Electronic device for generating feature data for prediction of business outcome based on ontology and method for generating feature data

Abstract

To provide a technology for predicting management or the like of a business risk.SOLUTION: According to an aspect of the present disclosure, there is provided an electronic apparatus for generating feature data for predicting a business result based on an ontology, the electronic apparatus including a processor configured to generate first data by extracting a feature or a pattern from ERP data based on an analysis, generate second data by extracting a feature or a pattern from the ERP data based on the ontology, generate third data by extracting an external variable at a first time point, and create a word collection suitable for a business domain from merged data obtained by merging the first to third data; A Biz2Vec model that learns syntax and context of a business domain may be generated using the merged dataset and the word collection, a vector space suitable for a business context may be configured using the Biz2Vec model, a similarity between vectors may be calculated in the vector space, and a relationship between objects may be quantified.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to a technology for predicting business risks using ERP data, and more particularly to a method for generating feature data for business performance prediction using an ontology and a vector space. [Background technology]

[0002] Enterprise resource planning (ERP) systems are widely used as tools for centrally collecting and analyzing a variety of data to manage a company's business activities in an integrated manner. ERP systems generally integrate data from various departments, such as finance, human resources, supply chain management, and production, and analyze it in real time, thereby improving corporate efficiency and supporting decision-making. However, traditional ERP systems primarily provide data in the form of simple reports or focus on basic trend analysis, limiting their ability to predict and respond to complex business processes and internal and external variability in real time.

[0003] Ontology is a method for structuring and representing concepts and relationships within a specific domain, and is commonly used in knowledge-based systems and artificial intelligence technologies. Data analysis using ontologies is advantageous for grasping semantic relationships between data and supports data analysis and decision-making processes by hierarchically representing complex data structures. However, ontology-based analysis is generally used primarily to structure unstructured data or improve the accuracy of information retrieval, and does not provide advanced functions for analyzing business flows in real time and predicting risks based on the results.

[0004] Today's corporate environment is complex and rapidly changing, making it important to respond in real time to not only internal business processes but also external economic indicators and market changes. Traditional ERP systems and ontology-based analysis tools are useful for integrating and analyzing data, but they have limitations in tracking and predicting business processes in real time. In particular, the ability to predict business risks in advance and effectively manage them is essential to strengthening a company's competitiveness.

[0005] Therefore, an advanced method is required to predict business risks by combining ERP data and ontology to quantify the impact of external variables that change in real time and generate feature data that includes numerical data. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Korean Patent No. 10-1677106B1 "Ontology-based knowledge inference collaborative system" Summary of the Invention [Problem to be solved by the invention]

[0007] The present disclosure has been made in consideration of the above circumstances, and its purpose is to provide an electronic device that provides a more efficient and accurate solution than conventional methods in managing business risks and predicting operational results by combining ontology and ERP data, reflecting external variables in real time, and improving machine learning performance.

[0008] However, the problems that the present disclosure aims to solve are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0009] According to the present disclosure, an electronic device for generating feature data for predicting business outcomes based on an ontology to achieve the above technical objectives may include a memory storing at least one instruction code and at least one processor executing the at least one instruction code, wherein the at least one processor generates first data by extracting features and patterns from ERP data based on an analysis, generates second data by extracting features and patterns from the ERP data based on an ontology, generates third data by extracting external variables at a first time point, generates merged data by merging the first, second, and third data, generates a vocabulary suitable for a business domain from the merged data, generates a Biz2Vec model that learns syntax and context of the business domain using the merged data and the vocabulary, constructs a vector space suitable for the business context using the Biz2Vec model, calculates similarities between vectors in the vector space, and quantifies relationships between objects using cosine similarity or Euclidean distance between each vector, thereby generating the feature data.

[0010] In one embodiment, the at least one processor calculates a weighted value that combines the relative similarity of each vector and the distance-based relationship strength based on at least one of the cosine similarity and Euclidean distance of each vector in the vector space, and can generate the feature data including numerical information regarding the complex relationships between objects using the weighted value.

[0011] In one embodiment, the at least one processor adds the generated feature data to the company's ERP data to generate final analysis data, applies the final analysis data to a machine learning algorithm to train an analytical model for predicting the company's operating results, and improves the accuracy of prediction and classification tasks by enhancing the performance of a conventional machine learning model based on the added feature data.

[0012] In one embodiment, the at least one processor can analyze time series data of enterprise operation information included in the ERP data and identify at least one of an abnormal pattern and a regular change pattern from the time series data.

[0013] In one embodiment, the at least one processor can analyze correlations between variables contained in the features and patterns extracted from the ERP data and derive causal relationships between the variables based on statistical significance.

[0014] In one embodiment, the at least one processor generates an ontology tree that connects relationships between at least two objects included in the ERP data, and calculates relationship strength based on the hierarchical structure of the at least two objects in the ontology tree.

[0015] In one embodiment, the at least one processor can extract the external variables at the first time point from external events collected in real time based on at least one of the external economic environment, market volatility, or competitor variables, compare the external variables with business rules included in ERP data, and quantify the impact of changes in the external variables on specific business processes or resource allocations of the company.

