is a satisfaction analysis result and optimized recommendations.

[0956] Specific actions

[0957] The device uses IBM Watson API to analyze emotions and send the results to the server. For example, call the API as follows:

[0958] python

[0959] from ibm_watson import NaturalLanguageUnderstandingV1

[0960] from ibm_watson.natural_language_understanding_v1 import Features, EmotionOptions

[0961] nlu = NaturalLanguageUnderstandingV1(version='2021-03-25')

[0962] response = nlu.analyze(text='"user input text"', features=Features(emotion=EmotionOptions())).get_result()

[0963] Step 7: Display in the user interface

[0964] Device behavior

[0965] 1. Visually display prediction results and suggestions to the user.

[0966] 2. Present information in an easy-to-understand manner using graphs and dashboards, encouraging users to take appropriate action.

[0967] Input and Output

[0968] The input includes prediction results and suggestion notifications received from the server, and the output is visualized information provided to the user.

[0969] Specific actions

[0970] The terminal uses the React.js or Vue.js framework to display the prediction results in a dashboard format. For example, the display process can be done with the following code:

[0971] javascript

[0972] import React from 'react';

[0973] import { LineChart, Line, CartesianGrid, XAxis, YAxis, Tooltip} from 'recharts';

[0974] const data = [...] / / received prediction data

[0975] const Dashboard = () => (

[0976] <linechart width="{600}" height="{300}" data="{data}">

[0977] <line type="monotone" dataKey="predicted_demand" stroke="8884d8" / >

[0978] <cartesiangrid stroke="ccc" / >

[0979] <xaxis dataKey="date" / >

[0980] <yaxis / >

[0981] <tooltip / >

[0982] < / linechart>

[0983] );

[0984] export default Dashboard;

[0985] Through the above processing steps, companies can realize highly accurate supply and demand forecasts and proposals that take user emotions into consideration.

[0986] (Application example 2)

[0987] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0988] The accuracy of supply and demand forecasts in inventory management is low, which leads to a high risk of overstocking and shortages. In addition, there is a lack of proposals that take user emotions into account, so a method to improve user satisfaction is also needed.

[0989] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: [means for collecting inventory turnover, sales data, sales data of competitors' products, and trend data of social network services from data sources;] [means for preprocessing the collected data, imputing missing values, data cleansing, normalizing, and dividing the collected data into training data and test data;] [means for constructing a generative algorithm model based on the preprocessed data and predicting product demand;] [means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities;] [means for displaying the proposal on a user interface and notifying the user;] [means for analyzing user emotions using an emotion recognition engine and collecting feedback; and [means for optimizing the prediction model based on the feedback and improving the effectiveness of the proposal content.] This enables highly accurate supply and demand prediction and improved user satisfaction.

[0990] "Data Source" refers to information sources that provide inventory turns, sales data, competitor product sales data, and SNS trend data, such as a company's ERP system, POS system, and social network service API.

[0991] "Preprocessing" refers to the process of filling in missing values, cleansing, and normalizing the collected data, and then dividing the data into training data and test data.

[0992] A "generative algorithm model" is a predictive model that is constructed using machine learning algorithms using pre-processed data.

[0993] "Supply and demand forecasting" is a technique that uses generative algorithm models to predict future demand for a product.

[0994] The "user interface" is an interface for visually displaying to the user the proposals and prediction results presented by the server.

[0995] An "emotion recognition engine" is a technology that recognizes and analyzes a user's emotions from the tone of their voice and the content of their text messages.

[0996] "Feedback" refers to data collected about users' reactions and emotions to the proposals, and is used to optimize the predictive model.

[0997] This invention is a system called "Inventory Perfect Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[0998] Data collection

[0999] Server Operation

[1000] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from data sources, utilizing the company's ERP system, POS system, and SNS API. The data is updated periodically and automatically imported.

[1001] Device behavior

[1002] The device provides a user interface, and users input information about new products and promotional plans into the device. The device is also equipped with an emotion recognition engine that recognizes emotions from the tone of voice and content of text messages generated during user operations. This information is also stored in a database and used to improve the accuracy of the predictive model.

[1003] Data Preprocessing

[1004] Server Operation

[1005] The server preprocesses the collected data: missing values ​​are filled with the average or surrounding values, and inaccurate data and noise are removed through data cleansing. The data is then normalized and organized into a unified format.

[1006] Building a supply and demand forecast model

[1007] Server Operation

[1008] The server uses a machine learning library (e.g., TensorFlow, PyTorch) to build a generative algorithm model. Feature engineering is performed to define appropriate features and input them into the model. The model is trained using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[1009] Future demand forecast

[1010] Server Operation

[1011] The server inputs the newly collected data into a generative algorithm model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and trend data from social networking services.

[1012] Saving and outputting prediction results

[1013] Server Operation

[1014] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[1015] Emotion Recognition and Feedback

[1016] Device behavior

[1017] The device is equipped with an emotion recognition engine that recognizes the user's emotions from the tone of voice when the user operates the device and the content of text messages. The emotion recognition engine collects the user's emotion data and sends it to the server.

[1018] Server Operation

[1019] The server analyzes the emotion data sent from the emotion recognition engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the suggestion content and timing, and reflects the user's emotional feedback in optimizing the system.

[1020] Display on the user interface

[1021] Device behavior

[1022] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[1023] Specific examples

[1024] For example, suppose a new product D is scheduled to be released next month. The server collects existing sales data, competitor data, and social media trend information, and uses a generative AI model to predict supply and demand. As a result, the predictive model predicts that demand for product D will increase by 40%. Using emotion recognition feedback, the store manager receives positive feedback that the prediction is "very reliable."

[1025] Prompt Sentence Examples

[1026] The supply and demand forecast for next month shows a 40% increase in demand for new product D. Please arrange the necessary inventory based on this forecast. You can provide feedback, which will be evaluated by the sentiment recognition engine. For example, you can enter feedback such as "The forecast is very reliable."

[1027] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1028] Step 1:

[1029] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources. The input is data from ERP systems, POS systems, SNS APIs, etc., which is collected periodically and stored in a database. The output is the various collected data.

[1030] Step 2:

[1031] The server preprocesses the collected data. Specifically, it imputes missing values ​​with the average or surrounding values, and removes inaccurate data and noise through data cleansing. Next, it normalizes the data, organizes it into a unified format, and splits it into training data and test data. The input is the data collected in step 1, and the output is the preprocessed training data and test data.

[1032] Step 3:

[1033] The server uses machine learning libraries to build a generative algorithm model. Specifically, it uses TensorFlow or PyTorch to extract features from the preprocessed data and trains the model using algorithms such as LSTM or Transformer. The input is the preprocessed data from step 2, and the output is the trained generative algorithm model.

[1034] Step 4:

[1035] The server inputs new data into the generative algorithm model to predict future product supply and demand. The inputs are the newly collected data and the trained model, and the output is the future demand forecast.

[1036] Step 5:

[1037] The server calculates future demand fluctuations based on the forecast results and proposes appropriate inventory and order quantities. The input is the supply and demand forecast results obtained in step 4, and the output is the proposed inventory and order quantities.

[1038] Step 6:

[1039] The terminal displays the prediction results and suggestions sent from the server on a user interface. The input is the prediction results and suggestions sent from the server, and the output is a visual display on the user interface.

[1040] Step 7:

[1041] The user operates the device to input information about new products and promotion plans. Emotional data is also collected from the tone of voice and the content of text messages while operating the device. The input is the user's new product information and emotional data, and the output is the collected data.

[1042] Step 8:

[1043] The emotion recognition engine analyzes the user's emotion data and sends it to the server. The input is the emotion data sent from the device, and the output is the analyzed emotion information.

[1044] Step 9:

[1045] The server analyzes the emotion data sent from the emotion recognition engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the suggestion content and timing and reflects this in the next prediction. The input is the analyzed emotion data, and the output is the adjusted suggestion content.

[1046] Step 10:

[1047] The server finally optimizes the prediction model based on the prediction results and user feedback to improve the effectiveness of the recommendations. The input is the feedback data and the prediction results, and the output is the optimized prediction model.

[1048] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1049] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1050] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1051] [Third embodiment]

[1052] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1053] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[1054] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1055] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1056] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1057] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1058] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1059] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1060] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1061] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1062] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1063] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[1064] This invention is a system called "Perfect Inventory Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[1065] Data collection

[1066] Server Operation

[1067] The server collects product inventory turnover, sales data, competitor product sales data, and SNS trend data from various data sources (e.g., the company's ERP system, POS system, social networking services). The data is monitored to ensure it is updated regularly, and new information is automatically incorporated.

[1068] Device behavior

[1069] The terminal provides an interface for users to operate. Users input data such as information on new products and promotional plans into the terminal, which adds the information necessary for prediction to the database.

[1070] Data Preprocessing

[1071] Server Operation

[1072] The server stores the collected data in a database, imputes missing values ​​in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes the data, and splits the dataset into training data and test data to build a highly accurate model.

[1073] Building a demand forecast model

[1074] Server Operation

[1075] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[1076] Running a prediction

[1077] Server Operation

[1078] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[1079] Output of results

[1080] Server Operation

[1081] The server stores the prediction results in a database, generates reports if necessary, calculates appropriate inventory and order quantities based on the prediction results, and makes suggestions to the user.

[1082] Device behavior

[1083] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[1084] Specific examples

[1085] Example 1: Demand forecast due to rising sales of competitors' products

[1086] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[1087] Example 2: Increased demand due to price reduction

[1088] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[1089] Based on these specific examples, this system can realize highly accurate supply and demand forecasts, prevent overstocks and shortages, and improve the efficiency of inventory management.

[1090] The processing flow will be explained below.

[1091] Step 1: Data collection

[1092] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[1093] Step 2: User Data Entry

[1094] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[1095] Step 3: Data Preprocessing

[1096] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[1097] Step 4: Splitting the Dataset

[1098] The server splits the preprocessed data into training data and test data, storing 80% of the data in the database as training data and 20% as test data.

[1099] Step 5: Feature Engineering

[1100] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates of social media trends, and prepares these features for input into the generative AI model.

[1101] Step 6: Building a generative AI model

[1102] The server uses machine learning libraries to build generative artificial intelligence models such as LSTM and Transformer. First, the model is trained using training data to learn patterns. After training is complete, the model is evaluated using test data to check accuracy and generalization performance.

[1103] Step 7: Forecast future demand

[1104] The server inputs new data into the generative AI model to predict future demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[1105] Step 8: Save and output prediction results

[1106] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[1107] Step 9: Display in the User Interface

[1108] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[1109] Step 10: User Actions

[1110] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocking or shortages.

