JP7692966B2 - Information providing device, information providing method, and information providing program - Google Patents

Description

The present invention relates to an information providing device, an information providing method, and an information providing program.

In recent years, so-called no-code data analysis tools (e.g., applications) have been provided (see, for example, Patent Literature 1 or Non-Patent Literature 1). No-code data analysis tools have the advantage that they do not require programming skills and can be used intuitively even by beginners.

JP 2023-27964 A

“No-code AI development tool Node-AI”, [online], [Retrieved July 21, 2023], Internet <https://sdpf.ntt.com/services/nodeai/>

However, traditional data analysis tools can be difficult for users to use.

For example, even with no-code data analysis tools, it is necessary to understand the concepts and methods of data analysis in order to use them effectively. When using a tool, users may conduct research to understand concepts and methods related to data analysis using the Internet or books, or through question and answer sessions with experts. However, if the results of the research that the user obtains do not match their own skill level, the research may be meaningless.

In particular, beginners in data analysis have difficulty distinguishing whether the results of their own research on analytical methods, etc., are basic or applied. For this reason, it is easy to run into the problem of not knowing whether the results of the research really match your own skill level.

The same goes for how to use analytical tools; for example, a beginner in data analysis may be confused when taught how to set up tools for advanced users.

In order to solve the above-mentioned problems and achieve the objective, the information providing device is characterized by having a level selection unit that selects a level related to data analysis for a user, a prompt generation unit that generates a prompt for causing a language model to generate a sentence corresponding to the level, which is a sentence for supporting the use of the data analysis tool, based on a request received from a terminal device on which the data analysis tool is executed and the level selected by the level selection unit, a sentence generation unit that generates a sentence by inputting the prompt generated by the prompt generation unit into the language model, and a display control unit that displays the sentence generated by the sentence generation unit on the terminal device.

The present invention can improve the ease of use of data analysis tools.

FIG. 1 is a diagram illustrating an example of the configuration of an information providing system according to the first embodiment. FIG. 2 is a diagram showing an example of a screen. FIG. 3 is a diagram showing an example of a screen. FIG. 4 is a diagram showing an example of the screen. FIG. 5 is a diagram showing a modification of the information providing system according to the first embodiment. FIG. 6 is a flowchart showing a flow of processing of the information providing system according to the first embodiment. FIG. 7 is a diagram illustrating an example of the configuration of a computer that executes an information provision program.

Below, embodiments of the information providing device, information providing method, and information providing program according to the present application will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiments described below.

[First embodiment]
[Configuration of the first embodiment]
The configuration of the information providing system will be described with reference to Fig. 1. Fig. 1 is a diagram showing an example of the configuration of the information providing system according to the first embodiment. As shown in Fig. 1, the information providing system 1 includes an information providing device 10 and a terminal device 20.

The information providing device 10 is a server connected to a terminal device 20 via a network. The terminal device 20 is a personal computer (PC), a tablet terminal, a smartphone, etc.

The information providing device 10 has the functionality of a data analysis tool such as Node-AI described in Non-Patent Document 1. The information providing device 10 may have the same functionality as the information processing device described in Patent Document 1. For example, a user executes a data analysis tool via a terminal device 20 to generate canvas data.

Node-AI employs flow-based programming, which allows designing a processing pipeline for data without coding. Node-AI also allows intuitive description of a processing pipeline for data by connecting the data to be processed with the functions that can execute each process. In this embodiment, a function that executes a specific process on data is referred to as a "card." Furthermore, data for which a pipeline is described is referred to as "canvas data."

In addition to the function of the data analysis tool, the information providing device 10 also has a function of providing natural language sentences to the user. The information providing device 10 supports the user in using the data analysis tool by providing messages.

The information providing device 10 provides written answers to the user's questions about the data analysis tool, hints on how to use the data analysis tool, recommended actions for the user, insights into the data, and corrections of the user's errors. At this time, the information providing device 10 provides written text that is appropriate for the user's level of data analysis.