[0016] In one embodiment, the at least one processor can calculate vector distances between business terms and objects in the vector space to construct a vector space using the Biz2Vec model, and generate vector clusters in which similar business processes or objects are clustered based on the vector distances.

[0017] In one embodiment, the at least one processor evaluates the importance of each of the at least one external event based on the magnitude of the impact of the at least one external event on the company's operations, the frequency of occurrence, and changes in the external events collected in real time, assigns weights of the external variables to the at least one external event based on the importance of the at least one external event, and generates an ontology that numerically reflects the importance among multiple external events using the weights of the external variables.

[0018] Furthermore, a method for generating feature data for predicting business performance based on an ontology according to the present disclosure may include the following operations: generating first data by extracting features and patterns from ERP data based on an analysis; generating second data by extracting features and patterns from the ERP data based on an ontology; generating third data by extracting external variables at a first time point; generating merged data by merging the first data, the second data, and the third data; generating a vocabulary list appropriate for a business domain from the merged data; generating a Biz2Vec model that learns the syntax and context of the business domain using the merged data and the vocabulary list; constructing a vector space appropriate for the business context using the Biz2Vec model; and generating feature data by calculating the similarity between vectors in the vector space and quantifying the relationship between objects using the cosine similarity or Euclidean distance between each vector.

[0019] In addition, a computer program stored on a computer-readable recording medium for embodying the present disclosure may also be provided.

[0020] In addition, a computer-readable recording medium having a computer program for implementing the present disclosure recorded thereon may also be provided. [Effects of the Invention]

[0021] According to the solution to the above-mentioned problem of the present disclosure, the electronic device of the present disclosure can provide a Biz2Vec model that analyzes corporate data and constructs a vector space that conforms to business syntax. Therefore, the electronic device of the present disclosure can generate vectors that are tailored to the business context rather than the everyday context, thereby providing accurate analysis of specific situations and processes that occur in business.

[0022] The electronic device disclosed herein can provide a more accurate predictive model by combining ERP data with an ontology to quantify the impact of external variables that change in real time and converting this into feature data. Furthermore, the electronic device disclosed herein can efficiently reflect important variables in a company's specific situation by defining relationships between objects within the ontology and assigning weights to each external event based on its importance. Therefore, the electronic device disclosed herein can more accurately grasp business risks in response to changes in the external environment.

[0023] The effects of the present disclosure are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description. [Brief explanation of the drawings]

[0024] [Figure 1] FIG. 1 is a diagram showing a block configuration of an electronic device according to the present disclosure. [Figure 2] FIG. 1 is a conceptual diagram illustrating the operation of an electronic device according to the present disclosure. [Figure 3] FIG. 2 is a flowchart illustrating the operation of the electronic device of the present disclosure. [Figure 4] FIG. 10 is an exemplary diagram illustrating how an electronic device of the present disclosure generates feature data based on an external event. [Figure 5] FIG. 10 is an exemplary diagram showing a business ontology based on ERP data and data based on external events. [Figure 6] FIG. 10 is an exemplary diagram illustrating a process in which an electronic device of the present disclosure generates a Biz2Vec model. [Figure 7] FIG. 1 is an exemplary diagram illustrating how the electronic device of the present disclosure improves the performance of a conventional machine learning model based on feature data. DETAILED DESCRIPTION OF THE INVENTION

[0025] The same reference numerals refer to the same elements throughout this disclosure. This disclosure does not describe all elements of the embodiments, and general content in the technical field to which this disclosure belongs or content that is redundant in the embodiments will be omitted. The terms "unit, module, component, block" used in this specification may be embodied in software or hardware, and depending on the embodiment, multiple "units, modules, components, blocks" may be embodied as one component, or one "unit, module, component, block" may include multiple components.

[0026] Throughout this specification, when a part is said to be "connected" to another part, this includes not only direct connection but also indirect connection, including connection via a wireless communication network.

[0027] Furthermore, when a part is described as "comprising" a certain element, this does not mean that it excludes other elements, but that it may further include other elements, unless otherwise specified.

[0028] Throughout this specification, when an element is said to be "on" another element, this includes not only when the element is in contact with the other element, but also when there is another element between the two elements.

[0029] The terms "first," "second," etc. are used to distinguish one component from another, and the components are not limited to the terms described above.

[0030] The singular expression includes the plural expression unless the context clearly indicates otherwise.

[0031] The identification numbers used in each step are for convenience of explanation, and do not dictate the order of the steps; the steps may be performed in a different order than specified unless the context clearly dictates a particular order.

[0032] The working principle and embodiments of the present disclosure will be described below with reference to the accompanying drawings.

[0033] In this specification, the term "device according to the present disclosure" includes all of the various devices that can perform computations and provide results to a user. For example, the device according to the present disclosure may include all of a computer, a server device, and a portable terminal, or may take any one of these forms.

[0034] Here, the computer may include, for example, a notebook computer, a desktop computer, a laptop computer, a tablet PC, a slate PC, etc., equipped with a web browser.