[1111] Example 1

[1112] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1113] Current inventory management systems are prone to inventory shortages and surpluses due to inaccurate demand forecasts. In addition, inaccuracies in collected data affect the accuracy of forecasts, so there is a need for more efficient data preprocessing. Furthermore, it is difficult to forecast demand while taking into account competitor trends and social media trends, making strategic inventory management difficult. There is a need for a system that can solve these issues and achieve highly accurate inventory management.

[1114] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1115] In this invention, the server includes: [means for collecting inventory turnover, sales data, competitor product sales data, and social network trend data from multiple data sources;] [means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing it, and dividing it into training data and test data; and] [means for constructing a generative artificial intelligence model based on the preprocessed data and predicting product demand.] This enables [highly accurate demand forecasting and efficient inventory management].

[1116] A "data source" is a system or resource for obtaining data from various sources.

[1117] "Inventory turnover" is a value that indicates the number of times an inventory product is sold within a certain period of time.

[1118] "Sales data" is data that includes numerical information about product sales during a specific period.

[1119] "Sales data of competitors' products" refers to data that includes numerical information on sales of competitors' products that exist in the same market.

[1120] "Social network trend data" is data that includes information about trends and topics on online social networking services.

[1121] A "missing value" is a value that indicates a portion of a data set where information is missing.

[1122] "Data cleansing" is the process of removing inaccurate data and noise from a dataset to improve its accuracy.

[1123] "Normalization" is a process that unifies the scale of data to make it easier to compare data with different units or scales.

[1124] "Training data" is a data set used to train an artificial intelligence model.

[1125] "Test data" is a data set used to evaluate the performance of a trained artificial intelligence model.

[1126] A "generative artificial intelligence model" is a machine learning model built on collected and preprocessed data to perform a specific task (e.g., supply and demand forecasting).

[1127] "Demand forecasting" is the prediction of the demand for a product for a specific period in the future.

[1128] "Inventory quantity" refers to the number of products in a warehouse or on a storefront at a particular point in time.

[1129] The "order quantity" is the quantity of the product that should be newly ordered based on the demand forecast.

[1130] The "user interface" refers to the part that includes the means of interaction and screen display that allows the user to interact with the system.

[1131] "Notification" refers to messages or alerts sent by the system to convey information to the user.

[1132] This invention is a system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[1133] System Configuration

[1134] The system mainly consists of the following components:

[1135] 1. Server: Collects data, preprocesses data, builds models, executes predictions, and outputs results.

[1136] 2. Terminal: Provides the interface that the user operates and displays prediction results and suggestions.

[1137] 3. User: Operates the system, inputs the necessary data, and makes decisions based on the prediction results provided.

[1138] Hardware and Software

[1139] The server consists of server equipment with high-performance computing resources (e.g., workstations equipped with GPUs) and uses the following software:

[1140] Operating System: Linux

[1141] Database: MySQL or PostgreSQL

[1142] Programming language: Python

[1143] Machine learning libraries: TensorFlow, PyTorch

[1144] Data processing libraries: pandas, numpy

[1145] The devices consist of common PCs, tablets, smartphones, and other devices, and use the following software:

[1146] Web browser: Google Chrome, Firefox

[1147] Web application framework: Django or Flask

[1148] Data collection

[1149] Server Operation

[1150] The server collects data from multiple data sources, including the company's ERP system, POS system, and social media. Specifically, it uses SQL queries to retrieve inventory turnover and sales data from the ERP system and pulls sales data from the POS system via API. Social media data is collected by using APIs from Twitter and Facebook. All collected data is integrated into a database.

[1151] Data Preprocessing

[1152] Server Operation

[1153] The server performs data preprocessing on the collected data. It imputes missing values ​​with the mean or median, detects and removes outliers, and performs normalization processes such as standardizing date formats and converting data types. It also splits the dataset into training data and test data.

[1154] Building a demand forecast model

[1155] Server Operation

[1156] The server uses machine learning libraries such as TensorFlow and PyTorch to build generative AI models. It performs feature engineering to define features suitable for model input. It trains the model using algorithms suitable for time-series data, such as LSTM and Transformer. After training, it evaluates the model's performance and tunes hyperparameters as necessary.

[1157] Running a prediction

[1158] Server Operation

[1159] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[1160] Output of results

[1161] Server Operation

[1162] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[1163] Device behavior

[1164] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[1165] Specific examples

[1166] Example 1: Demand forecast due to rising sales of competitors' products

[1167] The server collects sales data and social media trend data from other companies' similar products, performs data preprocessing, and then uses a generative AI model to predict that demand for Product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[1168] Example 2: Increased demand due to price reduction

[1169] The server collects promotion information from the POS system and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[1170] This allows users to manage their inventory appropriately and prevent shortages and surpluses.

[1171] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1172] Step 1: Data collection

[1173] Server Operation

[1174] The server collects the necessary data from multiple data sources (e.g., the company's ERP system, POS system, and social media). Specifically, it uses SQL queries to obtain inventory turnover and sales data from the ERP system, and sales data from the POS system via API. Social media data is collected through APIs from Twitter and Facebook.

[1175] Specific operations and inputs / outputs

[1176] Input: SQL query, API request

[1177] Data manipulation: Pulling data from ERP systems, getting data from POS systems, collecting data from social media APIs

[1178] Output: Consolidated dataset in a database

[1179] Step 2: Data Preprocessing

[1180] Server Operation

[1181] The server performs data preprocessing on the collected data. Specifically, it imputes missing values ​​with the mean or median, and detects and removes outliers. It also performs normalization processes such as standardizing date formats and converting data types. Next, it splits the dataset into training data and test data.

[1182] Specific operations and inputs / outputs

[1183] Input: Integrated dataset

[1184] Data manipulation: Imputing missing values ​​(using Python's pandas library), removing outliers, normalizing (scaling) the data, and splitting the data (using scikit-learn's train_test_split function).

[1185] Output: Preprocessed training and test data

[1186] Step 3: Build a demand forecast model

[1187] Server Operation

[1188] The server uses machine learning libraries (TensorFlow, PyTorch) to build a generative AI model. First, feature engineering is performed to define features suitable for the model. Then, the model is trained using algorithms specialized for time-series data, such as LSTM and Transformer, and the model's performance is evaluated. Hyperparameter tuning is also performed as necessary.

[1189] Specific operations and inputs / outputs

[1190] Input: Training data

[1191] Data manipulation: feature engineering (creating new features), model training (building an LSTM model using TensorFlow), model evaluation (calculating precision, recall, and F-measure)

[1192] Output: A trained generative AI model

[1193] Step 4: Run a prediction

[1194] Server Operation

[1195] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[1196] Specific operations and inputs / outputs

[1197] Input: Newly collected data, trained generative AI model

[1198] Data manipulation: Inputting data into the model, running demand forecasts (predicting future demand through the model)

[1199] Output: Forecast result (future demand curve)

[1200] Step 5: Output the results

[1201] Server Operation

[1202] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[1203] Specific operations and inputs / outputs

[1204] Input: Prediction result

[1205] Data manipulation: Prediction results saved in database, report generation (automatic PDF report generation)

[1206] Output: Report, proposals

[1207] Device behavior

[1208] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[1209] Specific operations and inputs / outputs

[1210] Input: Prediction results and suggestions sent from the server

[1211] Data manipulation: visualization (display in graphs and dashboards)

[1212] Output: Display on the user interface, notifications

[1213] The above processing steps enable highly accurate supply and demand forecasts and efficient inventory management.

[1214] (Application example 1)

[1215] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1216] Product demand forecasts are extremely important in modern inventory management, but achieving highly accurate forecasts remains difficult. In particular, it is difficult to integrate information from diverse data sources and perform detailed analysis of factors that cause demand fluctuations, and the uncertainty of forecasts is likely to have a negative impact on corporate management. Furthermore, while there is a demand for timely supply and demand forecasts that reflect promotions and price reduction information, there is a lack of technology to achieve this. Therefore, there is a need for a system that can solve these problems and achieve highly accurate supply and demand forecasts and inventory management.

[1217] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1218] In this invention, the server includes: means for collecting inventory turnover, sales data, sales data on competitors' products, and trend data on social network services from data sources; means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing the data, and dividing it into training data and test data; means for constructing a generative artificial intelligence model based on the preprocessed data and forecasting product demand; means for calculating future demand fluctuations based on the forecast results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; and means for providing the user with sales promotion information and making supply and demand forecasts based on that information. This enables companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[1219] "Data Source" refers to an information source that provides data necessary for supply and demand forecasting, including inventory turns, sales data, sales data of competitors' products, and trend data of social network services.

[1220] "Inventory turnover" is an index that indicates the frequency of product inventory replacement within a certain period of time.

[1221] "Sales data" refers to data relating to the number of products sold and the amount of sales within a certain period of time.

[1222] "Sales data of competitors' products" refers to data on the sales performance of competitors' products that are similar to or belong to the same market as our company's products.

[1223] "Social network service trend data" refers to data on popular topics and topics of interest extracted from user activities and posts on platforms such as social networking sites.

[1224] "Missing value imputation" is the process of filling in missing values ​​in a dataset in an appropriate way.

[1225] "Data cleansing" is the process of removing inaccurate or unnecessary information from a data set.

[1226] "Normalization" is a preprocessing process that improves the learning effect of an algorithm by constraining data within a certain range.

[1227] "Training data" is the dataset used to train a machine learning model.

[1228] "Test data" is a dataset used to evaluate the performance of a machine learning model.

[1229] A "generative artificial intelligence model" is a supply and demand forecasting model built using machine learning algorithms based on collected and preprocessed data.

[1230] "Prediction results" are data and information regarding future demand calculated by the generated artificial intelligence model.

[1231] "Future demand fluctuations" refers to trends that indicate future increases or decreases in demand based on forecast results.

[1232] "Inventory" is the quantity of a product in stock at a particular point in time.

[1233] "Order quantity" refers to the quantity of the product to be newly ordered to meet future demand.

[1234] A "user interface" is a screen or operating means through which a system interacts with a user.

[1235] "Sales promotion information" is information about measures to increase sales, such as product promotions and price reductions.

[1236] This invention is a "Perfect Inventory Manager" system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users. The details of the system's operation are described below.

[1237] Data collection

[1238] Server Operation

[1239] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from various data sources, such as the company's ERP system, POS system, and social networking API platforms. The data is monitored for regular updates, and new information is automatically incorporated.

