Large language model intelligent analysis and decision-making method for data collection of low-code platform

Abstract

The invention discloses a large language model intelligent analysis and decision-making method for data collection of a low-code platform. Comprising the following steps: S1, data collection and storage; s2, data accumulation; s3, data preprocessing; s4, performing deep analysis on the large language model; s5, automatically generating a statistical report; s6, providing decision suggestions; according to the method, data collection, form creation and form preprocessing are carried out through a plurality of low-code platforms, the forms are analyzed and processed through a large language model, customer purchase evaluation is subjected to sentiment analysis by extracting key information, the satisfaction degree of customers to commodities is judged, potential association and trends in data are further identified, and the customer purchase evaluation efficiency is improved. According to the method, sales conditions of commodities and customer behavior characteristics under different dimensions are analyzed, a comprehensive and multi-dimensional statistical report is automatically generated, decision suggestions with insight are automatically generated by a large language model, powerful support is provided for business development, and implementation and adjustment are carried out according to actual conditions.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a method for intelligent analysis and decision-making of a large language model for data collected by a low-code platform. Background Art

[0002] With the development of information technology and artificial intelligence, more and more companies and organizations are looking for efficient and flexible data processing and decision support solutions. Especially with the increasing amount of data, traditional manual data processing methods and decision-making processes are no longer able to cope with the complex and changing needs of modern society. Therefore, low-code platforms and large language models have become important tools for digital transformation and intelligent decision-making.

[0003] A low-code platform is a platform that allows users to develop applications with minimal or even no coding, aiming to simplify the development process and improve development efficiency. Through a visual drag-and-drop interface and preset templates, a low-code platform enables even users without a programming background to build, deploy, and manage application systems.

[0004] Data collection and integration: Low-code platforms allow users to quickly create data input interfaces, data collection processes, and database integration. Through visual tools, users can easily connect to external data sources such as databases, APIs, IoT devices, etc., and integrate this data into the platform.

[0005] Automated workflow: Low-code platforms usually provide automation functions, such as automated form filling, data validation, data cleaning, etc. This is very important for the collection, organization, and pre-processing of data.

[0006] Business logic modeling: Low-code platforms enable users to define and implement business logic without relying on complex coding. This allows data collection to be not limited to forms and basic queries, but also to apply more complex business rules.

[0007] The big language model is based on a deep learning algorithm. Through pre-training on massive text data, it can perform well in natural language processing tasks such as text generation, text classification, sentiment analysis, question-answering systems, etc.

[0008] Intelligent data analysis: The big language model can automatically extract valuable information from data and perform in-depth analysis. For example, it can identify trends, predict results, or provide insights from structured data, unstructured text, or log data.

[0009] Natural language understanding and generation: The large language model can understand natural language input and generate corresponding decisions or recommendations. For example, users can ask the model in natural language, "Based on the data of the past three months, which product has the fastest sales growth?" The model can make automated analysis based on the collected data.

[0010] Decision support and optimization: The large language model can also provide decision suggestions based on data, simulate the consequences of different decision paths, and even help users make decisions. By deducing multiple possible situations, the model can provide decision makers with risk assessment and benefit analysis.

[0011] However, there are some problems with the existing technology: the current low-code mainly focuses on basic functions such as form design and process building. For the collected data, it usually only implements simple storage operations and carries out basic statistical work. Users often need to rely on their own experience and subjective judgment when making decisions. Therefore, we propose a method of intelligent analysis and decision-making of large language models for data collected by low-code platforms. Summary of the invention

[0012] In response to the problems existing in the prior art, the purpose of the present invention is to provide a method for intelligent analysis and decision-making of a large language model for data collected by a low-code platform. Data is efficiently collected and stored through a low-code platform. When the amount of data accumulates to a certain scale, the integrated large language model immediately activates its powerful functions, conducts in-depth mining and analysis of these data, and autonomously generates comprehensive and multi-dimensional statistical reports.

[0013] The present invention is implemented in this way: a method for intelligent analysis and decision-making of a large language model for collecting data on a low-code platform comprises the following steps:

[0014] S1. Data collection and storage: Data is collected through multiple low-code platforms, and forms are quickly created through visual interfaces and components, and then stored in the database;

[0015] S2. Data accumulation: Monitor the number of forms in the database to determine whether the forms in the database meet the minimum number requirement for subsequent analysis;

[0016] S3. Data preprocessing: The low-code platform cleans the forms in the database and processes outliers;

[0017] S4. Deep analysis of large language model: The large language model performs deep analysis on the preprocessed data to extract key information. The large language model further identifies potential correlations and trends in the data.