[0035] The server device is a server that communicates with external devices and processes information, and may include an application server, a computing server, a database server, a file server, a game server, a mail server, a proxy server, and a web server.

[0036] The portable terminal may be, for example, a wireless communication device that ensures portability and mobility, and may include any kind of handheld-based wireless communication device such as PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), WiBro (Wireless Broadband Internet) terminal, smartphone, etc., as well as wearable devices such as watches, rings, bracelets, anklets, necklaces, glasses, contact lenses, or head-mounted devices (HMD).

[0037] The AI-related functions of the present disclosure are operated by a processor and a memory. The processor may be composed of one or more processors. Here, the one or more processors may be general-purpose processors such as a CPU, AP, or DSP (Digital Signal Processor), dedicated graphics processors such as a GPU or VPU (Vision Processing Unit), or dedicated AI processors such as an NPU. The one or more processors control the processing of input data according to predefined operating rules or AI models stored in memory. Alternatively, if the one or more processors are dedicated AI processors, the dedicated AI processors may be designed with a hardware structure specialized for processing a specific AI model.

[0038] The predefined behavioral rules or artificial intelligence models are generated through learning. Here, "generated through learning" means that a basic artificial intelligence model is trained using a large amount of training data by a learning algorithm to generate predefined behavioral rules or artificial intelligence models configured to achieve desired characteristics (or objectives). The learning may be performed on the device itself where the artificial intelligence according to the present disclosure is executed, or may be performed through a separate server and / or system. Examples of learning algorithms include supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning, but are not limited to the above examples.

[0039] The artificial intelligence model may be composed of multiple neural network layers. Each of the multiple neural network layers has multiple weight values, and performs neural network operations through operations between the operation results of previous layers and the multiple weight values. The multiple weight values ​​of the multiple neural network layers may be optimized according to the learning results of the artificial intelligence model. For example, the multiple weight values ​​may be updated so that the loss value or cost value acquired by the artificial intelligence model during the learning process is reduced or minimized. The artificial neural network may include a deep neural network (DNN), such as a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), a restricted Boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or a deep Q-network, but is not limited to the above examples.

[0040] According to an exemplary embodiment of the present disclosure, a processor may embody artificial intelligence. Artificial intelligence refers to a machine learning method based on an artificial neural network that allows a machine to learn by imitating human biological neurons. Artificial intelligence methodologies can be classified into supervised learning, in which both input data and output data are provided as training data according to a learning method, thereby determining the answer (output data) to a problem (input data); unsupervised learning, in which only input data is provided without output data, thereby determining the answer (output data) to a problem (input data); and reinforcement learning, in which a reward is provided from an external environment each time an action is taken in a current state, and learning is performed in a direction to maximize the reward. Artificial intelligence methodologies can also be categorized by their architecture, which is the structure of the learning model. The architectures of widely used deep learning technologies can be categorized into convolutional neural networks (CNNs), recurrent neural networks (RNNs), transformers, generative adversarial networks (GANs), etc.

[0041] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

[0042] FIG. 1 is a diagram showing a block configuration of an electronic device 100 of the present disclosure, and FIG. 2 is a conceptual diagram 200 showing the operation of the electronic device 100 of the present disclosure.

[0043] 1, an electronic device 100 according to an embodiment may include a processor 120, a memory 110, and a communication unit (not shown). The configuration shown in FIG. 1 is for illustrating a configuration related to the description of an embodiment according to the present disclosure, and other configurations (e.g., a communication unit) for performing functions of the electronic device may be added in addition to the configuration shown, or the configuration shown may be omitted.

[0044] The memory 110 according to an embodiment is a storage medium used by the electronic device 100 and may store data such as at least one command or setting information corresponding to at least one program, which may include an operating system (OS) program and various application programs.

[0045] In one embodiment, the memory 110 may include at least one type of storage medium selected from the group consisting of a flash memory type, a hard disk type, a multimedia card micro type, a card-type memory (e.g., SD or XD memory), a random access memory (RAM), a static random access memory (SRAM), a read only memory (ROM), an electrically erasable programmable ROM (EEPROM), a programmable ROM (PROM), a magnetic memory, a magnetic disk, and an optical disk.

[0046] In one embodiment, the communication unit may include a wireless communication unit (e.g., a cellular communication module, a short-range wireless communication module, or a global navigation satellite system (GNSS) communication module) or a wired communication unit. The communication unit may communicate with an external electronic device via a network (e.g., a legacy cellular network, a 5G network, a next-generation communication network, the Internet, or a long-range communication network such as a computer network (e.g., a LAN or a WAN)).

[0047] In one embodiment, the communication unit may support 5G networks and next-generation communication technologies, such as new radio access (NR) technology. NR technology may support high-speed transmission of high-capacity data (eMBB (enhanced mobile broadband)), minimizing terminal power and connecting multiple terminals (mMTC (massive machine-type communications)), or high reliability and low latency (URLC (ultra-reliable and low-latency communications)). For example, the communication unit may support high-frequency bands (e.g., mmWave bands) to achieve high data rates. The communication unit may support various technologies for ensuring performance in high-frequency bands, such as beamforming, massive MIMO (multiple-input and multiple-output), full-dimensional MIMO (FD-MIMO), array antennas, analog beamforming, or large-scale antennas.