[1240] User Actions

[1241] Users can input data such as sales promotion information, new product information, and promotion schedules through their devices, and the information provided by users is also used to improve the accuracy of the prediction model.

[1242] Data Preprocessing

[1243] Server Operation

[1244] The server stores the collected data in a database and imputes missing values ​​in an appropriate manner. It also performs data cleansing to remove inaccurate data and noise and normalizes the data. Furthermore, it splits the dataset into training data and test data to build a highly accurate predictive model. This process utilizes data integration tools and the Python pandas library.

[1245] Building a demand forecast model

[1246] Server Operation

[1247] The server builds a generative AI model using machine learning libraries such as TensorFlow and PyTorch. It performs feature engineering and inputs appropriate features into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[1248] Running a prediction

[1249] Server Operation

[1250] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are analyzed. At the same time, factors behind demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[1251] Output of results

[1252] Device behavior

[1253] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, prompting the user to take appropriate action. For example, the terminal notifies the user of suggested inventory levels and order quantities.

[1254] Specific examples

[1255] When a user inputs new promotional information into the device, the system predicts changes in demand based on that information. For example, if the user inputs information like "20% off all products in the Christmas sale starting December 1, 2023," the generative AI model predicts an increase in sales for the next month and alerts the user that "demand will increase next month, so you should increase your inventory."

[1256] Prompt Sentence Examples

[1257] New promotion information:

[1258] Event: Christmas Sale

[1259] Start date: December 1, 2023

[1260] Discount: 20%

[1261] Predict the percentage increase in sales that this promotion will bring

[1262] This allows companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[1263] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1264] Step 1: Data collection

[1265] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources (e.g., a company's ERP system, POS system, or SNS API). The collected data is sent to the server in JSON or CSV format and stored in a database. This provides a variety of data necessary for demand forecasting.

[1266] Input: ERP system, POS system, SNS API, etc.

[1267] Output: Inventory turnover, sales data, competitor product sales data, SNS trend data (stored in database)

[1268] Step 2: Data Preprocessing

[1269] The server imputes missing values ​​in the collected data in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes (scales) the data to the range of 0 to 1, and splits it into training data and test data. This process uses Python's pandas library and scikit-learn's MinMaxScaler.

[1270] Input: Collected raw data (taken from database)

[1271] Output: Preprocessed data (scaled data after missing value imputation and data cleansing)

[1272] Step 3: Building a generative AI model

[1273] The server uses the preprocessed data to build a generative AI model. It uses TensorFlow and PyTorch to train a time-series prediction model using algorithms such as LSTM and Transformer. It separates the training data from the test data to evaluate the model's accuracy and tune it accordingly.

[1274] Input: Preprocessed data (training data and test data)

[1275] Output: Highly accurate generative AI model (supply and demand forecast model)

[1276] Step 4: Run a prediction

[1277] The server inputs the newly collected data into the generative AI model to predict future product demand. The prediction results are generated as data showing future demand fluctuations, and also generate detailed analysis results of the factors behind demand fluctuations, taking into account sales data of competitors' products and social media trend data.

[1278] Input: New data (latest data from data collection step)

[1279] Output: Forecast results (future demand fluctuation data and analysis results of demand fluctuation factors)

[1280] Step 5: Output the results

[1281] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, allowing the user to take appropriate action based on the results. For example, the terminal may notify the user of suggested inventory levels or order quantities.

[1282] Input: Prediction results and suggestions (data sent from the server)

[1283] Output: Data displayed in a user interface in the form of graphs and dashboards

[1284] Step 6: Enter promotion information

[1285] Users input sales promotion information (e.g., promotions and price reductions) through their terminals. This information is sent to the server and stored in a database, which is then used to forecast supply and demand for future purchases.

[1286] Input: Sales promotion information (promotions and price reduction information)

[1287] Output: Saved promotion information (stored in database)

[1288] Step 7: Promotion-based forecasting

[1289] The server predicts supply and demand based on sales promotion information entered by the user. For example, it predicts an increase in demand after a Christmas sale based on information about the sale. This prediction result is also sent to the terminal and notified to the user.

[1290] Input: Sales promotion information (obtained from database)

[1291] Output: Prediction results and proposals that take into account sales promotion information (sent from the server to the device)

[1292] Through these steps, the system achieves highly accurate supply and demand forecasts and inventory management.

[1293] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1294] This invention is a "Perfect Inventory Manager" system that uses generative AI to perform highly accurate supply and demand forecasts in inventory management, and by combining it with an emotion engine that recognizes user emotions, it improves the effectiveness of proposals and user satisfaction. This system operates in cooperation with the server, terminals, and users.

[1295] Data collection

[1296] Server Operation

[1297] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social network service APIs. The data is updated regularly and is set to be automatically imported.

[1298] Device behavior

[1299] The device provides an interface for users to operate. Users input information about new products and upcoming promotions into the device. This information is also stored in the database and used to improve the accuracy of the prediction model.

[1300] Data Preprocessing

[1301] Server Operation

[1302] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[1303] Building a demand forecast model

[1304] Server Operation

[1305] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It then trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[1306] Future demand forecast

[1307] Server Operation

[1308] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[1309] Saving and outputting prediction results

[1310] Server Operation

[1311] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[1312] Emotion Recognition and Feedback

[1313] Device behavior

[1314] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and text message content when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server.

[1315] Server Operation

[1316] The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the content and timing of the suggestions, and reflects the user's emotional feedback in optimizing the system.

[1317] Display on the user interface

[1318] Device behavior

[1319] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[1320] Specific examples

[1321] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[1322] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the device of this result and issues an alert to the user that "demand will increase next month, so you should increase your inventory." The device uses an emotion engine to collect the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional feedback, evaluates whether the user is satisfied, and reflects this in the next proposal.

[1323] Example 2: Increased demand due to price reduction and emotional feedback

[1324] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased." The terminal collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[1325] Based on these specific examples, this system can achieve highly accurate supply and demand forecasts, and by taking user emotions into consideration, it can improve the effectiveness of its proposals. As a result, it is possible to prevent overstocks and shortages and improve the efficiency of inventory management.

[1326] The processing flow will be explained below.

[1327] Step 1: Data collection

[1328] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[1329] Step 2: User Data Entry

[1330] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[1331] Step 3: Data Preprocessing

[1332] The server preprocesses the collected data. First, it performs imputation processing using a method to impute missing values ​​using the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and arranges it into a unified format.

[1333] Step 4: Splitting the Dataset

[1334] The server splits the preprocessed data into training data and test data. Specifically, 80% of the data is stored in the database as training data and 20% as test data.

[1335] Step 5: Feature Engineering

[1336] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates for social media trends, and prepares the selected features for input into the generative AI model.

[1337] Step 6: Building a generative AI model

[1338] The server uses machine learning libraries to build generative AI models such as LSTM and Transformer. First, it trains the model using training data to learn patterns. After training is complete, it evaluates the model using test data to check its accuracy and generalization performance.

[1339] Step 7: Forecast future demand

[1340] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[1341] Step 8: Save and output prediction results

[1342] The server stores the forecast results and analysis reports in a database, and calculates appropriate inventory and order quantities based on the stored results and makes recommendations to the user.

[1343] Step 9: Emotion Recognition and Feedback

[1344] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and the content of text messages when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server. The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. The server adjusts the suggestion content and timing according to the level of satisfaction, and reflects the user's emotional feedback in optimizing the system.

[1345] Step 10: Display in the User Interface

[1346] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[1347] Step 11: User Actions

[1348] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocks and shortages.

[1349] Example 2

[1350] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1351] In inventory management, not only is it difficult to make accurate supply and demand forecasts, but the inability to properly consider user emotions and feedback reduces the effectiveness of proposals and user satisfaction.In addition, the inability to fully utilize external data such as the trends of competitors and social media trends prevents the accuracy of supply and demand forecasts from improving.

[1352] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for collecting inventory turnover, sales data, sales data of competitors, and trend information on social network services from information sources; means for preprocessing the collected data, filling in missing data, cleaning the data, standardizing the data, and dividing it into training data and test data; means for constructing a generative AI model based on the preprocessed data and predicting product demand; means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; means for recognizing emotions from user operations, voice, and text messages on the terminal and transmitting the data to the server; means for analyzing the emotion data transmitted to the server and determining user satisfaction; and means for optimizing the content and timing of proposals based on user satisfaction. This not only enables highly accurate supply and demand predictions that take into account the trends of competitors and social networking services trends, but also enables proposals that reflect user emotions, thereby improving inventory management efficiency and user satisfaction.

[1353] The "information source" is a data supplier that provides a variety of data such as inventory turnover, sales data, sales data of competitors, and trend information of social networking services.

[1354] "Inventory turnover" is an indicator of the number of times inventory items are turned over within a certain period of time. It is typically used to evaluate how quickly inventory is sold.

[1355] "Sales data" refers to data that includes information such as the quantity, price, and sales period of products and services sold.

[1356] "Competitor sales data" refers to data on the sales of competitors' products and services in the market. This data is used to understand the competitive landscape and to forecast supply and demand.

[1357] "Social network service trend information" refers to information about popular topics, products, and services on social media. This allows us to understand consumer trends and market trends.

[1358] "Missing data completion" is a process of completing missing values ​​in a dataset by filling in the average value or previous or next values.

[1359] "Data cleaning" is the process of removing outliers and noise from a dataset to improve data quality.

[1360] "Data standardization" is the process of converting data into a unified format to make it consistent, which makes it easier to analyze and train models.

[1361] A "generative artificial intelligence model" is a model that uses machine learning algorithms to predict future situations based on past data. It is also called a generative AI model.

[1362] "User interface" refers to the screens and operating methods that allow users to interact with the system, input information, and check results.

[1363] "Emotion recognition" is the process by which the device analyzes emotions from the user's voice or text messages and determines whether the emotion is positive or negative.

[1364] "Satisfaction" is an index that indicates the degree of satisfaction that a user feels with the services or proposals provided.

[1365] Optimization is the adjustment and improvement of systems and processes to make them operate most efficiently and effectively.

[1366] This invention is a system for implementing highly accurate inventory management for a company. This system operates in cooperation with a server, terminals, and users. The hardware and software used to build the system of this invention include the following elements:

[1367] Hardware and Software

[1368] server

[1369] The server collects data, preprocesses it, builds generative AI models, and stores the prediction results. The server uses the following software:

[1370] Data collection: HTTP client for API communication (e.g., requests library)

[1371] Data preprocessing: Python's pandas and scikit-learn libraries

[1372] Building generative AI models: TensorFlow or PyTorch

[1373] Terminal

[1374] The device provides a user interface and performs emotion recognition. Specifically, it uses the following software:

[1375] User Interface: React.js or Vue.js

[1376] Emotion Recognition: IBM Watson's Natural Language Understanding API

[1377] User

[1378] The user operates the terminal to input the data necessary for inventory management and receives suggestions from the system.