[0018] S5. Automatically generate statistical reports: The large language model automatically generates comprehensive and multi-dimensional statistical reports;

[0019] S6. Provide decision recommendations: The large language model automatically generates insightful decision recommendations and reviews the generated decision recommendations.

[0020] Specifically, the steps for multiple low-code platforms in S1 to collect data are as follows:

[0021] Use the visual interface and components of the low-code platform to quickly create forms for collected customer purchase history, browsing behavior records, and evaluation records;

[0022] Use low-code to create an automated data collection process, and collect customer purchase history, browsing behavior records, and evaluation records through RPA nodes, data update nodes, and large model nodes;

[0023] Generate a form with the customer's purchase history, browsing behavior record and evaluation record, and store the form in the database.

[0024] Specifically, the multiple low-code platforms collect data through crawler algorithms, and the collection steps are as follows:

[0025] Send request: The crawler program sends a request to the server of the target website through HTTP request to obtain the HTML page of the web page;

[0026] Parsing web page content: After receiving the HTML content of a web page, the crawler needs to parse it and extract useful information. The parsing targets include text and table data in the web page.

[0027] Extract data: Use regular expressions, CSS selectors, and XPath methods to extract data from HTML pages, including customer purchase history, browsing behavior records, and evaluation records;

[0028] Storing data: After parsing the extracted data, quickly create a form through the visual interface and components, and then store the form in the database.

[0029] Specifically, the steps of data preprocessing in S3 are as follows:

[0030] When the database detects that there are enough forms;

[0031] Low-code platforms begin data preprocessing;

[0032] First, process the form for outliers, set a threshold, and treat values ​​exceeding the threshold as outliers and remove them;

[0033] Then clean the form data, remove duplicate data, and fill in the gaps with the mean after eliminating missing values ​​and outliers.

[0034] Specifically, the outlier processing performs outlier detection on the data information in the form, and the outlier is calculated as follows:

[0035]

[0036] Among them, F is the outlier factor, D is the collected data information, and N j is the number of samples in the jth cluster group, d(x i ,c j ) is the data information x i The distance to the jth cluster center, and c j is the jth cluster center;

[0037] For the distance away from the cluster center, it is set to average the distance of the data information from the center position, and the data results processed by the cluster analysis algorithm are calculated:

[0038] That is Among them, c j is the jth cluster center,

[0039] If the distance of the data information is greater than 3d, it is judged as abnormal data information.

[0040] Specifically, the missing values ​​and outliers are removed and the vacancies are filled with the mean, and the mean is calculated as follows:

[0041] Among them, x i is the i-th value in the data information; n is the total number of data information; Represents the sum of all data information.

[0042] Specifically, the steps of deep analysis of the large language model in S4 are as follows:

[0043] The large language model conducts in-depth analysis on the pre-processed forms, extracts key information, and performs sentiment analysis on customer purchase reviews to determine customer satisfaction with the product.

[0044] The large language model further identifies potential connections and trends in data information, divides the data according to multiple dimensions such as region, age and consumption capacity, and analyzes product sales and customer behavior characteristics in different dimensions.

[0045] Specifically, the steps of automatically generating a statistical report in S5 are as follows:

[0046] The big language model automatically generates comprehensive and multi-dimensional statistical reports based on the potential correlations and trends in the data information;

[0047] Comprehensive and multi-dimensional statistical reports count the sales quantity and sales amount of various commodities by quarter, province, and age.

[0048] Specifically, the step of providing decision suggestions in S6 is as follows:

[0049] Based on comprehensive and multi-dimensional statistical reports, the large language model automatically generates insightful decision-making recommendations;

[0050] The large language model reviews the generated decision recommendations and assesses risks and uncertainties. Risk assessment methods include: risk matrix, probability analysis and scenario analysis.

