Spreadsheet intelligent interaction method, device and equipment and medium

Abstract

The invention discloses a spreadsheet intelligent interaction method and device, equipment and a medium, relates to the technical field of artificial intelligence, and is used for solving the technical problems of low efficiency, high threshold, poor flexibility, poor user experience and the like in the existing Excel interaction technology. The method comprises the following steps: receiving a natural language instruction input by a user; wherein the natural language instruction is input through text or voice; processing the natural language instruction by adopting the trained large language model and the trained word mapping model in sequence to generate an Excel operation instruction; wherein the trained large language model is used for determining the intention of the user, the trained large language model is obtained based on a GPT model, and the GPT model has a Transform architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on a Word2Vec model; according to the Excel operation instruction, corresponding operation processing is carried out in the Excel software, so that the efficiency, the flexibility and the user experience can be improved, and the threshold can be reduced.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and provides a method, device, equipment and medium for intelligent interaction of spreadsheets. Background Art

[0002] As is well known, Excel spreadsheet software has a wide range of applications in data storage, analysis, calculation, etc. Among them, traditional Excel operations are completed through a graphical interface or programming with Visual Basic for Applications (VBA). However, although these methods can achieve a certain degree of automation and flexibility, they also have many limitations.

[0003] For example, for graphical interface operations, users need to manually click and enter formulas, which takes a lot of time and effort for complex analysis and chart making, especially when the amount of data is large; for VBA programming, writing scripts through VBA for automated operations requires certain programming knowledge, and the debugging and modification of the code are relatively cumbersome, and it is impossible to flexibly respond to changes in user needs; for third-party plugins, although some third-party plugins can expand the functions of Excel, these plugins usually need to be purchased or integrated separately, and users also need to learn how to use these plugins. Obviously, the existing Excel interaction has technical problems such as low efficiency, high threshold, poor flexibility, and poor user experience. Summary of the Invention

[0004] This application provides a method, device, equipment and medium for intelligent interaction of spreadsheets, which is used to solve the technical problems existing in the existing Excel interaction technology, such as low efficiency, high threshold, poor flexibility, and poor user experience.

[0005] On the one hand, a method for intelligent interaction of spreadsheets is provided, and the method includes: Receiving a natural language instruction input by a user; wherein, the natural language instruction is input through text or voice; Successively processing the natural language instruction by a trained large language model and a trained word mapping model to generate an Excel operation instruction; wherein, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the Generative Pre-trained Transformer (GPT) model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; Perform corresponding operation processing in Excel software according to the Excel operation instructions.

[0006] Optionally, the step of sequentially processing the natural language instruction by using the trained large language model and the trained word mapping model to generate the Excel operation instruction includes: Parse and transform the natural language instruction by using the trained large language model to obtain user intent words and key parameters; wherein, the key parameters include an analysis dimension, a time range, and a target variable; Perform vector mapping on the user intent words by using the trained word mapping model to determine the target Excel operation template with the highest similarity to the user intent words; Generate the Excel operation instruction according to the key parameters and the target Excel operation template.

[0007] Optionally, the step of performing corresponding operation processing in Excel software according to the Excel operation instruction includes: Automatically perform missing value filling, outlier detection, outlier processing, and data format standardization processing in Excel software according to the Excel operation instruction.

[0008] Optionally, the step of performing corresponding operation processing in Excel software according to the Excel operation instruction includes: Automatically generate a target chart in Excel software according to the Excel operation instruction and execute a specified calculation formula.

[0009] Optionally, after parsing and transforming the natural language instruction by using the trained large language model to obtain user intent words and key parameters, the method further includes: Determine a task analysis type according to the user intent words; wherein, the task analysis type includes descriptive analysis, diagnostic analysis, predictive analysis, classification analysis, clustering analysis, and association rule analysis; Determine a target analysis model according to the task analysis type and the key parameters; Analyze a target Excel file by using the target analysis model to obtain an analysis result.