[1379] Specific processing of the system

[1380] Data collection

[1381] The server periodically retrieves inventory turnover, sales data, competitor sales data, and trend information from social networking services from information sources, thereby incorporating specific data that reflects the latest market trends into the system.

[1382] Data Preprocessing

[1383] The collected data undergoes preprocessing, including missing data imputation, data cleaning, and data standardization, to transform it into a consistent dataset, which is then ready for model training.

[1384] Building generative AI models

[1385] Based on the preprocessed data, the server builds a generative AI model using TensorFlow or PyTorch. It uses algorithms specialized for time series data, such as LSTM and Transformer, to predict product demand. After training is complete, the model is evaluated using test data.

[1386] Demand forecasting and proposals

[1387] The server inputs newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are calculated and appropriate inventory and order quantities are calculated. The prediction results are notified to the user via the user interface.

[1388] Emotion recognition and feedback collection

[1389] The device uses an emotion engine to analyze the tone of voice and text message content when the user confirms the proposal, and sends the user's emotional data to the server. The server analyzes the emotional data, determines the user's level of satisfaction, and optimizes the content and timing of the next proposal.

[1390] Specific examples

[1391] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[1392] The server collects sales data and social media trend data from other companies' products and performs data preprocessing. Using a generative AI model, it predicts that demand for the product will increase by 50% next month. This result is notified to the device, and an alert is issued stating, "Demand will increase next month, so you should increase your inventory." The device then uses an emotion engine to collect the user's emotional feedback on this suggestion and sends it to the server. The server analyzes the emotional feedback and reflects it in the next suggestion.

[1393] Example 2: Increased demand due to price reduction and emotional feedback

[1394] The server collects promotional information and obtains information that the product price will be reduced by 10%. Based on this data, it uses a generative AI model to predict that demand will increase by 30% after the price reduction. The server sends the prediction result to the device and provides a notification that "product demand will increase after the price reduction, so production should be increased." The device collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[1395] As described above, the system of the present invention is capable of highly accurate supply and demand forecasting and proposals that take into account the user's emotions, preventing overstocks and shortages and improving the efficiency of inventory management.

[1396] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1397] Step 1: Data collection

[1398] Server Operation

[1399] 1. The server collects data from information sources via API communication. Specifically, it obtains inventory turnover and sales data from the ERP system, real-time sales data from the POS system, and trend information from the API of social media.

[1400] 2. Data inputs include API endpoints and authentication information for each information source, while outputs include collected inventory turnover, sales data, competitor sales data, and social media trend information.

[1401] Specific actions

[1402] The server uses the Python requests library to collect data from each information source, for example by sending an API request like this:

[1403] python

[1404] import requests

[1405] response = requests.get('https: / / api.example.com / inventory')

[1406] data = response.json()

[1407] Step 2: Data Preprocessing

[1408] Server Operation

[1409] 1. Store the collected data in a database.

[1410] 2. The server uses SQL queries to extract missing values ​​from the data and impute them with the average or surrounding values.

[1411] 3. The server performs data cleansing to remove outliers and noise.

[1412] 4. Finally, standardize the data to create a consistent dataset.

[1413] Input and Output

[1414] The input includes the raw data collected in step 1. The output is the formatted data that has been imputed and cleaned.

[1415] Specific actions

[1416] The server uses the Python pandas library to impute and clean the data, for example imputing missing values ​​as follows:

[1417] python

[1418] import pandas as pd

[1419] df = pd.DataFrame(data)

[1420] df.fillna(df.mean(), inplace=True)

[1421] Step 3: Building a generative AI model

[1422] Server Operation

[1423] 1. Using the preprocessed data, we build a generative AI model. Specifically, we define features for the time series data we will use.

[1424] 2. The server uses TensorFlow or PyTorch to train the model using algorithms such as LSTM or Transformer.

[1425] 3. After training, the model is evaluated using test data to check accuracy and generalization performance.

[1426] Input and Output

[1427] The input is a pre-processed dataset, and the output is a trained generative AI model.

[1428] Specific actions

[1429] Perform feature engineering and train an AI model. For example, train an LSTM model with the following code:

[1430] python

[1431] import tensorflow as tf

[1432] from tensorflow.keras.models import Sequential

[1433] from tensorflow.keras.layers import LSTM, Dense

[1434] model = Sequential()

[1435] model.add(LSTM(50, return_sequences=True, input_shape=(n_timesteps, n_features)))

[1436] model.add(LSTM(50))

[1437] model.add(Dense(1))

[1438] model.compile(optimizer='adam', loss='mse')

[1439] model.fit(train_X, train_y, epochs=50, batch_size=72, validation_data=(test_X, test_y))

[1440] Step 4: Forecast future demand

[1441] Server Operation

[1442] 1. Newly collected data is fed into a generative AI model to predict future product demand.

[1443] 2. Based on the demand forecast, a future demand curve is generated and factors affecting demand are analyzed taking into account sales data of competitors and social media trend information.

[1444] Input and Output

[1445] The inputs include new data collected in real time, and the outputs are future demand forecasts and analysis of the factors that cause fluctuations.

[1446] Specific actions

[1447] The server inputs the new data into the model and uses the model's forecasting capabilities to calculate demand. The model forecasts as follows:

[1448] python

[1449] future_predictions = model.predict(new_data)

[1450] Step 5: Save and output prediction results

[1451] Server Operation

[1452] 1. Save the prediction results and analysis reports in a database.

[1453] 2. The server calculates the appropriate inventory and order quantities based on the prediction results and generates a notification to suggest to the user.

[1454] Input and Output

[1455] The input includes future demand forecast results. The output is a saved forecast and a proposed notification.

[1456] Specific actions

[1457] The server stores the prediction results in a MySQL database and generates notifications to send to the device, for example using the following SQL query:

[1458] sql

[1459] INSERT INTO predictions (date, product_id, predicted_demand) VALUES (%s, %s, %s)

[1460] Step 6: Emotion recognition and feedback collection

[1461] Device behavior

[1462] 1. The device analyzes the user's operations, voice, and text messages using an emotion engine and sends the emotion data to the server.

[1463] Server Operation

[1464] 1. The server analyzes the transmitted emotion data and determines the user's satisfaction level.

[1465] 2. Optimize the content and timing of proposals based on customer satisfaction.

[1466] Input and Output

[1467] The input includes user emotion data sent from the device, and the output is a satisfaction analysis result and optimized recommendations.

[1468] Specific actions

[1469] The device uses IBM Watson API to analyze emotions and send the results to the server. For example, call the API as follows:

[1470] python

[1471] from ibm_watson import NaturalLanguageUnderstandingV1

[1472] from ibm_watson.natural_language_understanding_v1 import Features, EmotionOptions

[1473] nlu = NaturalLanguageUnderstandingV1(version='2021-03-25')

[1474] response = nlu.analyze(text='"user input text"', features=Features(emotion=EmotionOptions())).get_result()

[1475] Step 7: Display in the user interface

[1476] Device behavior

[1477] 1. Visually display prediction results and suggestions to the user.

[1478] 2. Present information in an easy-to-understand manner using graphs and dashboards, encouraging users to take appropriate action.

[1479] Input and Output

[1480] The input includes prediction results and suggestion notifications received from the server, and the output is visualized information provided to the user.

[1481] Specific actions

[1482] The terminal uses the React.js or Vue.js framework to display the prediction results in a dashboard format. For example, the display process can be done with the following code:

[1483] javascript

[1484] import React from 'react';

[1485] import { LineChart, Line, CartesianGrid, XAxis, YAxis, Tooltip} from 'recharts';

[1486] const data = [...] / / received prediction data

[1487] const Dashboard = () => (

[1488] <linechart width="{600}" height="{300}" data="{data}">

[1489] <line type="monotone" dataKey="predicted_demand" stroke="8884d8" / >

[1490] <cartesiangrid stroke="ccc" / >

[1491] <xaxis dataKey="date" / >

[1492] <yaxis / >

[1493] <tooltip / >

[1494] < / linechart>

[1495] );

[1496] export default Dashboard;

[1497] Through the above processing steps, companies can realize highly accurate supply and demand forecasts and proposals that take user emotions into consideration.

[1498] (Application example 2)

[1499] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1500] The accuracy of supply and demand forecasts in inventory management is low, which leads to a high risk of overstocking and shortages. In addition, there is a lack of proposals that take user emotions into account, so a method to improve user satisfaction is also needed.

[1501] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: [means for collecting inventory turnover, sales data, sales data of competitors' products, and trend data of social network services from data sources;] [means for preprocessing the collected data, imputing missing values, data cleansing, normalizing, and dividing the collected data into training data and test data;] [means for constructing a generative algorithm model based on the preprocessed data and predicting product demand;] [means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities;] [means for displaying the proposal on a user interface and notifying the user;] [means for analyzing user emotions using an emotion recognition engine and collecting feedback; and [means for optimizing the prediction model based on the feedback and improving the effectiveness of the proposal content.] This enables highly accurate supply and demand prediction and improved user satisfaction.

[1502] "Data Source" refers to information sources that provide inventory turns, sales data, competitor product sales data, and SNS trend data, such as a company's ERP system, POS system, and social network service API.

[1503] "Preprocessing" refers to the process of filling in missing values, cleansing, and normalizing the collected data, and then dividing the data into training data and test data.

[1504] A "generative algorithm model" is a predictive model that is constructed using machine learning algorithms using pre-processed data.

[1505] "Supply and demand forecasting" is a technique that uses generative algorithm models to predict future demand for a product.

[1506] The "user interface" is an interface for visually displaying to the user the proposals and prediction results presented by the server.

[1507] An "emotion recognition engine" is a technology that recognizes and analyzes a user's emotions from the tone of their voice and the content of their text messages.

[1508] "Feedback" refers to data collected about users' reactions and emotions to the proposals, and is used to optimize the predictive model.

[1509] This invention is a system called "Inventory Perfect Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[1510] Data collection

[1511] Server Operation

[1512] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from data sources, utilizing the company's ERP system, POS system, and SNS API. The data is updated periodically and automatically imported.

[1513] Device behavior

[1514] The device provides a user interface, and users input information about new products and promotional plans into the device. The device is also equipped with an emotion recognition engine that recognizes emotions from the tone of voice and content of text messages generated during user operations. This information is also stored in a database and used to improve the accuracy of the predictive model.