[0051] Specifically, the risk matrix evaluates the risk levels of different decision-making options by drawing a risk matrix;

[0052] The probability analysis calculates the probability of different events occurring in order to estimate the possible consequences of different decisions;

[0053] The scenario analysis analyzes the results of decision-making under different assumptions and scenarios;

[0054] The calculation of the probability analysis is as follows:

[0055]

[0056] Among them, P(A|B) is the probability of event A occurring when it is known that event B has occurred, P(B|A) is the probability of event B occurring when event A has occurred, P(A) is the prior probability of event A, and P(B) is the marginal probability of event B.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] The present invention collects data through multiple low-code platforms, creates forms through a visual interface and components, and enables a large language model to perform centralized analysis and processing based on the number of forms collected by detection. Before the large language model performs centralized analysis and processing, the low-code platform preprocesses the form to ensure the accuracy of subsequent analysis, and the large language model analyzes and processes the form, extracts key information to perform sentiment analysis on customer purchase evaluations, judges customer satisfaction with the product, further identifies potential associations and trends in the data, divides data according to dimensions such as region, age, and consumption capacity, analyzes the sales of products and customer behavior characteristics under different dimensions, and automatically generates comprehensive and multi-dimensional statistical reports, which count the sales quantity and sales volume of various products by quarter, province, age, and other dimensions. The large language model automatically generates insightful decision-making suggestions, provides strong support for business development, and implements and adjusts according to actual conditions to achieve business goals, significantly improving the efficiency of data collection and analysis, and reducing dependence on professional programmers. At the same time, through the support of large language models and intelligent decision-making, it provides enterprises with more accurate and rapid data insights and decision-making basis, and promotes business innovation and optimization.

[0059] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of the step flow provided by the present invention. DETAILED DESCRIPTION

[0061] In order to further understand the content, features and effects of the present invention, the following embodiments are given as examples and described in detail with reference to the accompanying drawings.

[0062] like Figure 1 As shown, the method for intelligent analysis and decision-making of a large language model for collecting data on a low-code platform provided in an embodiment of the present invention includes the following steps:

[0063] S1. Data collection and storage: Data is collected through multiple low-code platforms, and forms are quickly created through visual interfaces and components, and then stored in the database;

[0064] S2. Data accumulation: Monitor the number of forms in the database to determine whether the forms in the database meet the minimum number requirement for subsequent analysis;

[0065] S3. Data preprocessing: The low-code platform cleans the forms in the database and processes outliers;

[0066] S4. Deep analysis of large language model: The large language model performs deep analysis on the preprocessed data to extract key information. The large language model further identifies potential correlations and trends in the data.

[0067] S5. Automatically generate statistical reports: The large language model automatically generates comprehensive and multi-dimensional statistical reports;

[0068] S6. Provide decision recommendations: The large language model automatically generates insightful decision recommendations and reviews the generated decision recommendations.

[0069] In this embodiment, preferably, the steps of data collection by the multiple low-code platforms in S1 are as follows:

[0070] Use the visual interface and components of the low-code platform to quickly create forms for collected customer purchase history, browsing behavior records, and evaluation records;

[0071] Use low-code to create an automated data collection process, and collect customer purchase history, browsing behavior records, and evaluation records through RPA nodes, data update nodes, and large model nodes;

[0072] Generate a form based on the customer's purchase history, browsing behavior record and evaluation record, and store the form in the database;

[0073] It should be noted that data can be quickly collected through multiple low-code platforms and in a variety of ways, which makes it easier to increase the diversity of data collection sources and generate data into forms to facilitate subsequent observation and storage.

[0074] In this embodiment, preferably, the multiple low-code platforms collect data through crawler algorithms, and the collection steps are as follows:

[0075] Send request: The crawler program sends a request to the server of the target website through HTTP request to obtain the HTML page of the web page;

[0076] Parsing web page content: After receiving the HTML content of a web page, the crawler needs to parse it and extract useful information. The parsing targets include text and table data in the web page.

[0077] Extract data: Use regular expressions, CSS selectors, and XPath methods to extract data from HTML pages, including customer purchase history, browsing behavior records, and evaluation records;

[0078] Storing data: After parsing the extracted data, quickly create a form through the visual interface and components, and then store the form in the database;

[0079] It should be noted that crawler algorithms are used to search and query data information on web pages or programs, and extract data information, which facilitates the creation of forms and storage of data information.

[0080] In this embodiment, preferably, the steps of data preprocessing in S3 are as follows:

[0081] When the database detects that there are enough forms;

[0082] Low-code platforms begin data preprocessing;

[0083] First, process the form for outliers, set a threshold, and treat values ​​exceeding the threshold as outliers and remove them;

[0084] Then clean the data in the form, remove duplicate data, and fill in the gaps with the mean after eliminating missing values ​​and outliers;

[0085] It should be noted that the low-code platform cleans the data, such as removing duplicate data and processing missing values; and processes outliers, such as setting thresholds, treating values ​​exceeding the thresholds as outliers and performing corresponding processing, to ensure the accuracy of subsequent analysis.