[0010] Optionally, the step of determining a target analysis model according to the task analysis type and the key parameters includes: If it is determined that the task analysis type is predictive analysis, then determine linear regression, the time series model ARIMA, or LSTM as the target analysis model; If it is determined that the task analysis type is classification analysis, then logistic regression, decision tree, random forest, or support vector machine (SVM) is determined as the target analysis model; If it is determined that the task analysis type is clustering analysis, then K-means, hierarchical clustering, or DBSCAN is determined as the target analysis model; If it is determined that the task analysis type is association rule analysis, then Apriori or FP-Growth is determined as the target analysis model.

[0011] Optionally, after performing corresponding operation processing in Excel software according to the Excel operation instruction, the method further includes: Recording multiple operation behaviors of the user in Excel software; wherein, the operation behaviors include common format settings, formulas, functions, chart types, chart styles, data input, and editing modes; Extracting multiple key features from the multiple operation behaviors; wherein, the key features include operation frequency, usage time, and operation sequence; Constructing a user behavior feature vector according to the multiple key features; Constructing a user preference model according to the user behavior feature vector; wherein, the user preference model is obtained based on the Collaborative Filtering algorithm and the Latent Semantic Model; Automatically generating an Excel preference template that conforms to the user's habits according to the user preference model.

[0012] On the one hand, an intelligent spreadsheet interaction device is provided, and the device includes: A natural language instruction receiving unit, configured to receive a natural language instruction input by a user; wherein, the natural language instruction is input through text or voice; An operation instruction generating unit, configured to sequentially process the natural language instruction by using a trained large language model and a trained word mapping model to generate an Excel operation instruction; wherein, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; An operation processing unit, configured to perform corresponding operation processing in Excel software according to the Excel operation instruction.

[0013] On the one hand, an electronic device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, any of the above methods is implemented.

[0014] On the one hand, a storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, any of the above methods is implemented.

[0015] Compared with the prior art, the beneficial effects of this application are as follows: In this application, when performing intelligent interaction of spreadsheets, first, a natural language instruction input by the user can be received; wherein, the natural language instruction is input through text or voice; then, the trained large language model and the trained word mapping model can be sequentially used to process the natural language instruction to generate an Excel operation instruction; wherein, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; finally, corresponding operation processing can be performed in the Excel software according to the Excel operation instruction.

[0016] Based on this, in this application, since the user can complete complex Excel operations through natural language instructions, therefore, this application simplifies the operation, avoids manual clicks and complex programming, and significantly improves work efficiency; moreover, even users without programming experience or a data analysis background can easily interact with the system through natural language to complete professional data analysis and chart generation, greatly reducing the threshold and improving flexibility; in addition, since the user only needs to input a simple natural language instruction, the corresponding operation processing can be automatically completed in the Excel software, significantly enhancing the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the drawings in the following description are only the embodiments of this application. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0018] Figure 1 An electronic device provided for an embodiment of this application; Figure 2 A schematic diagram of a method for intelligent interaction of spreadsheets provided for an embodiment of this application; Figure 3It is a schematic diagram of an electronic spreadsheet intelligent interaction device provided by an embodiment of the present application.

[0019] Markings in the figure: 10 - Electronic spreadsheet intelligent interaction device, 101 - Processor, 102 - Memory, 103 - I / O interface, 104 - Database, 30 - Electronic spreadsheet intelligent interaction device, 301 - Natural language instruction receiving unit, 302 - Operation instruction generating unit, 303 - Operation processing unit, 304 - Analysis and prediction unit, 305 - Preference model building unit. Detailed implementation manners

[0020] To make the objectives, technical solutions and advantages of the present application clearer and more understandable, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application. Without conflict, the embodiments in the present application and the features in the embodiments can be arbitrarily combined with each other. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0021] As is well known, Excel spreadsheet software has wide applications in data storage, analysis, calculation, etc. Among them, traditional Excel operations are completed through a graphical interface or programming with Visual Basic for Applications (VBA). However, although these methods can achieve a certain degree of automation and flexibility, they also have many limitations.

[0022] For example, for graphical interface operations, users need to manually click and enter formulas, which takes a lot of time and effort for complex analysis and chart making, especially when the amount of data is large; for VBA programming, writing scripts through VBA for automated operations requires certain programming knowledge, and the debugging and modification of the code are relatively cumbersome, and it is impossible to flexibly respond to changes in user needs; for third-party plugins, although some third-party plugins can expand the functions of Excel, these plugins usually need to be purchased or integrated additionally, and users also need to learn how to use these plugins. Obviously, there are technical problems such as low efficiency, high threshold, poor flexibility and poor user experience in existing Excel interactions.