[1515] Data Preprocessing

[1516] Server Operation

[1517] The server preprocesses the collected data: missing values ​​are filled with the average or surrounding values, and inaccurate data and noise are removed through data cleansing. The data is then normalized and organized into a unified format.

[1518] Building a supply and demand forecast model

[1519] Server Operation

[1520] The server uses a machine learning library (e.g., TensorFlow, PyTorch) to build a generative algorithm model. Feature engineering is performed to define appropriate features and input them into the model. The model is trained using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[1521] Future demand forecast

[1522] Server Operation

[1523] The server inputs the newly collected data into a generative algorithm model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and trend data from social networking services.

[1524] Saving and outputting prediction results

[1525] Server Operation

[1526] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[1527] Emotion Recognition and Feedback

[1528] Device behavior

[1529] The device is equipped with an emotion recognition engine that recognizes the user's emotions from the tone of voice when the user operates the device and the content of text messages. The emotion recognition engine collects the user's emotion data and sends it to the server.

[1530] Server Operation

[1531] The server analyzes the emotion data sent from the emotion recognition engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the suggestion content and timing, and reflects the user's emotional feedback in optimizing the system.

[1532] Display on the user interface

[1533] Device behavior

[1534] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[1535] Specific examples

[1536] For example, suppose a new product D is scheduled to be released next month. The server collects existing sales data, competitor data, and social media trend information, and uses a generative AI model to predict supply and demand. As a result, the predictive model predicts that demand for product D will increase by 40%. Using emotion recognition feedback, the store manager receives positive feedback that the prediction is "very reliable."

[1537] Prompt Sentence Examples

[1538] The supply and demand forecast for next month shows a 40% increase in demand for new product D. Please arrange the necessary inventory based on this forecast. You can provide feedback, which will be evaluated by the sentiment recognition engine. For example, you can enter feedback such as "The forecast is very reliable."

[1539] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1540] Step 1:

[1541] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources. The input is data from ERP systems, POS systems, SNS APIs, etc., which is collected periodically and stored in a database. The output is the various collected data.

[1542] Step 2:

[1543] The server preprocesses the collected data. Specifically, it imputes missing values ​​with the average or surrounding values, and removes inaccurate data and noise through data cleansing. Next, it normalizes the data, organizes it into a unified format, and splits it into training data and test data. The input is the data collected in step 1, and the output is the preprocessed training data and test data.

[1544] Step 3:

[1545] The server uses machine learning libraries to build a generative algorithm model. Specifically, it uses TensorFlow or PyTorch to extract features from the preprocessed data and trains the model using algorithms such as LSTM or Transformer. The input is the preprocessed data from step 2, and the output is the trained generative algorithm model.

[1546] Step 4:

[1547] The server inputs new data into the generative algorithm model to predict future product supply and demand. The inputs are the newly collected data and the trained model, and the output is the future demand forecast.

[1548] Step 5:

[1549] The server calculates future demand fluctuations based on the forecast results and proposes appropriate inventory and order quantities. The input is the supply and demand forecast results obtained in step 4, and the output is the proposed inventory and order quantities.

[1550] Step 6:

[1551] The terminal displays the prediction results and suggestions sent from the server on a user interface. The input is the prediction results and suggestions sent from the server, and the output is a visual display on the user interface.

[1552] Step 7:

[1553] The user operates the device to input information about new products and promotion plans. Emotional data is also collected from the tone of voice and the content of text messages while operating the device. The input is the user's new product information and emotional data, and the output is the collected data.

[1554] Step 8:

[1555] The emotion recognition engine analyzes the user's emotion data and sends it to the server. The input is the emotion data sent from the device, and the output is the analyzed emotion information.

[1556] Step 9:

[1557] The server analyzes the emotion data sent from the emotion recognition engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the suggestion content and timing and reflects this in the next prediction. The input is the analyzed emotion data, and the output is the adjusted suggestion content.

[1558] Step 10:

[1559] The server finally optimizes the prediction model based on the prediction results and user feedback to improve the effectiveness of the recommendations. The input is the feedback data and the prediction results, and the output is the optimized prediction model.

[1560] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1561] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1562] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1563] [Fourth embodiment]

[1564] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1565] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1566] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1567] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1568] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1569] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1570] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1571] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1572] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1573] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1574] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1575] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1576] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1577] This invention is a system called "Perfect Inventory Manager" that uses generative AI to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[1578] Data collection

[1579] Server Operation

[1580] The server collects product inventory turnover, sales data, competitor product sales data, and SNS trend data from various data sources (e.g., the company's ERP system, POS system, social networking services). The data is monitored to ensure it is updated regularly, and new information is automatically incorporated.

[1581] Device behavior

[1582] The terminal provides an interface for users to operate. Users input data such as information on new products and promotional plans into the terminal, which adds the information necessary for prediction to the database.

[1583] Data Preprocessing

[1584] Server Operation

[1585] The server stores the collected data in a database, imputes missing values ​​in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes the data, and splits the dataset into training data and test data to build a highly accurate model.

[1586] Building a demand forecast model

[1587] Server Operation

[1588] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[1589] Running a prediction

[1590] Server Operation

[1591] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[1592] Output of results

[1593] Server Operation

[1594] The server stores the prediction results in a database, generates reports if necessary, calculates appropriate inventory and order quantities based on the prediction results, and makes suggestions to the user.

[1595] Device behavior

[1596] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[1597] Specific examples

[1598] Example 1: Demand forecast due to rising sales of competitors' products

[1599] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[1600] Example 2: Increased demand due to price reduction

[1601] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[1602] Based on these specific examples, this system can realize highly accurate supply and demand forecasts, prevent overstocks and shortages, and improve the efficiency of inventory management.

[1603] The processing flow will be explained below.

[1604] Step 1: Data collection

[1605] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[1606] Step 2: User Data Entry

[1607] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[1608] Step 3: Data Preprocessing

[1609] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[1610] Step 4: Splitting the Dataset

[1611] The server splits the preprocessed data into training data and test data, storing 80% of the data in the database as training data and 20% as test data.

[1612] Step 5: Feature Engineering

[1613] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates of social media trends, and prepares these features for input into the generative AI model.

[1614] Step 6: Building a generative AI model

[1615] The server uses machine learning libraries to build generative artificial intelligence models such as LSTM and Transformer. First, the model is trained using training data to learn patterns. After training is complete, the model is evaluated using test data to check accuracy and generalization performance.

[1616] Step 7: Forecast future demand

[1617] The server inputs new data into the generative AI model to predict future demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[1618] Step 8: Save and output prediction results

[1619] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[1620] Step 9: Display in the User Interface

[1621] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[1622] Step 10: User Actions

[1623] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocking or shortages.

[1624] Example 1

[1625] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1626] Current inventory management systems are prone to inventory shortages and surpluses due to inaccurate demand forecasts. In addition, inaccuracies in collected data affect the accuracy of forecasts, so there is a need for more efficient data preprocessing. Furthermore, it is difficult to forecast demand while taking into account competitor trends and social media trends, making strategic inventory management difficult. There is a need for a system that can solve these issues and achieve highly accurate inventory management.

[1627] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1628] In this invention, the server includes: [means for collecting inventory turnover, sales data, competitor product sales data, and social network trend data from multiple data sources;] [means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing it, and dividing it into training data and test data; and] [means for constructing a generative artificial intelligence model based on the preprocessed data and predicting product demand.] This enables [highly accurate demand forecasting and efficient inventory management].

[1629] A "data source" is a system or resource for obtaining data from various sources.

[1630] "Inventory turnover" is a value that indicates the number of times an inventory product is sold within a certain period of time.

[1631] "Sales data" is data that includes numerical information about product sales during a specific period.

[1632] "Sales data of competitors' products" refers to data that includes numerical information on sales of competitors' products that exist in the same market.

[1633] "Social network trend data" is data that includes information about trends and topics on online social networking services.

[1634] A "missing value" is a value that indicates a portion of a data set where information is missing.

[1635] "Data cleansing" is the process of removing inaccurate data and noise from a dataset to improve its accuracy.

[1636] "Normalization" is a process that unifies the scale of data to make it easier to compare data with different units or scales.

[1637] "Training data" is a data set used to train an artificial intelligence model.

[1638] "Test data" is a data set used to evaluate the performance of a trained artificial intelligence model.

[1639] A "generative artificial intelligence model" is a machine learning model built on collected and preprocessed data to perform a specific task (e.g., supply and demand forecasting).

[1640] "Demand forecasting" is the prediction of the demand for a product for a specific period in the future.

[1641] "Inventory quantity" refers to the number of products in a warehouse or on a storefront at a particular point in time.

[1642] The "order quantity" is the quantity of the product that should be newly ordered based on the demand forecast.

[1643] The "user interface" refers to the part that includes the means of interaction and screen display that allows the user to interact with the system.

[1644] "Notification" refers to messages or alerts sent by the system to convey information to the user.

[1645] This invention is a system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users.

[1646] System Configuration

[1647] The system mainly consists of the following components:

[1648] 1. Server: Collects data, preprocesses data, builds models, executes predictions, and outputs results.

[1649] 2. Terminal: Provides the interface that the user operates and displays prediction results and suggestions.

[1650] 3. User: Operates the system, inputs the necessary data, and makes decisions based on the prediction results provided.

[1651] Hardware and Software

[1652] The server consists of server equipment with high-performance computing resources (e.g., workstations equipped with GPUs) and uses the following software:

[1653] Operating System: Linux

[1654] Database: MySQL or PostgreSQL

[1655] Programming language: Python

[1656] Machine learning libraries: TensorFlow, PyTorch

[1657] Data processing libraries: pandas, numpy

[1658] The devices consist of common PCs, tablets, smartphones, and other devices, and use the following software:

[1659] Web browser: Google Chrome, Firefox

[1660] Web application framework: Django or Flask

[1661] Data collection

[1662] Server Operation

[1663] The server collects data from multiple data sources, including the company's ERP system, POS system, and social media. Specifically, it uses SQL queries to retrieve inventory turnover and sales data from the ERP system and pulls sales data from the POS system via API. Social media data is collected by using APIs from Twitter and Facebook. All collected data is integrated into a database.

[1664] Data Preprocessing

[1665] Server Operation

[1666] The server performs data preprocessing on the collected data. It imputes missing values ​​with the mean or median, detects and removes outliers, and performs normalization processes such as standardizing date formats and converting data types. It also splits the dataset into training data and test data.