[0086] In this embodiment, preferably, the outlier processing performs outlier detection on the data information in the form, and the outlier is calculated as follows:

[0087]

[0088] Among them, F is the outlier factor, D is the collected data information, and N j is the number of samples in the jth cluster group, d(x i ,c j ) is the data information x i The distance to the jth cluster center, and c j is the jth cluster center;

[0089] For the distance away from the cluster center, it is set to average the distance of the data information from the center position, and the data results processed by the cluster analysis algorithm are calculated:

[0090] That is Among them, c j is the jth cluster center,

[0091] If the distance of the data information is greater than 3d, it is judged as abnormal data information;

[0092] It should be noted that by calculating the distance between the data information and the cluster center, the anomaly of the data information can be determined, and the abnormal information can be eliminated to improve the accuracy of the data information.

[0093] In this embodiment, preferably, the missing values ​​and outliers are removed and the vacancies are filled with the mean, and the mean is calculated as follows:

[0094] Among them, x i is the i-th value in the data information; n is the total number of data information; Represents the sum of all data information;

[0095] It should be noted that by filling the missing values ​​and vacancies in the data information with the mean, the integrity of the data information can be improved, and it is convenient to improve the accuracy of the data information.

[0096] In this embodiment, preferably, the steps of deep analysis of the large language model in S4 are as follows:

[0097] The large language model conducts in-depth analysis on the pre-processed forms, extracts key information, and performs sentiment analysis on customer purchase reviews to determine customer satisfaction with the product.

[0098] The big language model further identifies potential connections and trends in data information, divides data by region, age, and spending power, and analyzes product sales and customer behavior characteristics in different dimensions.

[0099] The extraction of key information is called feature extraction;

[0100]

[0101] Among them, TF(x i ,d) reflects the importance of a word in the data information. The more frequently it appears, the higher the TF value, x i represents the word for calculation, and d represents the word x i The number of times it appears in the data information, It is used to measure the prevalence of a word in the entire corpus. The higher the value, the less common the word is in the corpus, expressed as the total number of texts in the corpus, DF(x i ) means that it contains word x i The number of texts;

[0102] If a word appears frequently in the data information to be detected and calculated, but appears less frequently in other texts, TF(x i ,The value of d) will be very high, indicating the importance of the word to the data information that needs to be detected and calculated;

[0103] On the contrary, if the word frequency is low or the word appears in many texts, then TF(x i,The value of ,d) will be low, indicating that the word is not very important to the data information that needs to be detected and calculated;

[0104] The calculation of sentiment analysis is as follows:

[0105] Assume there is a vocabulary containing words with positive sentiment (+1) and negative sentiment (-1). Given a text T = {w1,w2,…,w n}, where each word w i Corresponding to the sentiment score S(w i ), and each word w i That is the extracted key information, then the sentiment score S(T) of the text is expressed as:

[0106]

[0107] Among them, S(w i ) is the word w i The sentiment score of the text (positive number indicates positive sentiment, negative number indicates negative sentiment, and zero indicates neutral sentiment), S(T) is the sentiment score of the text, It is the sum of the sentiment scores of all keywords in the text;

[0108] It should be noted that the large language model analyzes and processes the form, and analyzes the data information based on potential associations and trends, so as to analyze the sales situation of goods and customer behavior characteristics in different dimensions.

[0109] In this embodiment, preferably, the step of automatically generating a statistical report in S5 is as follows:

[0110] The big language model automatically generates comprehensive and multi-dimensional statistical reports based on the potential correlations and trends in the data information;

[0111] Comprehensive and multi-dimensional statistical reports provide statistics on sales volume and sales of various commodities by quarter, province, and age;

[0112] Computational analysis of potential associations and trends in data information is achieved through neural networks:

[0113] Propagation from input layer to hidden layer:

[0114] Among them, a k is the input of the hidden layer neuron, w th is the weight from the input layer to the hidden layer, x i is the output of the input layer neuron, b k is the bias of the hidden layer neurons, n is the number of input layer neurons;

[0115] Propagation from hidden layer to output layer:

[0116] Among them, y i is the output of the output layer neuron, w hj is the weight from the hidden layer to the output layer, σ is the activation function, m is the number of neurons in the hidden layer, and b j is the bias of the output layer;

[0117] Through the output of the output layer, we can get the connection between associations and trends, and report the sales quantity and sales amount of various commodities by quarter, province, and age.