[0023] Based on this, an embodiment of the present application provides a method for intelligent interaction of spreadsheets. In this method, first, a natural language instruction input by the user can be received; among them, the natural language instruction is input through text or voice; then, the trained large language model and the trained word mapping model can be sequentially used to process the natural language instruction to generate an Excel operation instruction; among them, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; finally, corresponding operation processing can be performed in the Excel software according to the Excel operation instruction. Based on this, in the present application, since the user can complete complex Excel operations through natural language instructions, the present application simplifies the operation, avoids manual clicks and complex programming, and significantly improves work efficiency; moreover, even users without programming experience or a data analysis background can easily interact with the system through natural language, complete professional data analysis and chart generation, greatly reducing the threshold and improving flexibility; in addition, since the user only needs to input a simple natural language instruction, the corresponding operation processing can be automatically completed in the Excel software, significantly enhancing the user experience.

[0024] After introducing the design concept of the embodiment of the present application, the following briefly introduces the application scenarios applicable to the technical solution of the embodiment of the present application. It should be noted that the following introduced application scenarios are only used to illustrate the embodiment of the present application rather than to limit it. In the specific implementation process, the technical solution provided by the embodiment of the present application can be flexibly applied according to actual needs.

[0025] As Figure 1 shown, an electronic device provided by an embodiment of the present application, and this electronic device can specifically be an intelligent spreadsheet interaction device 10.

[0026] Among them, the spreadsheet intelligent interaction device 10 can perform intelligent interaction on spreadsheets. For example, it can be a personal computer (PC), a server, a laptop, etc. The spreadsheet intelligent interaction device 10 may include one or more processors 101, a memory 102, an I / O interface 103, and a database 104. Specifically, the processor 101 can be a central processing unit (CPU), or a digital processing unit, etc. The memory 102 can be a volatile memory, such as a random-access memory (RAM); the memory 102 can also be a non-volatile memory, such as a read-only memory, a flash memory, a hard disk drive (HDD), or a solid-state drive (SSD); or the memory 102 is any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 102 can be a combination of the above memories. Part of the program instructions of the spreadsheet intelligent interaction method provided in the embodiments of the present application can be stored in the memory 102. When these program instructions are executed by the processor 101, they can be used to implement the steps of the spreadsheet intelligent interaction method provided in the embodiments of the present application, so as to solve the technical problems existing in the existing Excel interaction technology, such as low efficiency, high threshold, poor flexibility, and poor user experience. The database 104 can be used to store data such as the trained large language model, the trained word mapping model, user intent words, key parameters, target charts, analysis results, and Excel preference templates involved in the solutions provided in the embodiments of the present application.

[0027] In the embodiments of the present application, the spreadsheet intelligent interaction device 10 can obtain natural language instructions through the I / O interface 103. Then, the processor 101 of the spreadsheet intelligent interaction device 10 will solve the technical problems existing in the existing Excel interaction technology, such as low efficiency, high threshold, poor flexibility, and poor user experience, according to the code instructions of the spreadsheet intelligent interaction method provided in the embodiments of the present application in the memory 102. In addition, data such as the trained large language model, the trained word mapping model, user intent words, key parameters, target charts, analysis results, and Excel preference templates can be stored in the database 104.

[0028] Of course, the method provided in the embodiments of the present application is not limited to Figure 1 the application scenarios shown, and can also be used in other possible application scenarios, which are not restricted by the embodiments of the present application. ForFigure 1 The functions that can be achieved by each device in the application scenario shown will be described together in the subsequent method embodiments, and will not be elaborated here too much. Next, the method of the embodiments of the present application will be introduced with reference to the accompanying drawings.

[0029] As Figure 2 shown, it is a schematic diagram of an intelligent spreadsheet interaction method provided by an embodiment of the present application. This method can be executed by the intelligent spreadsheet interaction device in Figure 1 . Specifically, the process of this method is introduced as follows.