[0048] In one embodiment, the processor 120 can, for example, execute software (e.g., a program) to control at least one other component (e.g., a hardware or software component) of the electronic device 100 connected to the processor 120, and can perform various data processing or calculations.

[0049] In one embodiment, processor 120 may perform at least some of the data processing or computation by storing instructions or data received from other components (e.g., communication units) in volatile memory, processing the instructions or data stored in volatile memory, and storing the resulting data in non-volatile memory.

[0050] In one embodiment, processor 120 may include a main processor (e.g., a central processing unit or an application processor) or a secondary processor (e.g., a graphics processing unit, a neural processing unit (NPU), or a communication processor) that may operate independently or in conjunction with a main processor. For example, processor 120 may include at least one of a central processing unit (CPU), an application processor (AP), or a microprocessor.

[0051] The processor 120 according to one embodiment can perform the operations of the electronic device 100 described in the following figures. For example, as shown in Figure 2, the processor 120 can provide a more efficient and accurate solution for managing business risks and predicting operational results than conventional methods by combining ontology and ERP data, reflecting external variables in real time, and enhancing the performance of machine learning.

[0052] Specifically, the processor 120 generates first data by extracting features and patterns from ERP data based on analysis, generates second data by extracting features and patterns from the ERP data based on an ontology, generates third data by extracting external variables at a first time point, generates merged data by merging the first data, the second data, and the third data, generates a vocabulary suitable for the business domain from the merged data, generates a Biz2Vec model that learns the syntax and context of the business domain using the merged data and the vocabulary, constructs a vector space suitable for the business context using the Biz2Vec model, calculates the similarity between vectors in the vector space, and generates feature data by quantifying the relationship between objects using the cosine similarity or Euclidean distance between each vector. ERP refers to, but is not limited to, a plan for the appropriate allocation and effective utilization of resource elements such as people, materials, money, and information that are fundamental to corporate management. ERP can also be described as a system for integrating, but is not limited to, core business operations such as accounting, human resources, production, logistics, and sales, in order to improve efficiency and centralize information. ERP data is data related to ERP.

[0053] In this way, the electronic device 100 of the present disclosure can provide a more accurate predictive model by combining ERP data and ontology to quantify the impact of external variables that change in real time and converting this into feature data. Furthermore, the electronic device 100 of the present disclosure can efficiently reflect important variables in a company's specific situation by defining relationships between objects within the ontology and assigning weights to each external event based on the importance of the event. Therefore, the electronic device 100 of the present disclosure can more accurately grasp business risks in response to changes in the external environment.

[0054] FIG. 3 is a flow chart illustrating the operation of the electronic device 100 of the present disclosure.

[0055] 3, the electronic device 100 of the present disclosure generates first data by extracting features and patterns from ERP data based on analysis (OPERATION 310), generates second data by extracting features and patterns from the ERP data based on an ontology (OPERATION 320), generates third data by extracting external variables at a first time point (OPERATION 330), generates merged data by merging the first data, the second data, and the third data (OPERATION 340), generates a vocabulary appropriate for the business domain from the merged data (OPERATION 350), generates a Biz2Vec model that learns the syntax and context of the business domain using the merged data and the vocabulary (OPERATION 360), constructs a vector space appropriate for the business context using the Biz2Vec model (OPERATION 370), calculates the similarity between vectors in the vector space, and quantifies the relationship between objects using the cosine similarity or Euclidean distance between each vector, thereby generating feature data (OPERATION 380).

[0056] According to one example, in operation 310, the electronic device 100 can generate first data by analytically extracting features and patterns from the ERP data.

[0057] The electronic device 100 can collect enterprise operation data from an ERP system, analyze features and patterns based on the operation data, and generate first data. The electronic device 100 can first convert various business process, resource, human resource, inventory, financial information, etc. contained in the ERP data into time-series data and analyze changes over time.

[0058] The electronic device 100 can identify normal operating patterns and outliers of a company by analyzing time series data. For example, it can extract regular patterns based on patterns of fluctuations within a certain range in production volume, inventory, sales, etc. during a specific period, and detect abnormal patterns when unexpected fluctuations occur.

[0059] The electronic device 100 can analyze correlations between variables identified in the ERP data. In this process, the electronic device 100 can evaluate the relationships between variables to analyze the impact of a specific variable (e.g., exchange rate fluctuations, sales, raw material prices, personnel allocation, etc.) on other variables. This allows for the derivation of causal relationships between variables in the ERP data.

[0060] The electronic device 100 can generate first data reflecting the operating characteristics of a company based on correlation and causality analysis. The generated first data can include characteristic data summarizing ERP data and highlighting variables important to business decision-making. For example, the correlation between increases and decreases in sales over a specific period and personnel allocation or raw material prices can be analyzed and included as the first data.

[0061] According to one example, in operation 320, the electronic device 100 can generate second data by extracting features and patterns from the ERP data based on an ontology.