[1667] Building a demand forecast model

[1668] Server Operation

[1669] The server uses machine learning libraries such as TensorFlow and PyTorch to build generative AI models. It performs feature engineering to define features suitable for model input. It trains the model using algorithms suitable for time-series data, such as LSTM and Transformer. After training, it evaluates the model's performance and tunes hyperparameters as necessary.

[1670] Running a prediction

[1671] Server Operation

[1672] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[1673] Output of results

[1674] Server Operation

[1675] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[1676] Device behavior

[1677] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[1678] Specific examples

[1679] Example 1: Demand forecast due to rising sales of competitors' products

[1680] The server collects sales data and social media trend data from other companies' similar products, performs data preprocessing, and then uses a generative AI model to predict that demand for Product A will increase by 50% next month. The server notifies the terminal of this result and alerts the user that "demand will increase next month, so inventory should be increased."

[1681] Example 2: Increased demand due to price reduction

[1682] The server collects promotion information from the POS system and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased."

[1683] This allows users to manage their inventory appropriately and prevent shortages and surpluses.

[1684] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1685] Step 1: Data collection

[1686] Server Operation

[1687] The server collects the necessary data from multiple data sources (e.g., the company's ERP system, POS system, and social media). Specifically, it uses SQL queries to obtain inventory turnover and sales data from the ERP system, and sales data from the POS system via API. Social media data is collected through APIs from Twitter and Facebook.

[1688] Specific operations and inputs / outputs

[1689] Input: SQL query, API request

[1690] Data manipulation: Pulling data from ERP systems, getting data from POS systems, collecting data from social media APIs

[1691] Output: Consolidated dataset in a database

[1692] Step 2: Data Preprocessing

[1693] Server Operation

[1694] The server performs data preprocessing on the collected data. Specifically, it imputes missing values ​​with the mean or median, and detects and removes outliers. It also performs normalization processes such as standardizing date formats and converting data types. Next, it splits the dataset into training data and test data.

[1695] Specific operations and inputs / outputs

[1696] Input: Integrated dataset

[1697] Data manipulation: Imputing missing values ​​(using Python's pandas library), removing outliers, normalizing (scaling) the data, and splitting the data (using scikit-learn's train_test_split function).

[1698] Output: Preprocessed training and test data

[1699] Step 3: Build a demand forecast model

[1700] Server Operation

[1701] The server uses machine learning libraries (TensorFlow, PyTorch) to build a generative AI model. First, feature engineering is performed to define features suitable for the model. Then, the model is trained using algorithms specialized for time-series data, such as LSTM and Transformer, and the model's performance is evaluated. Hyperparameter tuning is also performed as necessary.

[1702] Specific operations and inputs / outputs

[1703] Input: Training data

[1704] Data manipulation: feature engineering (creating new features), model training (building an LSTM model using TensorFlow), model evaluation (calculating precision, recall, and F-measure)

[1705] Output: A trained generative AI model

[1706] Step 4: Run a prediction

[1707] Server Operation

[1708] The server inputs the newly collected data into a generative AI model to predict future demand. The forecast uses new sales data and social media trend data. It also takes into account sales data from competitors' products to analyze fluctuations in demand.

[1709] Specific operations and inputs / outputs

[1710] Input: Newly collected data, trained generative AI model

[1711] Data manipulation: Inputting data into the model, running demand forecasts (predicting future demand through the model)

[1712] Output: Forecast result (future demand curve)

[1713] Step 5: Output the results

[1714] Server Operation

[1715] The server stores the forecast results in a database, generates reports as needed, calculates appropriate inventory and order quantities based on the forecast results, and creates proposals for users.

[1716] Specific operations and inputs / outputs

[1717] Input: Prediction result

[1718] Data manipulation: Prediction results saved in database, report generation (automatic PDF report generation)

[1719] Output: Report, proposals

[1720] Device behavior

[1721] The device displays the prediction results and suggestions sent from the server, presenting the results in a visually easy-to-understand graph or dashboard format and prompting the user to take appropriate action.

[1722] Specific operations and inputs / outputs

[1723] Input: Prediction results and suggestions sent from the server

[1724] Data manipulation: visualization (display in graphs and dashboards)

[1725] Output: Display on the user interface, notifications

[1726] The above processing steps enable highly accurate supply and demand forecasts and efficient inventory management.

[1727] (Application example 1)

[1728] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1729] Product demand forecasts are extremely important in modern inventory management, but achieving highly accurate forecasts remains difficult. In particular, it is difficult to integrate information from diverse data sources and perform detailed analysis of factors that cause demand fluctuations, and the uncertainty of forecasts is likely to have a negative impact on corporate management. Furthermore, while there is a demand for timely supply and demand forecasts that reflect promotions and price reduction information, there is a lack of technology to achieve this. Therefore, there is a need for a system that can solve these problems and achieve highly accurate supply and demand forecasts and inventory management.

[1730] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1731] In this invention, the server includes: means for collecting inventory turnover, sales data, sales data on competitors' products, and trend data on social network services from data sources; means for preprocessing the collected data, imputing missing values, cleansing the data, normalizing the data, and dividing it into training data and test data; means for constructing a generative artificial intelligence model based on the preprocessed data and forecasting product demand; means for calculating future demand fluctuations based on the forecast results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; and means for providing the user with sales promotion information and making supply and demand forecasts based on that information. This enables companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[1732] "Data Source" refers to an information source that provides data necessary for supply and demand forecasting, including inventory turns, sales data, sales data of competitors' products, and trend data of social network services.

[1733] "Inventory turnover" is an index that indicates the frequency of product inventory replacement within a certain period of time.

[1734] "Sales data" refers to data relating to the number of products sold and the amount of sales within a certain period of time.

[1735] "Sales data of competitors' products" refers to data on the sales performance of competitors' products that are similar to or belong to the same market as our company's products.

[1736] "Social network service trend data" refers to data on popular topics and topics of interest extracted from user activities and posts on platforms such as social networking sites.

[1737] "Missing value imputation" is the process of filling in missing values ​​in a dataset in an appropriate way.

[1738] "Data cleansing" is the process of removing inaccurate or unnecessary information from a data set.

[1739] "Normalization" is a preprocessing process that improves the learning effect of an algorithm by constraining data within a certain range.

[1740] "Training data" is the dataset used to train a machine learning model.

[1741] "Test data" is a dataset used to evaluate the performance of a machine learning model.

[1742] A "generative artificial intelligence model" is a supply and demand forecasting model built using machine learning algorithms based on collected and preprocessed data.

[1743] "Prediction results" are data and information regarding future demand calculated by the generated artificial intelligence model.

[1744] "Future demand fluctuations" refers to trends that indicate future increases or decreases in demand based on forecast results.

[1745] "Inventory" is the quantity of a product in stock at a particular point in time.

[1746] "Order quantity" refers to the quantity of the product to be newly ordered to meet future demand.

[1747] A "user interface" is a screen or operating means through which a system interacts with a user.

[1748] "Sales promotion information" is information about measures to increase sales, such as product promotions and price reductions.

[1749] This invention is a "Perfect Inventory Manager" system that uses generative AI models to perform highly accurate supply and demand forecasts in inventory management. This system operates in cooperation with a server, terminals, and users. The details of the system's operation are described below.

[1750] Data collection

[1751] Server Operation

[1752] The server collects inventory turnover, sales data, competitor product sales data, and social networking service trend data from various data sources, such as the company's ERP system, POS system, and social networking API platforms. The data is monitored for regular updates, and new information is automatically incorporated.

[1753] User Actions

[1754] Users can input data such as sales promotion information, new product information, and promotion schedules through their devices, and the information provided by users is also used to improve the accuracy of the prediction model.

[1755] Data Preprocessing

[1756] Server Operation

[1757] The server stores the collected data in a database and imputes missing values ​​in an appropriate manner. It also performs data cleansing to remove inaccurate data and noise and normalizes the data. Furthermore, it splits the dataset into training data and test data to build a highly accurate predictive model. This process utilizes data integration tools and the Python pandas library.

[1758] Building a demand forecast model

[1759] Server Operation

[1760] The server builds a generative AI model using machine learning libraries such as TensorFlow and PyTorch. It performs feature engineering and inputs appropriate features into the model. It trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, it evaluates the model and tunes it as necessary.

[1761] Running a prediction

[1762] Server Operation

[1763] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are analyzed. At the same time, factors behind demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[1764] Output of results

[1765] Device behavior

[1766] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, prompting the user to take appropriate action. For example, the terminal notifies the user of suggested inventory levels and order quantities.

[1767] Specific examples

[1768] When a user inputs new promotional information into the device, the system predicts changes in demand based on that information. For example, if the user inputs information like "20% off all products in the Christmas sale starting December 1, 2023," the generative AI model predicts an increase in sales for the next month and alerts the user that "demand will increase next month, so you should increase your inventory."

[1769] Prompt Sentence Examples

[1770] New promotion information:

[1771] Event: Christmas Sale

[1772] Start date: December 1, 2023

[1773] Discount: 20%

[1774] Predict the percentage increase in sales that this promotion will bring

[1775] This allows companies to comprehensively analyze a variety of data and make highly accurate supply and demand forecasts. It also enables timely responses to promotions and price reductions, effectively preventing inventory surpluses and shortages.

[1776] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1777] Step 1: Data collection

[1778] The server collects inventory turnover, sales data, competitor product sales data, and social network service trend data from data sources (e.g., a company's ERP system, POS system, or SNS API). The collected data is sent to the server in JSON or CSV format and stored in a database. This provides a variety of data necessary for demand forecasting.

[1779] Input: ERP system, POS system, SNS API, etc.

[1780] Output: Inventory turnover, sales data, competitor product sales data, SNS trend data (stored in database)

[1781] Step 2: Data Preprocessing

[1782] The server imputes missing values ​​in the collected data in an appropriate way, performs data cleansing to remove inaccurate data and noise, normalizes (scales) the data to the range of 0 to 1, and splits it into training data and test data. This process uses Python's pandas library and scikit-learn's MinMaxScaler.

[1783] Input: Collected raw data (taken from database)

[1784] Output: Preprocessed data (scaled data after missing value imputation and data cleansing)

[1785] Step 3: Building a generative AI model

[1786] The server uses the preprocessed data to build a generative AI model. It uses TensorFlow and PyTorch to train a time-series prediction model using algorithms such as LSTM and Transformer. It separates the training data from the test data to evaluate the model's accuracy and tune it accordingly.