[0030] Step 201: Receive a natural language instruction input by the user.

[0031] In the embodiment of the present application, the natural language instruction is input through text or voice.

[0032] For example, the natural language instruction input by the user is "Please generate a bar chart showing the quarterly sales in 2024". After the artificial intelligence interaction system receives the natural language input by the user, it can then prepare for subsequent processing. Step 202: Process the natural language instruction by successively using a trained large language model and a trained word mapping model to generate an Excel operation instruction.

[0033] In the embodiment of the present application, the trained large language model is used to determine the user's intention. The trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and multiple layers of self-attention mechanisms to capture the context relationship in the natural language instruction. And this large language model is pre-trained on a large-scale text data to learn the statistical characteristics and semantic structure of the language, so as to have the ability to understand and generate natural language; the trained word mapping model is generated based on the Word2Vec model to map words to a continuous vector space and capture the semantic relationship between words. And this word mapping model is trained on a large-scale text corpus to learn the context information of words, so as to map words with similar semantics to similar vectors.

[0034] Specifically, first, a trained large language model can be used to parse and transform natural language instructions to obtain user intent words and key parameters; among them, the key parameters include analysis dimensions, time ranges, and target variables. For example, for the natural language instruction "Please generate a bar chart showing quarterly sales in 2024", the trained large language model can analyze the dependency relationships between words, aiming to identify the dependencies between words in the sentence and construct a dependency structure tree centered around the verb. In the above natural language instruction, "generate" can be identified as the main verb, "bar chart" as its object, "showing" as a verb phrase modifying the "bar chart", and "quarterly sales in 2024" as the object of "showing". Through this analysis, the subordinate relationships between each word can be clarified, thereby accurately understanding the user's intent. Based on this, the user intent of the above natural language instruction can be identified as "generate a bar chart", and key parameters can be extracted from the natural language instruction. For example, "2024" is used as the time range, and "quarterly sales" is used as the data type.

[0035] Then, a trained word mapping model can be used to vector map the user intent words to determine the target Excel operation template with the highest similarity to the user intent words. That is, a trained word mapping model can be used to match the user's intent with predefined Excel operation templates. Specifically, the user intent words and each Excel operation template can be mapped into vectors, and the semantic similarity between the user intent vector and each Excel operation template vector can be calculated (for example, using the cosine similarity calculation method to calculate the semantic similarity), so as to select the target Excel operation template with the highest similarity to the user intent vector (that is, the highest similarity of the user intent words).

[0036] Finally, an Excel operation instruction can be generated based on the key parameters and the target Excel operation template. That is, according to the extracted key parameters, the slots in the target Excel operation template can be filled to generate specific executable Excel operation instructions or scripts, such as OfficeScripts, VBA, or Python code, thereby realizing the automated operation of Excel.

[0037] Step 203: Perform corresponding operation processing in the Excel software according to the Excel operation instruction.

[0038] In a possible implementation, according to the Excel operation instructions, Excel file parsing and operations can also be performed. That is, various types of data in the Excel file can be intelligently parsed and automatically operated on. Specifically, the trained large language model described above can be combined with a table parsing algorithm (for example, a method of identifying data blocks based on structured document tree traversal and cell attribute recognition, merged cell processing, format analysis, etc.) to parse the worksheets, cells, row and column structures in the Excel file, identify structural information such as data regions, titles, and merged cells, and identify the types of content in each cell, such as text, numbers, dates, formulas, and charts, thereby providing a basis for subsequent operations.

[0039] Based on this, in the embodiments of the present application, when performing corresponding operation processing in the Excel software according to the Excel operation instructions (that is, according to the natural language instructions input by the user), operations such as missing value filling, outlier detection, outlier processing, and data format standardization processing can be automatically performed in the Excel software to improve data quality.

[0040] In addition, in the embodiments of the present application, when performing corresponding operation processing in the Excel software according to the Excel operation instructions (that is, based on user requirements), a target chart (such as a bar chart, line chart, and pie chart, etc.) can also be automatically generated in the Excel software according to the Excel operation instructions, and specified calculation formulas can be executed, supporting complex nested functions and cross-sheet references.