[0062] The electronic device 100 may generate second data by extracting features and patterns based on an ontology to analyze relationships between objects included in the ERP data. The electronic device 100 may first analyze relationships between various business objects (e.g., products, customers, materials, suppliers, and orders) defined in the ERP data using an ontology model. Each object has unique attributes, and the attributes may be defined in a linked state through the ontology.

[0063] The electronic device 100 can apply relationship rules defined within the ontology to understand interactions between objects. For example, the relationship between a product and a supplier may be an important element in a supply chain, and the relationship between a customer and an order may have direct relevance to sales. The electronic device 100 can analyze the relationships between the objects to understand the dependencies between the objects.

[0064] The electronic device 100 can also evaluate the strength of interactions between objects through a hierarchical structure. For example, if a supplier of raw materials for a particular product is changed, the impact of this change on the production cost of the product can be evaluated. The electronic device 100 can quantify the relationship strength to determine which objects play an important role in business decision-making.

[0065] Based on the ontology, the electronic device 100 can analyze the roles and characteristics of each object and define the relationships between the objects into a concrete pattern. For example, if the order pattern associated with a particular customer changes at a particular time, the change may reflect fluctuations in market demand. The electronic device 100 can visualize the relationships in an ontology tree or graph structure to clearly understand the interactions between the objects.

[0066] The electronic device 100 can generate second data based on the analysis, and the second data includes relationships and interaction patterns between objects derived based on the ontology. Thus, the electronic device 100 can obtain data reflecting relational meaning in addition to the simple information represented in the ERP data, and by utilizing this data, enable more comprehensive and insightful analysis of business operations.

[0067] According to one example, in operation 330, the electronic device 100 can generate the third data by extracting external variables at the first time point.

[0068] The electronic device 100 may analyze the external variables collected in real time to extract external variables at a first time point and generate third data based on the external variables at the first time point. The external variables may be various factors that may affect the operation of a company, such as economic indicators (e.g., GDP, changes in interest rates), fluctuations in exchange rates, stock prices, raw material prices, market trends, strategic changes by competitors, etc.

[0069] The electronic device 100 can first collect external variables through real-time data streams and analyze the changes and patterns of the external variables at a specific time point. For example, if the exchange rate changes suddenly at a specific time point, the impact of the change on the import and export activities, raw material prices, or financial status of a company can be evaluated. Here, the term "real-time data stream" refers to, but is not limited to, data that changes in real time over time. The real-time data stream is, for example, data collected from various data sources and processed in real time. The real-time data stream includes the external variables that change in real time.

[0070] The electronic device 100 can compare and analyze the ERP data with external variables to evaluate how changes in the external variables affect the business process of the enterprise. For example, a decrease in market demand as an external variable may cause changes in the ERP data, such as a decrease in order volume or an adjustment to the production plan, and the impact on the business process of the enterprise can be evaluated based on the changes.

[0071] In addition, the electronic device 100 can quantify external variables and numerically evaluate the impact of changes in each variable on a company's specific business processes, resource allocation, cost structure, sales, etc. For example, it can analyze how much an increase in raw material prices will affect a company's production costs, or how a competitor's price cuts will affect a company's market share.

[0072] The electronic device 100 can assign a weight to each variable depending on the importance of the external variable. For example, if fluctuations in exchange rates have a significant impact on a company's export ratio, the variable can be weighted more highly. If an economic downturn has a significant impact on a consumer goods company, the variable can also be weighted more highly.

[0073] Based on the analysis, the electronic device 100 may generate third data. The third data may include data reflecting the impact of real-time changes in external variables on the operation of the enterprise. The third data may help present countermeasures to the external variables during the enterprise's decision-making process.

[0074] According to one example, in operation 340, electronic device 100 may generate merged data by merging the first data, the second data, and the third data.

[0075] According to one example, in operation 340, the electronic device 100 may merge the first data, the second data, and the third data to generate merged data, which is a process of integrating data derived from different data sources to generate a comprehensive data set for analysis.

[0076] The electronic device 100 can first analyze the relationship between the features and patterns of ERP data extracted from the first data and the ontology-based objects derived from the second data, and identify the relevance and redundancy between them. For example, the electronic device 100 can integrate the relationship between the usage pattern of a specific resource identified in the ERP data and the resource defined in the ontology and a different object (e.g., production process or personnel allocation) to derive new semantic connectivity.

[0077] In addition, the electronic device 100 can analyze how real-time external variables included in the third data affect the relationship between the ERP data and the ontology-based object. For example, the electronic device 100 can integrate the first data and the second data to understand the effect of fluctuations in exchange rates collected from the third data on the cost of a specific resource and production plans.

[0078] During the merging process, the electronic device 100 can remove overlapping elements between the data and assign weights according to their importance. For example, if an object relationship defined in the ontology matches an operational pattern in the ERP data, the relationship can be assigned a higher weight so that it plays a more important role in the analysis.

[0079] The electronic device 100 can be designed to maintain the semantic consistency of data when combining different data types during the merging process, thereby converting the quantitative characteristics of ERP data and the qualitative relationship information of the ontology into a single integrated data set. During this process, the electronic device 100 can generate merged data that highlights relationships and patterns that are important for business decision-making.