[0118] It should be noted that after analysis and processing, statistical reports are obtained to facilitate statistics on the quantity and amount of goods sold and improve the clarity of observation.

[0119] In this embodiment, preferably, the step of providing decision suggestions in S6 is as follows:

[0120] Based on comprehensive and multi-dimensional statistical reports, the large language model automatically generates insightful decision-making recommendations;

[0121] The big language model reviews the generated decision recommendations and assesses risks and uncertainties. Risk assessment methods include: risk matrix, probability analysis and scenario analysis;

[0122] The risk matrix evaluates the risk levels of different decision-making options by drawing a risk matrix;

[0123] The probability analysis calculates the probability of different events occurring in order to estimate the possible consequences of different decisions;

[0124] The scenario analysis analyzes the results of decision-making under different assumptions and scenarios;

[0125] The calculation of the probability analysis is as follows:

[0126]

[0127] Where P(A|B) is the probability of event A occurring given that event B has occurred, P(B|A) is the probability of event B occurring given that event A has occurred, P(A) is the prior probability of event A, and P(B) is the marginal probability of event B.

[0128] It should be noted that by evaluating the plan and determining whether it is feasible, it is easy to predict that a certain product is about to become a popular trend, negotiate with suppliers in advance to increase inventory, ensure sufficient supply, meet market demand, avoid missing sales opportunities, and implement and adjust according to actual conditions to achieve business goals.

[0129] The specific operation process of the present invention is as follows:

[0130] Step 1: Data collection and storage: Collect data through multiple low-code platforms, quickly create forms through visual interfaces and components, and then store the forms in the database;

[0131] Use the visual interface and components of the low-code platform to quickly create forms for collected customer purchase history, browsing behavior records, and evaluation records;

[0132] Use low-code to create an automated data collection process, and collect customer purchase history, browsing behavior records, and evaluation records through RPA nodes, data update nodes, and large model nodes;

[0133] Generate a form based on the customer's purchase history, browsing behavior record and evaluation record, and store the form in the database;

[0134] Step 2: Data accumulation: Monitor the number of forms in the database to determine whether the forms in the database meet the minimum number requirement for subsequent analysis;

[0135] Step 3: Data preprocessing: The low-code platform cleans the data and processes outliers in the forms in the database;

[0136] When the database detects that there are enough forms;

[0137] Low-code platforms begin data preprocessing;

[0138] First, process the form for outliers, set a threshold, and treat values ​​exceeding the threshold as outliers and remove them;

[0139] Then clean the data in the form, remove duplicate data, and fill in the gaps with the mean after eliminating missing values ​​and outliers;

[0140] Step 4: In-depth analysis of the large language model: The large language model performs in-depth analysis on the preprocessed data to extract key information. The large language model further identifies potential correlations and trends in the data.

[0141] The large language model conducts in-depth analysis on the pre-processed forms, extracts key information, and performs sentiment analysis on customer purchase reviews to determine customer satisfaction with the product.

[0142] The big language model further identifies potential connections and trends in data information, divides data by region, age, and spending power, and analyzes product sales and customer behavior characteristics in different dimensions.

[0143] Step 5: Automatically generate statistical reports: The large language model automatically generates comprehensive and multi-dimensional statistical reports;

[0144] The big language model automatically generates comprehensive and multi-dimensional statistical reports based on the potential correlations and trends in the data information;

[0145] Comprehensive and multi-dimensional statistical reports provide statistics on sales volume and sales of various commodities by quarter, province, and age;

[0146] Step 6: Provide decision suggestions: The large language model automatically generates insightful decision suggestions and reviews the generated decision suggestions;

[0147] Based on comprehensive and multi-dimensional statistical reports, the large language model automatically generates insightful decision-making recommendations;

[0148] The large language model reviews the generated decision recommendations and assesses risks and uncertainties. Risk assessment methods include: risk matrix, probability analysis and scenario analysis.