[1787] Input: Preprocessed data (training data and test data)

[1788] Output: Highly accurate generative AI model (supply and demand forecast model)

[1789] Step 4: Run a prediction

[1790] The server inputs the newly collected data into the generative AI model to predict future product demand. The prediction results are generated as data showing future demand fluctuations, and also generate detailed analysis results of the factors behind demand fluctuations, taking into account sales data of competitors' products and social media trend data.

[1791] Input: New data (latest data from data collection step)

[1792] Output: Forecast results (future demand fluctuation data and analysis results of demand fluctuation factors)

[1793] Step 5: Output the results

[1794] The terminal displays the prediction results and suggestions sent from the server on a user interface. The results are displayed in a visually easy-to-understand graph or dashboard format, allowing the user to take appropriate action based on the results. For example, the terminal may notify the user of suggested inventory levels or order quantities.

[1795] Input: Prediction results and suggestions (data sent from the server)

[1796] Output: Data displayed in a user interface in the form of graphs and dashboards

[1797] Step 6: Enter promotion information

[1798] Users input sales promotion information (e.g., promotions and price reductions) through their terminals. This information is sent to the server and stored in a database, which is then used to forecast supply and demand for future purchases.

[1799] Input: Sales promotion information (promotions and price reduction information)

[1800] Output: Saved promotion information (stored in database)

[1801] Step 7: Promotion-based forecasting

[1802] The server predicts supply and demand based on sales promotion information entered by the user. For example, it predicts an increase in demand after a Christmas sale based on information about the sale. This prediction result is also sent to the terminal and notified to the user.

[1803] Input: Sales promotion information (obtained from database)

[1804] Output: Prediction results and proposals that take into account sales promotion information (sent from the server to the device)

[1805] Through these steps, the system achieves highly accurate supply and demand forecasts and inventory management.

[1806] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1807] This invention is a "Perfect Inventory Manager" system that uses generative AI to perform highly accurate supply and demand forecasts in inventory management, and by combining it with an emotion engine that recognizes user emotions, it improves the effectiveness of proposals and user satisfaction. This system operates in cooperation with the server, terminals, and users.

[1808] Data collection

[1809] Server Operation

[1810] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social network service APIs. The data is updated regularly and is set to be automatically imported.

[1811] Device behavior

[1812] The device provides an interface for users to operate. Users input information about new products and upcoming promotions into the device. This information is also stored in the database and used to improve the accuracy of the prediction model.

[1813] Data Preprocessing

[1814] Server Operation

[1815] The server preprocesses the collected data. First, it imputes missing values ​​using a method to impute the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and organizes it into a unified format.

[1816] Building a demand forecast model

[1817] Server Operation

[1818] The server builds a generative AI model using a machine learning library (e.g., TensorFlow, PyTorch). It performs feature engineering to define appropriate features and input them into the model. It then trains the model using algorithms specialized for time-series data, such as LSTM and Transformer. After training is complete, the model is evaluated using test data to check its accuracy and generalization performance.

[1819] Future demand forecast

[1820] Server Operation

[1821] The server inputs the newly collected data into the generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated. At the same time, factors that influence demand fluctuations are analyzed, taking into account sales data of competitors' products and social media trend data.

[1822] Saving and outputting prediction results

[1823] Server Operation

[1824] The server stores the prediction results and analysis reports in a database. Based on the stored results, it calculates appropriate inventory and order quantities and makes suggestions to the user.

[1825] Emotion Recognition and Feedback

[1826] Device behavior

[1827] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and text message content when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server.

[1828] Server Operation

[1829] The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. Depending on the level of satisfaction, the server adjusts the content and timing of the suggestions, and reflects the user's emotional feedback in optimizing the system.

[1830] Display on the user interface

[1831] Device behavior

[1832] The device displays the prediction results and suggestions sent from the server on a user interface, visually displaying them in easy-to-understand graphs and dashboard formats, and prompting the user to take appropriate action.

[1833] Specific examples

[1834] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[1835] To predict demand for a certain product A, the server collects sales data and social media trend data for a similar product B from another company. After preprocessing the data, a generative artificial intelligence model is used to predict that demand for product A will increase by 50% next month. The server notifies the device of this result and issues an alert to the user that "demand will increase next month, so you should increase your inventory." The device uses an emotion engine to collect the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional feedback, evaluates whether the user is satisfied, and reflects this in the next proposal.

[1836] Example 2: Increased demand due to price reduction and emotional feedback

[1837] The server collects promotion information and obtains the date when the price of product C will be reduced by 10%. Based on this data, it uses a generative artificial intelligence model to predict that demand will increase by 30% after the price reduction. The server sends this prediction result to the terminal and notifies the user that "demand for product C will increase after the price reduction, so production should be increased." The terminal collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[1838] Based on these specific examples, this system can achieve highly accurate supply and demand forecasts, and by taking user emotions into consideration, it can improve the effectiveness of its proposals. As a result, it is possible to prevent overstocks and shortages and improve the efficiency of inventory management.

[1839] The processing flow will be explained below.

[1840] Step 1: Data collection

[1841] The server collects inventory turnover, sales data, competitor product sales data, and SNS trend data from data sources such as the company's ERP system, POS system, and social networking service APIs. The data is updated regularly and is set to be automatically imported.

[1842] Step 2: User Data Entry

[1843] Users can input new product information and promotional plans through the device, which is then stored in a database to help improve the accuracy of the prediction model.

[1844] Step 3: Data Preprocessing

[1845] The server preprocesses the collected data. First, it performs imputation processing using a method to impute missing values ​​using the average or surrounding values. Then it performs data cleansing to remove inaccurate data and noise. Finally, it normalizes the data and arranges it into a unified format.

[1846] Step 4: Splitting the Dataset

[1847] The server splits the preprocessed data into training data and test data. Specifically, 80% of the data is stored in the database as training data and 20% as test data.

[1848] Step 5: Feature Engineering

[1849] The server selects features suitable for data analysis, such as sales data, inventory turnover, and engagement rates for social media trends, and prepares the selected features for input into the generative AI model.

[1850] Step 6: Building a generative AI model

[1851] The server uses machine learning libraries to build generative AI models such as LSTM and Transformer. First, it trains the model using training data to learn patterns. After training is complete, it evaluates the model using test data to check its accuracy and generalization performance.

[1852] Step 7: Forecast future demand

[1853] The server inputs the newly collected data into a generative AI model to predict future product demand. Based on this prediction, a future demand curve is generated and visualized as a graph. At the same time, it takes into account sales data of competitors' products and social media trend data to analyze factors behind fluctuations in demand.

[1854] Step 8: Save and output prediction results

[1855] The server stores the forecast results and analysis reports in a database, and calculates appropriate inventory and order quantities based on the stored results and makes recommendations to the user.

[1856] Step 9: Emotion Recognition and Feedback

[1857] The device is equipped with an emotion engine that recognizes emotions from the tone of voice and the content of text messages when the user operates the device. The emotion engine collects the user's emotion data and sends it to the server. The server analyzes the emotion data sent from the emotion engine and determines whether the user is satisfied with the suggestions. The server adjusts the suggestion content and timing according to the level of satisfaction, and reflects the user's emotional feedback in optimizing the system.

[1858] Step 10: Display in the User Interface

[1859] The device displays the prediction results and suggestions sent from the server on a user interface, presenting the results in easy-to-understand graphs and dashboards, and prompting the user to take appropriate action.

[1860] Step 11: User Actions

[1861] Based on the information provided by the terminal, users can take necessary actions such as adjusting inventory or placing orders, thereby ensuring a balance between supply and demand and preventing overstocks and shortages.

[1862] Example 2

[1863] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1864] In inventory management, not only is it difficult to make accurate supply and demand forecasts, but the inability to properly consider user emotions and feedback reduces the effectiveness of proposals and user satisfaction.In addition, the inability to fully utilize external data such as the trends of competitors and social media trends prevents the accuracy of supply and demand forecasts from improving.

[1865] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for collecting inventory turnover, sales data, sales data of competitors, and trend information on social network services from information sources; means for preprocessing the collected data, filling in missing data, cleaning the data, standardizing the data, and dividing it into training data and test data; means for constructing a generative AI model based on the preprocessed data and predicting product demand; means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities; means for displaying the proposal on a user interface and notifying the user; means for recognizing emotions from user operations, voice, and text messages on the terminal and transmitting the data to the server; means for analyzing the emotion data transmitted to the server and determining user satisfaction; and means for optimizing the content and timing of proposals based on user satisfaction. This not only enables highly accurate supply and demand predictions that take into account the trends of competitors and social networking services trends, but also enables proposals that reflect user emotions, thereby improving inventory management efficiency and user satisfaction.

[1866] The "information source" is a data supplier that provides a variety of data such as inventory turnover, sales data, sales data of competitors, and trend information of social networking services.

[1867] "Inventory turnover" is an indicator of the number of times inventory items are turned over within a certain period of time. It is typically used to evaluate how quickly inventory is sold.

[1868] "Sales data" refers to data that includes information such as the quantity, price, and sales period of products and services sold.

[1869] "Competitor sales data" refers to data on the sales of competitors' products and services in the market. This data is used to understand the competitive landscape and to forecast supply and demand.

[1870] "Social network service trend information" refers to information about popular topics, products, and services on social media. This allows us to understand consumer trends and market trends.

[1871] "Missing data completion" is a process of completing missing values ​​in a dataset by filling in the average value or previous or next values.

[1872] "Data cleaning" is the process of removing outliers and noise from a dataset to improve data quality.

[1873] "Data standardization" is the process of converting data into a unified format to make it consistent, which makes it easier to analyze and train models.

[1874] A "generative artificial intelligence model" is a model that uses machine learning algorithms to predict future situations based on past data. It is also called a generative AI model.

[1875] "User interface" refers to the screens and operating methods that allow users to interact with the system, input information, and check results.

[1876] "Emotion recognition" is the process by which the device analyzes emotions from the user's voice or text messages and determines whether the emotion is positive or negative.

[1877] "Satisfaction" is an index that indicates the degree of satisfaction that a user feels with the services or proposals provided.

[1878] Optimization is the adjustment and improvement of systems and processes to make them operate most efficiently and effectively.

[1879] This invention is a system for implementing highly accurate inventory management for a company. This system operates in cooperation with a server, terminals, and users. The hardware and software used to build the system of this invention include the following elements:

[1880] Hardware and Software

[1881] server

[1882] The server collects data, preprocesses it, builds generative AI models, and stores the prediction results. The server uses the following software:

[1883] Data collection: HTTP client for API communication (e.g., requests library)

[1884] Data preprocessing: Python's pandas and scikit-learn libraries

[1885] Building generative AI models: TensorFlow or PyTorch

[1886] Terminal

[1887] The device provides a user interface and performs emotion recognition. Specifically, it uses the following software:

[1888] User Interface: React.js or Vue.js

[1889] Emotion Recognition: IBM Watson's Natural Language Understanding API

[1890] User

[1891] The user operates the terminal to input the data necessary for inventory management and receives suggestions from the system.