[0080] The electronic device 100 can provide an integrated analytical data set that takes into account all of the company's internal data and external variables by merging the first, second, and third data. For example, the merged data can be used to build better predictive models and as a basis for analyzing business operation performance. The electronic device 100 can semantically associate the first data with the second data based on terms, features, patterns, etc. of the ERP data related to the first data, and terms, features, patterns, and relationships between ontology-based objects related to the second data. The electronic device 100 can also semantically associate external variables related to the third data with the first data and the second data based on, for example, the terms used in the external variables. As a result, the electronic device 100 can generate an integrated analysis dataset by semantically associating and merging the first, second, and third data.

[0081] According to one example, in operation 350, the electronic device 100 can generate a vocabulary appropriate for a business domain from the merged data, the vocabulary being a vocabulary that can semantically associate the first to third data.

[0082] The electronic device 100 can generate a corpus appropriate for a business domain from the merged data.

[0083] The electronic device 100 can analyze the ERP data, ontology data, and external variable data included in the merged data to extract business terms.

[0084] The electronic device 100 can identify core terms that are frequently used in a business domain and create a vocabulary taking into consideration the frequency of appearance of the terms, the context of use, related objects, and the like.

[0085] The electronic device 100 can evaluate the importance of the extracted terms according to the business situation and assign higher priority to terms that are important for business decision making.

[0086] The electronic device 100 can also analyze the association between words to group related terms in a business domain and define relationships between the terms.

[0087] For example, the electronic device 100 can generate a vocabulary appropriate for business processes and operations, and use the vocabulary for subsequent analysis and machine learning.

[0088] According to one example, in operation 360, the electronic device 100 can generate a Biz2Vec model that uses the merged data and the lexicon to learn the syntax and context of the business domain.

[0089] The electronic device 100 can generate a Biz2Vec model based on the merged data and vocabulary. This process can be a procedure for learning syntax and context that fit a business domain and building a model that provides vector representations appropriate for the business domain.

[0090] The electronic device 100 first analyzes the ERP data, ontology-based object relationship data, and external variable data extracted from the merged data to learn the interactions within each data, where the ERP data indicates the overall operational characteristics of the company, the ontology data indicates the relationships between objects, and the external variable data indicates the impact of real-time environmental changes on the company's operations.

[0091] The electronic device 100 can understand the interrelationships between syntaxes and terms commonly used in a business environment based on the business terms and syntax extracted from the corpus. For example, a syntax such as "increased sales" for a particular product can be related to "increased consumer demand" as an external variable, or can be linked to changes in production plans at a particular time in ERP data.

[0092] The electronic device 100 can vectorize the associations between the terms and syntax and train a Biz2Vec model. The Biz2Vec model, like Word2Vec, expresses word meanings in a vector space, but unlike general language models, it trains by reflecting business-specific syntax and context. That is, it can generate vectors that reflect meanings appropriate for the business context, rather than general word meanings.

[0093] For example, a Biz2Vec model can learn the relationship between words such as "inventory" and "cost" by reflecting their meaning in a business context, rather than in a general context. In the above case, "inventory" and "cost" are strongly connected to each other through the relationship between the ERP data and the ontology, and can be placed close to each other in the vector space.

[0094] In the process of training the Biz2Vec model, the electronic device 100 may consider the semantic similarity between words to accurately reflect business syntax and context. Here, if a particular syntax appears frequently or is highly relevant in a particular situation, the similarity between the vectors may be set high.

[0095] For example, the electronic device 100 can construct a vector space suited to a business context through the learned Biz2Vec model, and enable decision-making and predictive analysis suited to various situations occurring in the business domain based on the vector space suited to the business context.

[0096] According to one example, in operation 370, the electronic device 100 can use the Biz2Vec model to construct a vector space that fits a business context.

[0097] The electronic device 100 can use the Biz2Vec model to construct a vector space that fits a business context. This process converts terms, objects, syntax, etc. in a business domain into vectors to form a vector space based on semantic similarity.

[0098] The electronic device 100 can convert each data element into a vector based on the ERP data, ontology data, and external variable data in the merged data trained through the Biz2Vec model. For example, business objects such as products, customers, orders, and costs can be arranged as coordinates in the vector space.

[0099] The electronic device 100 can place objects at similar vector coordinates if the objects have similar meanings or relationships in a business context. For example, the terms "sales" and "order volume increase" have related meanings in business operations and therefore may be at similar positions in the vector space. External variables and business factors, such as exchange rate fluctuations and raw material costs, can also be placed at similar positions on the vector coordinates based on their relationship to each other.

[0100] The electronic device 100 can also set the distance between vector coordinates to reflect the strength of the relationship between objects. Using mathematical methods such as cosine similarity or Euclidean distance, the similarity between each vector can be calculated, and the vector space can be configured so that objects that are closely related maintain a shorter distance. For example, if the relationship between inventory and cost is strong, the distance between the vectors will be relatively short.