[0041] In a possible implementation, since the trained large language model of the present application supports users to perform advanced data analysis tasks through natural language instructions. That is, through semantic analysis, the analysis intent of the user can be identified, such as "analyze the quarterly trend of sales data" or "predict the sales volume for the next month". Key parameters, such as analysis dimensions, time ranges, target variables, etc., are extracted from the natural language instructions input by the user to construct a parameter set for the analysis task.

[0042] Therefore, in the embodiments of the present application, after parsing and converting the natural language instructions using the trained large language model to obtain the user intent words and key parameters, "data analysis and prediction" can also be performed.

[0043] Specifically, first, the task analysis type can be determined based on the user's intent words. Among them, the task analysis types include descriptive analysis (used to summarize and describe the basic characteristics of data, for example, calculating the mean, standard deviation, frequency distribution, etc.), diagnostic analysis (used to identify abnormal patterns or trends in data, for example, detecting abnormal fluctuations in sales), predictive analysis (used to predict future trends or values, for example, predicting next month's sales), classification analysis (used to classify data into different categories, for example, classifying customers as high-value or low-value customers based on their behavior), clustering analysis (used to group data so that data points in the same group have high similarity, for example, market segmentation), and association rule analysis (used to discover the association relationships between data, for example, finding that people who buy bread may also buy butter in basket analysis); then, an appropriate target analysis model can be automatically determined based on the task analysis type and key parameters; next, the determined target analysis model can be trained on the user data and verified to ensure the accuracy and reliability of the target analysis model; finally, the target analysis model can be used to analyze the target Excel file to obtain the analysis results and generate a result explanation described in natural language to help the user understand the analysis results.

[0044] Further, when determining the target analysis model based on the task analysis type and key parameters, if the task analysis type is determined to be predictive analysis, linear regression, time series model ARIMA, or LSTM can be determined as the target analysis model; if the task analysis type is determined to be classification analysis, logistic regression, decision tree, random forest, or support vector machine SVM can be determined as the target analysis model; if the task analysis type is determined to be clustering analysis, K-means, hierarchical clustering, or DBSCAN can be determined as the target analysis model; if the task analysis type is determined to be association rule analysis, Apriori or FP-Growth can be determined as the target analysis model.

[0045] In a possible implementation manner, after performing corresponding operation processing in Excel software according to the Excel operation instruction, the present application can also automatically generate an Excel template or style that meets the user's needs based on the user's historical operations and preferences, that is, "personalized template generation" can also be performed.

[0046] Specifically, first, multiple operation behaviors of the user in Excel software can be recorded. Among them, the operation behaviors include common format settings (such as font, color, and border, etc.), formulas, functions, chart types, chart styles, data input, and editing modes.

[0047] Then, multiple key features can be extracted from these multiple operation behaviors of the mobile phone; among them, the key features include operation frequency, usage time, and operation sequence. Next, a user behavior feature vector can be constructed based on these multiple key features.

[0048] Then, a user preference model can be constructed based on the user behavior feature vector; among them, the user preference model is obtained based on the Collaborative Filtering algorithm and the Latent Semantic Model.

[0049] Finally, an Excel preference template that conforms to the user's habits can be automatically generated according to the user preference model, such as preset formats, formulas, and charts, etc.

[0050] In addition, the user can also adjust the generated Excel preference template through natural language instructions and update the template content to achieve highly personalized template customization. Moreover, various operations of this application are performed on an intelligent operation interface, and the user only needs to input simple instructions to automatically complete background data processing and result display.

[0051] In summary, this application has the following advantages: (1) Improve efficiency: Since the user can complete complex Excel operations through natural language instructions, the operation is simplified, manual clicks and complex programming are avoided, and the work efficiency is significantly improved.

[0052] (2) Lower the threshold: Since various operations of Excel files can be automatically executed, even users without programming experience or a data analysis background can easily interact with the system through natural language to complete professional data analysis and chart generation, greatly reducing the usage threshold of Excel software.

[0053] (3) High flexibility: Since the operations can be flexibly adjusted according to the specific needs of the user, it can not only support a variety of complex data processing and analysis tasks, but also automatically adapt to different Excel worksheet structures.