[1892] Specific processing of the system

[1893] Data collection

[1894] The server periodically retrieves inventory turnover, sales data, competitor sales data, and trend information from social networking services from information sources, thereby incorporating specific data that reflects the latest market trends into the system.

[1895] Data Preprocessing

[1896] The collected data undergoes preprocessing, including missing data imputation, data cleaning, and data standardization, to transform it into a consistent dataset, which is then ready for model training.

[1897] Building generative AI models

[1898] Based on the preprocessed data, the server builds a generative AI model using TensorFlow or PyTorch. It uses algorithms specialized for time series data, such as LSTM and Transformer, to predict product demand. After training is complete, the model is evaluated using test data.

[1899] Demand forecasting and proposals

[1900] The server inputs newly collected data into a generative AI model to predict future product demand. Based on this prediction, future demand fluctuations are calculated and appropriate inventory and order quantities are calculated. The prediction results are notified to the user via the user interface.

[1901] Emotion recognition and feedback collection

[1902] The device uses an emotion engine to analyze the tone of voice and text message content when the user confirms the proposal, and sends the user's emotional data to the server. The server analyzes the emotional data, determines the user's level of satisfaction, and optimizes the content and timing of the next proposal.

[1903] Specific examples

[1904] Example 1: Demand forecast and emotional feedback due to rising sales of competitors' products

[1905] The server collects sales data and social media trend data from other companies' products and performs data preprocessing. Using a generative AI model, it predicts that demand for the product will increase by 50% next month. This result is notified to the device, and an alert is issued stating, "Demand will increase next month, so you should increase your inventory." The device then uses an emotion engine to collect the user's emotional feedback on this suggestion and sends it to the server. The server analyzes the emotional feedback and reflects it in the next suggestion.

[1906] Example 2: Increased demand due to price reduction and emotional feedback

[1907] The server collects promotional information and obtains information that the product price will be reduced by 10%. Based on this data, it uses a generative AI model to predict that demand will increase by 30% after the price reduction. The server sends the prediction result to the device and provides a notification that "product demand will increase after the price reduction, so production should be increased." The device collects the user's emotional feedback on this proposal and sends it to the server. The server analyzes the emotional data, evaluates the effectiveness of the proposal, and reflects it in system optimization.

[1908] As described above, the system of the present invention is capable of highly accurate supply and demand forecasting and proposals that take into account the user's emotions, preventing overstocks and shortages and improving the efficiency of inventory management.

[1909] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1910] Step 1: Data collection

[1911] Server Operation

[1912] 1. The server collects data from information sources via API communication. Specifically, it obtains inventory turnover and sales data from the ERP system, real-time sales data from the POS system, and trend information from the API of social media.

[1913] 2. Data inputs include API endpoints and authentication information for each information source, while outputs include collected inventory turnover, sales data, competitor sales data, and social media trend information.

[1914] Specific actions

[1915] The server uses the Python requests library to collect data from each information source, for example by sending an API request like this:

[1916] python

[1917] import requests

[1918] response = requests.get('https: / / api.example.com / inventory')

[1919] data = response.json()

[1920] Step 2: Data Preprocessing

[1921] Server Operation

[1922] 1. Store the collected data in a database.

[1923] 2. The server uses SQL queries to extract missing values ​​from the data and impute them with the average or surrounding values.

[1924] 3. The server performs data cleansing to remove outliers and noise.

[1925] 4. Finally, standardize the data to create a consistent dataset.

[1926] Input and Output

[1927] The input includes the raw data collected in step 1. The output is the formatted data that has been imputed and cleaned.

[1928] Specific actions

[1929] The server uses the Python pandas library to impute and clean the data, for example imputing missing values ​​as follows:

[1930] python

[1931] import pandas as pd

[1932] df = pd.DataFrame(data)

[1933] df.fillna(df.mean(), inplace=True)

[1934] Step 3: Building a generative AI model

[1935] Server Operation

[1936] 1. Using the preprocessed data, we build a generative AI model. Specifically, we define features for the time series data we will use.

[1937] 2. The server uses TensorFlow or PyTorch to train the model using algorithms such as LSTM or Transformer.

[1938] 3. After training, the model is evaluated using test data to check accuracy and generalization performance.

[1939] Input and Output

[1940] The input is a pre-processed dataset, and the output is a trained generative AI model.

[1941] Specific actions

[1942] Perform feature engineering and train an AI model. For example, train an LSTM model with the following code:

[1943] python

[1944] import tensorflow as tf

[1945] from tensorflow.keras.models import Sequential

[1946] from tensorflow.keras.layers import LSTM, Dense

[1947] model = Sequential()

[1948] model.add(LSTM(50, return_sequences=True, input_shape=(n_timesteps, n_features)))

[1949] model.add(LSTM(50))

[1950] model.add(Dense(1))

[1951] model.compile(optimizer='adam', loss='mse')

[1952] model.fit(train_X, train_y, epochs=50, batch_size=72, validation_data=(test_X, test_y))

[1953] Step 4: Forecast future demand

[1954] Server Operation

[1955] 1. Newly collected data is fed into a generative AI model to predict future product demand.

[1956] 2. Based on the demand forecast, a future demand curve is generated and factors affecting demand are analyzed taking into account sales data of competitors and social media trend information.

[1957] Input and Output

[1958] The inputs include new data collected in real time, and the outputs are future demand forecasts and analysis of the factors that cause fluctuations.

[1959] Specific actions

[1960] The server inputs the new data into the model and uses the model's forecasting capabilities to calculate demand. The model forecasts as follows:

[1961] python

[1962] future_predictions = model.predict(new_data)

[1963] Step 5: Save and output prediction results

[1964] Server Operation

[1965] 1. Save the prediction results and analysis reports in a database.

[1966] 2. The server calculates the appropriate inventory and order quantities based on the prediction results and generates a notification to suggest to the user.

[1967] Input and Output

[1968] The input includes future demand forecast results. The output is a saved forecast and a proposed notification.

[1969] Specific actions

[1970] The server stores the prediction results in a MySQL database and generates notifications to send to the device, for example using the following SQL query:

[1971] sql

[1972] INSERT INTO predictions (date, product_id, predicted_demand) VALUES (%s, %s, %s)

[1973] Step 6: Emotion recognition and feedback collection

[1974] Device behavior

[1975] 1. The device analyzes the user's operations, voice, and text messages using an emotion engine and sends the emotion data to the server.

[1976] Server Operation

[1977] 1. The server analyzes the transmitted emotion data and determines the user's satisfaction level.

[1978] 2. Optimize the content and timing of proposals based on customer satisfaction.

[1979] Input and Output

[1980] The input includes user emotion data sent from the device, and the output is a satisfaction analysis result and optimized recommendations.

[1981] Specific actions

[1982] The device uses IBM Watson API to analyze emotions and send the results to the server. For example, call the API as follows:

[1983] python

[1984] from ibm_watson import NaturalLanguageUnderstandingV1

[1985] from ibm_watson.natural_language_understanding_v1 import Features, EmotionOptions

[1986] nlu = NaturalLanguageUnderstandingV1(version='2021-03-25')

[1987] response = nlu.analyze(text='"user input text"', features=Features(emotion=EmotionOptions())).get_result()

[1988] Step 7: Display in the user interface

[1989] Device behavior

[1990] 1. Visually display prediction results and suggestions to the user.

[1991] 2. Present information in an easy-to-understand manner using graphs and dashboards, encouraging users to take appropriate action.

[1992] Input and Output

[1993] The input includes prediction results and suggestion notifications received from the server, and the output is visualized information provided to the user.

[1994] Specific actions

[1995] The terminal uses the React.js or Vue.js framework to display the prediction results in a dashboard format. For example, the display process can be done with the following code:

[1996] javascript

[1997] import React from 'react';

[1998] import { LineChart, Line, CartesianGrid, XAxis, YAxis, Tooltip} from 'recharts';

[1999] const data = [...] / / received prediction data

[2000] const Dashboard = () => (

[2001] <linechart width="{600}" height="{300}" data="{data}">

[2002] <line type="monotone" dataKey="predicted_demand" stroke="8884d8" / >

[2003] <cartesiangrid stroke="ccc" / >

[2004] <xaxis dataKey="date" / >

[2005] <yaxis / >

[2006] <tooltip / >

[2007] < / linechart>

[2008] );

[2009] export default Dashboard;

[2010] Through the above processing steps, companies can realize highly accurate supply and demand forecasts and proposals that take user emotions into consideration.

[2011] (Application example 2)

[2012] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[2013] The accuracy of supply and demand forecasts in inventory management is low, which leads to a high risk of overstocking and shortages. In addition, there is a lack of proposals that take user emotions into account, so a method to improve user satisfaction is also needed.

[2014] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: [means for collecting inventory turnover, sales data, sales data of competitors' products, and trend data of social network services from data sources;] [means for preprocessing the collected data, imputing missing values, data cleansing, normalizing, and dividing the collected data into training data and test data;] [means for constructing a generative algorithm model based on the preprocessed data and predicting product demand;] [means for calculating future demand fluctuations based on the prediction results and proposing appropriate inventory and order quantities;] [means for displaying the proposal on a user interface and notifying the user;] [means for analyzing user emotions using an emotion recognition engine and collecting feedback; and [means for optimizing the prediction model based on the feedback and improving the effectiveness of the proposal content.] This enables highly accurate supply and demand prediction and improved user satisfaction.

[2015] "Data Source" refers to information sources that provide inventory turns, ...

Claims

1. a means for collecting inventory turns, sales data, competitor product sales data, and social network service trend data from data sources; A means to preprocess the collected data, impute missing values, cleanse the data, normalize it, and split it into training data and test data; A means for constructing a generative artificial intelligence model based on preprocessed data to forecast product demand; A method for calculating future demand fluctuations based on the forecast results and proposing appropriate inventory and order quantities; The system includes means for displaying the suggestions in a user interface and notifying the user.

2. 2. The system according to claim 1, wherein sales data of competitors' products and trend data of social network services are input into the model to perform a detailed analysis of factors that cause fluctuations in demand.

3. 2. The system according to claim 1, wherein the system calculates future increases in demand based on information about price reductions and promotions, and proposes appropriate production and inventory amounts to the user.

Images