[0101] Furthermore, the electronic device 100 can adjust the weight of the vector if a particular term or object is used as important in the business domain. For example, if an increase in customer demand is determined to be an important factor at a particular time, it can be emphasized more in the vector space to further strengthen its relationship with other related terms.

[0102] For example, the electronic device 100 can form a vector group suited to a business context and construct a vector cluster between similar objects that is useful for business operations and decision-making. The vector space can be used in an analytical model or a forecasting system to build a decision support system tailored to the business situation, and can be useful for real-time risk management, outcome prediction, etc.

[0103] According to one example, in operation 380, the electronic device 100 can generate feature data by calculating the similarity between vectors in the vector space and quantifying the relationship between objects using cosine similarity or Euclidean distance between the vectors. Operation 380 of the electronic device 100 will be described in detail below with reference to FIGS. 4 to 7.

[0104] FIG. 4 is an exemplary diagram showing a method 400 in which the electronic device 100 of the present disclosure generates feature data based on external events, and FIG. 5 is an exemplary diagram 500 showing a business ontology based on ERP data and data based on external events.

[0105] 4 and 5, the electronic device 100 can generate feature data by calculating the similarity between vectors in a vector space. The electronic device 100 first compares the vectors of each object in the vector space generated through the Biz2Vec model, and can quantify the similarity between the objects using cosine similarity and Euclidean distance based on the comparison.

[0106] The electronic device 100 can measure the directional similarity between two vectors using cosine similarity. For example, if an increase in sales and an increase in customer demand have similar directions, the cosine similarity value of the two vectors may be close to 1. On the other hand, between objects that are weakly related or contradictory, the cosine similarity value may be close to 0. This allows the electronic device 100 to identify objects that are similar to each other or highly correlated.

[0107] The electronic device 100 can also calculate the absolute distance between objects using Euclidean distance. The electronic device 100 can analyze that the shorter the distance between two objects, the stronger the relationship between the objects. For example, if the distance between two vectors in the relationship between raw material prices and production costs is short, the two objects can be evaluated as having a high correlation with each other.

[0108] The electronic device 100 can quantify the strength of the relationship between objects by combining the calculated cosine similarity and Euclidean distance. For example, if a decrease in inventory and an increase in order volume have a high similarity, the relationship between them can be reflected in feature data and analyzed as a meaningful business pattern. Through this process, the electronic device 100 can define relationships important to business decision-making as feature data and further improve the performance of data analysis models based on this.

[0109] For example, the electronic device 100 can calculate the similarity and relationship strength between each object and cluster objects with high relationship strength to derive meaningful business patterns. The feature data can be applied to subsequent predictive models and machine learning algorithms to improve the accuracy of analysis and prediction.

[0110] FIG. 6 is an exemplary diagram showing a process 600 in which the electronic device 100 of the present disclosure generates a Biz2Vec model, and FIG. 7 is an exemplary diagram 700 in which the electronic device 100 of the present disclosure improves the performance of a conventional machine learning model based on feature data.

[0111] 6 and 7, in operation 380, the electronic device 100 can apply the generated feature data to a conventional machine learning model to improve the model's performance. The electronic device 100 can combine the previously generated feature data with conventional ERP data to create a final analysis dataset. The final analysis dataset includes more feature information than an analysis using only conventional ERP data, and is composed of data that reflects external variables and ontology relationships, enabling more precise analysis.

[0112] The electronic device 100 can apply the final dataset with the added feature data to a machine learning algorithm to improve the learning model. For example, the electronic device 100 can input the feature data into a regression analysis model, a classification model, or a deep learning-based prediction model to obtain better prediction performance. In particular, the feature data contains relationship and similarity information not found in conventional data, and therefore can more accurately reflect interactions and complex relationships than conventional models.

[0113] In addition, the electronic device 100 can improve the prediction accuracy and performance indicators of conventional models using the generated feature data. For example, in business problems such as sales forecasting and customer demand analysis, the electronic device 100 can calculate more accurate predictions by using models containing feature data. The feature data can more clearly explain the relationships between business objects and how external variables affect business outcomes, thereby significantly improving the reliability of predictions.

[0114] The electronic device 100 can retrain a conventional machine learning model based on the feature data, thereby improving the model's predictive performance. For example, in an inventory management system, the feature data can more accurately understand the relationship between order patterns and inventory fluctuations, which can be more useful for real-time decision-making such as inventory optimization.

[0115] As a result, the electronic device 100 can maximize the performance of machine learning algorithms by adding the generated feature data to conventional models, providing a more sophisticated analytical environment that supports accurate predictions and decision-making.

[0116] In this way, the electronic device 100 of the present disclosure can provide a more accurate predictive model by combining ERP data and ontology to quantify the impact of external variables that change in real time and converting this into feature data. Furthermore, the electronic device 100 of the present disclosure can efficiently reflect important variables in a company's specific situation by defining relationships between objects within the ontology and assigning weights to each external event based on the importance of the event. Therefore, the electronic device 100 of the present disclosure can more accurately grasp business risks in response to changes in the external environment.

[0117] However, since this has been described above, a duplicate explanation will be omitted.