[0054] (4) High sense of experience: Since this application not only provides an intelligent operation interface, but also can automatically complete background data processing and result display by only inputting simple instructions, the user experience can be significantly improved.

[0055] Based on the same inventive concept, an embodiment of this application provides a spreadsheet intelligent interaction device 30, as Figure 3 shown, the spreadsheet intelligent interaction device 30 includes: A natural language instruction receiving unit 301 for receiving natural language instructions input by a user; wherein, the natural language instructions are input via text or voice; An operation instruction generating unit 302 for sequentially processing the natural language instructions using a trained large language model and a trained word mapping model to generate Excel operation instructions; wherein, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; An operation processing unit 303 for performing corresponding operation processing in the Excel software according to the Excel operation instructions.

[0056] Optionally, the operation instruction generating unit 302 is further configured to: Parse and transform the natural language instructions using the trained large language model to obtain user intention words and key parameters; wherein, the key parameters include analysis dimensions, time ranges, and target variables; Perform vector mapping on the user intention words using the trained word mapping model to determine the target Excel operation template with the highest similarity to the user intention words; Generate Excel operation instructions based on the key parameters and the target Excel operation template.

[0057] Optionally, the operation processing unit 303 is further configured to: Automatically perform missing value filling, outlier detection, outlier processing, and data format standardization processing in the Excel software according to the Excel operation instructions.

[0058] Optionally, the operation processing unit 303 is further configured to: Automatically generate a target chart and execute a specified calculation formula in the Excel software according to the Excel operation instructions.

[0059] Optionally, the spreadsheet intelligent interaction device 30 further includes an analysis and prediction unit 304 for: Determine the task analysis type according to the user intention words; wherein, the target analysis models include descriptive analysis, diagnostic analysis, predictive analysis, classification analysis, clustering analysis, and association rule analysis; Determine the target analysis model according to the task analysis type and the key parameters; Analyze the target Excel file using the target analysis model to obtain an analysis result.

[0060] Optionally, the analysis and prediction unit 304 is further configured to: If it is determined that the task analysis type is predictive analysis, then linear regression, time series model ARIMA, or LSTM is determined as the target analysis model; If it is determined that the task analysis type is classification analysis, then logistic regression, decision tree, random forest, or support vector machine SVM is determined as the target analysis model; If it is determined that the task analysis type is clustering analysis, then K-means, hierarchical clustering, or DBSCAN is determined as the target analysis model; If it is determined that the task analysis type is association rule analysis, then Apriori or FP-Growth is determined as the target analysis model.

[0061] Optionally, the spreadsheet intelligent interaction device 30 further includes a preference model construction unit 305 for: Recording multiple operation behaviors of the user in the Excel software; wherein, the operation behaviors include common format settings, formulas, functions, chart types, chart styles, data input, and editing modes; Extracting multiple key features from the multiple operation behaviors; wherein, the key features include operation frequency, usage time, and operation sequence; Constructing a user behavior feature vector based on the multiple key features; Constructing a user preference model based on the user behavior feature vector; wherein, the user preference model is obtained based on the Collaborative Filtering algorithm and the Latent Semantic Model; Automatically generating an Excel preference template that conforms to the user's habits according to the user preference model.

[0062] The spreadsheet intelligent interaction device 30 can be used to execute Figure 2 the methods executed in the embodiments shown, and therefore, for the functions that can be achieved by each functional module of the spreadsheet intelligent interaction device 30, reference can be made to Figure 2 the description of the embodiments shown, which will not be elaborated here.

[0063] In some possible implementation manners, each aspect of the method provided in this application can also be implemented in the form of a program part, which includes program code. When the program part runs on a computer device, the program code is used to cause the computer device to execute the steps in the methods according to various exemplary implementation manners of this application described above in this specification. For example, the computer device can execute the methods executed in the embodiments shown in Figure 2 the embodiments shown.

[0064] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including those of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs and other various media that can store program codes. Alternatively, if the above integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent components, they can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of software components. The computer software components are stored in a storage medium and include several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. And the aforementioned storage medium includes: mobile storage devices, ROM, RAM, magnetic disks, or optical discs and other various media that can store program codes.