As shown in FIG. 1, the information providing device 10 has an interface unit 11, a request processing unit 12, a prompt generating unit 13, a history DB 14, a language model group 15, a sentence generating unit 16, and a level selection unit 17.

The interface unit 11 exchanges data with the terminal device 20. For example, the interface unit 11 accepts a request from the terminal device 20. Also, for example, the interface unit 11 causes the terminal device 20 to display a screen.

The interface unit 11 receives the text of a query entered by a user or the operation content of a data analysis tool as a request from the terminal device 20. The interface unit 11 passes the received request to the request processing unit 12.

When the request is an operation of a data analysis tool, the request processing unit 12 obtains the result for the request. The result is canvas data that the data analysis tool returns in response to the operation.

The request processing unit 12 saves the request and the result in the history DB 14. The request processing unit 12 also passes the request and the result to the prompt generation unit 13 and the level selection unit 17. If there are no results, the request processing unit 12 saves the request in the history DB 14 and passes the request to the prompt generation unit 13 and the level selection unit 17.

The level selection unit 17 selects the user's level regarding data analysis. The level is information that represents the user's level of knowledge regarding data analysis, proficiency with data analysis tools, etc. in stages. Here, the level is represented in two stages: advanced and beginner. However, the number of levels may be three or more.

The level selection unit 17 selects a level based on an input from a user. In this case, the interface unit 11 accepts the level input via the terminal device 20.

The level selection unit 17 may also automatically select the level. For example, the level selection unit 17 selects the level based on the user's past operation content in the data analysis tool and the accuracy of data analysis by a process generated based on the past operation content.

The prompt generation unit 13 generates a prompt for causing the language model to generate a sentence to assist in the use of the data analysis tool, based on a request (or a request and a result) received from the terminal device 20 on which the data analysis tool is executed. The prompt generation unit 13 passes the generated prompt to the sentence generation unit 16.

The prompt generation unit 13 generates a prompt for the language model to generate a sentence appropriate to the level, which is a sentence to assist in the use of the data analysis tool, based on a request received from the terminal device on which the data analysis tool is executed and the level selected by the level selection unit 17.

The sentence generation unit 16 generates a sentence by inputting the prompt generated by the prompt generation unit 13 into a language model included in the language model group 15. The sentence generation unit 16 outputs the obtained sentence to the terminal device 20 via the interface unit 11.

As shown in FIG. 1, the language model group 15 includes an advanced language model 151 and a beginner language model 152. The advanced language model 151 is a language model that corresponds to the advanced level. The beginner language model 152 is a model that corresponds to the beginner level.

The sentence generation unit 16 can generate sentences by inputting a prompt into a language model corresponding to a level selected by the level selection unit 17, among the language models corresponding to each of the multiple levels. The language model to be input may be specified by the prompt.

Even when the same prompt is input, the advanced language model 151 and the beginner language model 152 output different sentences.

The level selection unit 17 further selects a role. The prompt generation unit 13 generates a prompt to cause the language model to generate a sentence in a writing style according to the role. For example, the writing style may be either the desumasu style or the dearu style, whether honorific language is used, whether interjections are used, whether an imperative tone is used, etc.

The role may be selected by the user via a screen as shown in FIG. 2. FIG. 2 is a diagram showing an example of the screen. The interface unit 11 causes the terminal device 20 to display the role selection screen 54 of FIG. 2. "Instructor", "Secretary", and "University Professor" are examples of roles. The user selects one of the roles displayed on the role selection screen 54. The selected role is notified to the prompt generation unit 13 by the level selection unit 17.

Here, the language models included in the language model group 15 are large language models (LLMs). The language models may be neural networks. In addition, the language models are, for example, ChatGPT (registered trademark) (reference URL: https://openai.com/blog/chatgpt).

The language model performs inference based on the input prompt and outputs the natural language sentence obtained through inference.

Here, the data analysis tool generates a series of processes (pipeline) for data analysis using a machine learning model in response to user operations. The machine learning model that performs data analysis is called an analysis model. For example, the data analysis tool outputs an executable file that executes a series of processes.

The process can be divided into three parts: data preprocessing, training an analytical model using the data (updating parameters), and evaluating the trained analytical model.