[0149] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent analysis and decision-making of a large language model for collecting data on a low-code platform, characterized in that: The following steps are involved: S1. Data collection and storage: Data is collected through multiple low-code platforms, and forms are quickly created through visual interfaces and components, and then stored in the database; S2. Data accumulation: Monitor the number of forms in the database to determine whether the forms in the database meet the minimum number requirement for subsequent analysis; S3. Data preprocessing: The low-code platform cleans the forms in the database and processes outliers; S4. Deep analysis of large language model: The large language model performs deep analysis on the preprocessed data to extract key information. The large language model further identifies potential correlations and trends in the data. S5. Automatically generate statistical reports: The large language model automatically generates comprehensive and multi-dimensional statistical reports; S6. Provide decision recommendations: The large language model automatically generates insightful decision recommendations and reviews the generated decision recommendations.

2. The method for intelligent analysis and decision-making of a large language model for collecting data on a low-code platform according to claim 1 is characterized by: The steps for data collection by multiple low-code platforms in S1 are as follows: Use the visual interface and components of the low-code platform to quickly create forms for collected customer purchase history, browsing behavior records, and evaluation records; Use low-code to create an automated data collection process, and collect customer purchase history, browsing behavior records, and evaluation records through RPA nodes, data update nodes, and large model nodes; Generate a form with the customer's purchase history, browsing behavior record and evaluation record, and store the form in the database.

3. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 2 is characterized by: The multiple low-code platforms collect data through crawler algorithms, and the collection steps are as follows: Send request: The crawler program sends a request to the server of the target website through HTTP request to obtain the HTML page of the web page; Parsing web page content: After receiving the HTML content of a web page, the crawler needs to parse it and extract useful information. The parsing targets include text and table data in the web page. Extract data: Use regular expressions, CSS selectors, and XPath methods to extract data from HTML pages, including customer purchase history, browsing behavior records, and evaluation records; Storing data: After parsing the extracted data, quickly create a form through the visual interface and components, and then store the form in the database.

4. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 1 is characterized by: The steps of data preprocessing in S3 are as follows: When the database detects that there are enough forms; Low-code platforms begin data preprocessing; First, process the form for outliers, set a threshold, and treat values ​​exceeding the threshold as outliers and remove them; Then clean the form data, remove duplicate data, and fill in the gaps with the mean after eliminating missing values ​​and outliers.

5. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 4 is characterized by: The outlier processing performs outlier detection on the data information in the form, and the outlier is calculated as follows: Among them, F is the outlier factor, D is the collected data information, and N j is the number of samples in the jth cluster group, d(x i ,c j ) is the data information x i The distance to the jth cluster center, and c j is the jth cluster center; For the distance away from the cluster center, it is set to average the distance of the data information from the center position, and the data results processed by the cluster analysis algorithm are calculated: That is Among them, c j is the jth cluster center, If the distance of the data information is greater than 3d, it is judged as abnormal data information.

6. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 4 is characterized by: The missing values ​​and outliers are removed and the mean is filled, and the mean is calculated as follows: Among them, x i is the i-th value in the data information; n is the total number of data information; Represents the sum of all data information.

7. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 1 is characterized by: The steps of deep analysis of the large language model in S4 are as follows: The large language model conducts in-depth analysis on the pre-processed forms, extracts key information, and performs sentiment analysis on customer purchase reviews to determine customer satisfaction with the product. The large language model further identifies potential connections and trends in data information, divides the data according to multiple dimensions such as region, age and consumption capacity, and analyzes product sales and customer behavior characteristics in different dimensions.

8. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 7 is characterized by: The steps of automatically generating a statistical report in S5 are as follows: The big language model automatically generates comprehensive and multi-dimensional statistical reports based on the potential correlations and trends in the data information; Comprehensive and multi-dimensional statistical reports count the sales quantity and sales amount of various commodities by quarter, province, and age.

9. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 8 is characterized by: The steps of providing decision suggestions in S6 are as follows: Based on comprehensive and multi-dimensional statistical reports, the large language model automatically generates insightful decision-making recommendations; The large language model reviews the generated decision recommendations and assesses risks and uncertainties. Risk assessment methods include: risk matrix, probability analysis and scenario analysis.

10. The method for intelligent analysis and decision-making of large language models for collecting data on a low-code platform according to claim 9 is characterized by: The risk matrix evaluates the risk levels of different decision-making options by drawing a risk matrix; The probability analysis calculates the probability of different events occurring in order to estimate the possible consequences of different decisions; The scenario analysis analyzes the results of decision-making under different assumptions and scenarios; The calculation of the probability analysis is as follows: Among them, P(A|B) is the probability of event A occurring when it is known that event B has occurred, P(B|A) is the probability of event B occurring when event A has occurred, P(A) is the prior probability of event A, and P(B) is the marginal probability of event B.

Images