[0065] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0066] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these changes and modifications.

Claims

1. An intelligent interaction method for spreadsheets, characterized in that, The method includes: Receiving a natural language instruction input by a user; wherein, the natural language instruction is input through text or voice; Successively processing the natural language instruction by a trained large language model and a trained word mapping model to generate an Excel operation instruction; wherein, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; Performing corresponding operation processing in the Excel software according to the Excel operation instruction.

2. The method according to claim 1, wherein The step of successively processing the natural language instruction by a trained large language model and a trained word mapping model to generate an Excel operation instruction includes: Parsing and converting the natural language instruction by a trained large language model to obtain user intention words and key parameters; wherein, the key parameters include an analysis dimension, a time range, and a target variable; Performing vector mapping on the user intention words by a trained word mapping model to determine a target Excel operation template with the highest similarity to the user intention words; Generating the Excel operation instruction according to the key parameters and the target Excel operation template.

3. The method according to claim 1, wherein The step of performing corresponding operation processing in the Excel software according to the Excel operation instruction includes: Automatically performing missing value filling, outlier detection, outlier processing, and data format standardization processing in the Excel software according to the Excel operation instruction.

4. The method according to claim 1, characterized in that The step of performing corresponding operation processing in the Excel software according to the Excel operation instruction includes: Automatically generating a target chart in the Excel software according to the Excel operation instruction and executing a specified calculation formula.

5. The method according to claim 2, wherein After parsing and converting the natural language instruction by a trained large language model to obtain user intention words and key parameters, the method further includes: Determining a task analysis type according to the user intention words; wherein, the task analysis type includes descriptive analysis, diagnostic analysis, predictive analysis, classification analysis, clustering analysis, and association rule analysis; Determining a target analysis model according to the task analysis type and the key parameters; Analyzing a target Excel file by using the target analysis model to obtain an analysis result.

6. The method according to claim 5, wherein The step of determining a target analysis model according to the task analysis type and the key parameters includes: If it is determined that the task analysis type is predictive analysis, determining a linear regression, a time series model ARIMA, or an LSTM as the target analysis model; If it is determined that the task analysis type is classification analysis, determining a logistic regression, a decision tree, a random forest, or a support vector machine SVM as the target analysis model; If it is determined that the task analysis type is clustering analysis, then K-means, hierarchical clustering, or DBSCAN is determined as the target analysis model; If it is determined that the task analysis type is association rule analysis, then Apriori or FP-Growth is determined as the target analysis model.

7. The method according to claim 1, characterized in that, After performing corresponding operation processing in Excel software according to the Excel operation instruction, the method further includes: Recording multiple operation behaviors of the user in Excel software; wherein, the operation behaviors include common format settings, formulas, functions, chart types, chart styles, data input, and editing modes; Extracting multiple key features from the multiple operation behaviors; wherein, the key features include operation frequency, usage time, and operation sequence; Constructing a user behavior feature vector according to the multiple key features; Constructing a user preference model according to the user behavior feature vector; wherein, the user preference model is obtained based on the Collaborative Filtering algorithm and the Latent Semantic Model; Automatically generating an Excel preference template that conforms to the user's habits according to the user preference model.

8. An intelligent spreadsheet interaction device, characterized in that, The device includes: A natural language instruction receiving unit, configured to receive a natural language instruction input by a user; wherein, the natural language instruction is input via text or voice; An operation instruction generating unit, configured to process the natural language instruction by sequentially using a trained large language model and a trained word mapping model to generate an Excel operation instruction; wherein, the trained large language model is used to determine the user's intention, the trained large language model is obtained based on the GPT model, and the GPT model has a Transformer architecture and a multi-layer self-attention mechanism; the trained word mapping model is generated based on the Word2Vec model; An operation processing unit, configured to perform corresponding operation processing in Excel software according to the Excel operation instruction.

9. An electronic device, characterized in that, The device includes: A memory, configured to store program instructions; A processor, configured to call the program instructions stored in the memory and execute the method according to any one of claims 1-7 according to the obtained program instructions.

10. A storage medium, characterized in that, The storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute the method according to any one of claims 1-7.

Images