The language models may include language models tuned for each of the three processes. For example, the language model group 15 includes a language model tuned for preprocessing, a language model tuned for learning, and a language model tuned for evaluation.

For example, a language model tuned for preprocessing learns prompts based on the operation content that specifies the preprocessing of the data used for training, and written advice from an experienced user for the operation content. A language model tuned for preprocessing can generate sentences that are more appropriate for preprocessing than other language models.

The division of the processes of the data analysis tool is not limited to the above, and may be further subdivided. The language model group 15 includes language models tuned to each of the subdivided processes. For example, the language model group 15 may include language models tuned to each of the cards in the data analysis tool.

The language model group 15 may also include language models that are subdivided into processes and correspond to each of a plurality of levels. For example, the language model group 15 includes six language models: a beginner-level language model tuned for preprocessing, a beginner-level language model tuned for learning, a beginner-level language model tuned for evaluation, an advanced-level language model tuned for preprocessing, an advanced-level language model tuned for learning, and an advanced-level language model tuned for evaluation.

The sentence generation unit 16 selects an appropriate language model based on the prompt and generates a sentence using the selected language model. For example, if the prompt contains the keyword "preprocessing" and the user's level is beginner, the sentence generation unit 16 selects a beginner-level language model tuned to preprocessing.

Below, the operation of the information providing device 10 will be described in detail, divided into cases where the request is the text of an inquiry entered by a user and where the request is the operation content of a data analysis tool.

(1. If the request is a query text entered by the user)
As shown in Fig. 3, the screen (user interface) of the data analysis tool displays canvas data created in response to the operation content. Fig. 3 is a diagram showing an example of the screen. The screen is displayed by the terminal device 20 in response to the control of the information providing device 10.

Canvas data represents a series of processes (pipeline) related to data analysis using an analytical model. First, data and functions specified by user operations are placed as cards on the canvas data.

In the example of FIG. 3, cards 511, 512, and 513 are arranged. Card 511 represents data to be input into the analysis model. Cards 512 and 513 represent pre-processing of the data. Furthermore, comments related to the cards are displayed in areas 521 and 522.

The chat screen 53 is a screen that displays text generated by the information providing device 10. The chat screen 53 may be displayed as a pop-up superimposed on a screen that displays the canvas data. For example, the interface unit 11 functions as a display control unit that causes the terminal device 20 to display the canvas data and the chat screen 53.

Box 5311 displays the text of the inquiry entered by the user. The text of the inquiry is entered via input field 533. Icon 532 is an image indicating a pseudo-subject of the statement (a bot) when the text generated by the information providing device 10 is output as a statement in chat. Furthermore, box 5312 displays the text generated by the information providing device 10.

In the example of FIG. 3, the user inputs the text "Please tell me the difference between lasso and ridge." In response, the information providing device 10 generates and outputs the following sentence: "You don't know the difference between lasso and ridge? Have you studied it properly, you idiot? First of all, lasso is an abbreviation for Least Absolute Shrinkage and Selection Operator, and is a regularization method that selects features by moving coefficients closer to 0. On the other hand, ridge is a method that prevents overfitting by using L2 regularization. In other words, lasso is used when you want to remove specific features, and ridge is suitable when you want to stabilize the entire model. Do you get it, you idiot?"

The prompt generator 13 receives text input as a request by a user who uses the data analysis tool via the terminal device 20. The prompt generator 13 generates a prompt based on the text input by the user. The prompt generator 13 generates a prompt that includes the following elements:
(1) Instructions: the specific task or command you want the language model to perform; (2) Context: external information or additional context to guide the model to a better response; (3) Input data: the request; (4) Output indicator: the type or format of the output.

In the example of FIG. 3, the request is text entered by the user. For example, when text is entered in the input field 533 of the chat screen 53, the request processing unit 12 determines that the request is text entered by the user and notifies the prompt generation unit 13 of the determination result.

If the request is text entered by the user, the prompt generation unit 13 sets element (1) to "generate a sentence to reply to the user." If the user's statement displayed on the chat screen 53 (e.g., the text in box 5311) matches the request, the prompt generation unit 13 may determine that the request is text entered by the user.