[0118] The devices described above may be implemented using hardware components, software components, and / or a combination of hardware and software components. For example, the devices and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as a processor, controller, arithmetic logic unit (ALU), digital signal processor, microcomputer, field programmable array (FPA), programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, the description may refer to a single processing device. However, those skilled in the art will recognize that a processing device may include multiple processing elements and / or multiple types of processing elements. For example, a processing device may include multiple processors or one processor and one controller. Other processing configurations, such as parallel processors, are also possible.

[0119] Methods according to the embodiments may be embodied in the form of program instructions that can be executed by various computer means and recorded on a computer-readable medium. The computer-readable medium may include, alone or in combination, program instructions, data files, data structures, and the like. The program instructions recorded on the medium may be specially designed and constructed for the embodiments, or may be known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape; optical media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include not only machine language code, such as generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

[0120] Although the above embodiments have been described with limited illustrations, those skilled in the art will appreciate that various modifications and variations may be made from the above description. For example, the described techniques may be performed in a different order than described, and / or the described system, structure, device, circuit, or other components may be combined or combined in a different manner than described, or may be substituted or replaced by other components or equivalents, and still achieve suitable results. Accordingly, other implementations, other embodiments, and equivalents of the claims are also intended to be within the scope of the following claims. [Explanation of symbols]

[0121] 100 Electronic equipment 110 memory 120 processors

Claims

1. An electronic device that generates feature data for predicting business outcomes based on an ontology, comprising: a memory for storing at least one instruction word; at least one processor that executes the at least one instruction; Including, The at least one processor generating first data by analytically extracting features and patterns from the ERP data; generating second data by extracting features and patterns from the ERP data based on an ontology; generating third data by extracting external variables at the first time point; generating merged data by merging the first data, the second data, and the third data; Creating a vocabulary list appropriate for the business domain from the merged data; generating a Biz2Vec model that learns the syntax and context of the business domain using the merged data and the vocabulary; Using the Biz2Vec model, construct a vector space that fits the business context; An electronic device that generates feature data by calculating the similarity between vectors in the vector space and quantifying the relationship between objects using the cosine similarity or Euclidean distance between each vector.

2. The at least one processor calculating a weighted value that combines the relative similarity and distance-based relationship strength of each vector based on at least one of the cosine similarity and Euclidean distance of each vector in the vector space; The electronic device of claim 1 , wherein the weighted values ​​are used to generate the feature data including numerical information regarding complex relationships between objects.

3. The at least one processor adding the generated feature data to the enterprise resource planning (ERP) data of the enterprise to generate data for final analysis; Applying the final analysis data to a machine learning algorithm to train an analytical model for predicting the business performance of the company; The electronic device of claim 2 , further comprising: a processor for processing the generated data; a processor for processing the generated data; a processor for processing the generated data;

4. The at least one processor Analyzing time series data of enterprise operation information included in the ERP data; The electronic device according to claim 1 , wherein at least one of an abnormal pattern and a regular change pattern is identified from the time series data.

5. The at least one processor Analyzing correlations between variables included in the features and patterns extracted from the ERP data; The electronic device according to claim 1 , wherein the causal relationships between the variables are derived based on statistical significance.

6. The at least one processor generating an ontology tree that connects relationships between at least two objects included in the ERP data; The electronic device of claim 1 , wherein the relationship strength is calculated based on a hierarchical structure of the at least two objects in the ontology tree.

7. The at least one processor extracting the external variables at the first time point from external events collected in real time based on at least one of an external economic environment, market volatility, or a competitor's variable; comparing the external variables with business rules contained in the ERP data; The electronic device according to claim 1 , wherein the electronic device quantifies the influence of a change in the external variable on a specific business process or resource allocation of a company.

8. The at least one processor 2. The electronic device according to claim 1, wherein, in order to construct a vector space using the Biz2Vec model, a vector distance between a business term and an object in the vector space is calculated, and vector clusters in which similar business processes or objects are clustered together are generated based on the vector distance.

9. The at least one processor Evaluating the importance of at least one external event based on the magnitude of the impact of the at least one external event on the business operations of the company, the frequency of occurrence, and changes in the external events collected in real time; assigning a weight value of the external variable to the at least one external event based on the importance of each of the at least one external event; The electronic device according to claim 1 , wherein an ontology is generated that numerically reflects the importance of a plurality of external events using the weighted values ​​of the external variables.

10. 1. A method for generating feature data for predicting business outcomes based on an ontology executed by a processor of an electronic device, comprising: generating first data by analytically extracting features and patterns from the ERP data; generating second data by extracting features and patterns from the ERP data based on an ontology; generating third data by extracting external variables at a first time point; generating merged data by merging the first data, the second data, and the third data; generating a vocabulary list appropriate for a business domain from the merged data; generating a Biz2Vec model that learns syntax and context of the business domain using the merged data and the vocabulary; constructing a vector space that fits a business context using the Biz2Vec model; An operation of calculating similarities between vectors in the vector space and quantifying relationships between objects using cosine similarity or Euclidean distance between each vector to generate feature data; A method comprising:

Images