The prompt generation unit 13 also determines element (2) through the following procedure. First, the prompt generation unit 13 extracts predetermined terms related to data analysis from the text entered by the user. It is assumed that "lasso" and "ridge" are predetermined terms related to data analysis. Then, the prompt generation unit 13 adds the fact that "lasso" and "ridge" are terms related to data analysis to element (2) as external information or additional context.

This, for example, prevents the language model from generating sentences that simply answer the difference between "lasso" and "ridge" as character strings, or from generating sentences that explain the difference in meaning as listed in a dictionary, and encourages the language model to generate sentences that explain the difference in data analysis. In other words, element (2) makes it easier to provide the information that users need.

Element (3) is the text entered by the user. However, the prompt generation unit 13 may set the text to element (3) after correcting obvious errors (spelling mistakes, typos, etc.).

The prompt generation unit 13 also sets element (4) to "a sentence in a natural language." The prompt generation unit 13 may also set element (4) to include a language, such as "an English sentence" or "a Japanese sentence." For example, the prompt generation unit 13 sets element (4) to be a sentence in a language that is already set in the data analysis tool itself, or in the same language as the language of the text already displayed on the chat screen 53.

In the example of FIG. 3, the prompt generating unit 13 generates a prompt, for example, as follows:
(1) Command: Generate a sentence to respond to the user. (2) Context: "lasso" and "ridge" are data analysis terms. (3) Input data: "What is the difference between lasso and ridge?"
(4) Output indicator: Japanese text

As described herein, if the request is a query text entered by a user, the information providing device 10 may generate the sentence without considering the level.

(2. When the request is for the operation of a data analysis tool)
An example of a case where a request is an operation for a data analysis tool will be described with reference to Fig. 4. Fig. 4 is a diagram showing an example of a screen. Canvas data is displayed on the screen of Fig. 4.

In the example of FIG. 4, card 514 is placed. Card 514 represents that the analytical model is a "linear model." An analytical model may be rephrased as an algorithm for analyzing data.

Boxes 5313 and 5314 on the chat screen 53 display text generated by the information providing device 10. In the example of FIG. 4, the information providing device 10 generates text in response to the user's operation and the selected level, and outputs the generated text.

The level selection unit 17 selects a level. For example, it selects a level specified by the user. For example, the level is either advanced or beginner.

The prompt generation unit 13 receives the operation content for the data analysis tool as a request. For example, the prompt generation unit 13 generates a prompt in response to the operation content (request) of "specify that the analysis model is a linear model." The elements included in the prompt are as described above.

When the request is the user's operation content, the prompt generation unit 13 sets element (1) to "generate advice sentences according to the operation content." When the user's statement displayed on the chat screen 53 (e.g., the text in box 5311) does not match the request, the prompt generation unit 13 may determine that the request is the user's operation content.

The prompt generation unit 13 also determines element (2) through the following procedure. First, the prompt generation unit 13 determines whether a specific operation has been performed before the target operation. For example, when the operation of the request is the specification of an analysis model, the prompt generation unit 13 determines whether preprocessing has already been specified. The prompt generation unit 13 determines the result of this determination as element (2). Furthermore, element (3) is the operation itself.

The prompt generation unit 13 also sets element (4) to be a "sentence in natural language." The prompt generation unit 13 may also set element (4) to be a language, such as an "English sentence" or an "Japanese sentence."

The prompt generation unit 13 also adds the level selected by the level selection unit 17 to element (4). Here, it is assumed that the level selection unit 17 has selected "Beginner" as the level.

In the example of FIG. 4, the prompt generating unit 13 generates a prompt, for example, as follows:
(1) Command: Generate advice sentences according to the operation content (2) Context: Preprocessing not specified (3) Input data: Specify "linear model" as the analysis model (4) Output indicator: Japanese sentence, Level: Beginner

In this way, the prompt generation unit 13 receives, as a request, operation content that specifies an algorithm for analyzing data in a data analysis tool that generates a series of processes related to data analysis, and generates a prompt that commands the generation of advice sentences according to the operation content. Furthermore, the sentence generation unit 16 inputs the prompt into a language model, and generates sentences according to the level, which indicate operations that should be performed after the operation content (e.g., specifying preprocessing). Furthermore, the style of the sentence changes depending on the selected role.

When the language model group 15 includes one language model, the sentence generation unit 16 can input the level specified in element (4) (e.g., "Level: Beginner") directly into the language model as part of the prompt.

On the other hand, as shown in FIG. 1, when the language model group 15 includes an advanced language model 151 and a beginner language model 152, the language model to input the prompt may be determined according to the level specified in element (4).

Here, we will explain the difference between sentences generated by the language model when the level is "advanced" (hereinafter, "sentences for advanced learners") and sentences generated by the language model when the level is "beginner" (hereinafter, "sentences for beginners"). The difference explained here arises from the inference process performed by the language model.

For example, a sentence for advanced learners contains less information than a sentence for beginners. For example, the sentence shown in box 5313 in Figure 4 is considered to be a sentence for beginners. When a prompt in which "Level: Beginner" in element (4) in the prompt in Figure 4 is replaced with "Level: Advanced" is input to the language model, a sentence is generated in which the part "And the next thing to do is preprocessing" is omitted from the sentence shown in box 5313.

Texts for advanced learners can be said to be more difficult than texts for beginners. In other words, the level determines the difficulty of the text.

Also, for example, texts aimed at beginners tend to evoke stronger emotions in users than texts aimed at advanced users.

For example, the prompt generation unit 13 generates a prompt to cause the language model to generate a sentence that will move the user's emotions more strongly the lower the level selected by the level selection unit 17. For example, the prompt generation unit 13 adds the phrase "emotionally" to element (4) of the prompt along with "Level: Beginner".

For example, the language model generates sentences that strongly affect the user's emotions by using exclamations and exaggerating the user's operations. For example, when praising an operation that specifies a certain preprocessing, a sentence with a more matter-of-fact impression such as "I think that preprocessing is good" is added to sentences for advanced users, while a sentence with an exaggerated impression such as "I think that preprocessing is very good!" is added to sentences for beginners.

The level selection unit 17 may automatically select a level using at least one of the requests and results obtained from the request processing unit 12 and past requests and results obtained from the history DB 14.

The operation contents are, for example, a specified analysis model (linear model, Lasso, Ridge, etc.), preprocessing, set parameters, etc. Furthermore, the results are, for example, whether the processing was successful or not, an accuracy index (for example, RMSE (Root Mean Squared Error) of a linear model), etc. Furthermore, the operation contents and results are expressed as canvas data.

For example, if the accuracy index shown in the result obtained from the request processing unit 12 is equal to or greater than a threshold, the level selection unit 17 selects the advanced level, and if the accuracy index is less than the threshold, the level selection unit 17 selects the beginner level.

The level selection unit 17 may also calculate the success rate and average accuracy index when the process is executed from a predetermined number of past canvas data (e.g., 10 cases) of the user obtained from the history DB 14, and select a level based on the calculation results.

The prompt generation unit 13 may also generate a prompt by referring to past requests and results obtained from the history DB 14. For example, the prompt generation unit 13 selects the canvas data with the highest evaluation value from among the canvas data that indicates the evaluation value of the accuracy of the final trained analytical model, which is the canvas data when a "linear model" was specified as the analytical model in the past. Then, the prompt generation unit 13 generates a prompt that encourages the user to generate a sentence that recommends the preprocessing specified in the selected canvas data.

Furthermore, the history DB 14 may store the data contents to be analyzed. For example, the request processing unit 12 stores the data to be analyzed in the history DB 14 together with the request and the results. The history DB 14 may store data in a specified format, such as CSV format or table format, as is, or may store a portion of the data, or may store statistics calculated from the data.

The prompt generation unit 13 can generate a prompt by referring to not only past requests and results acquired from the history DB 14, but also past data of the analysis target acquired from the history DB 14. This allows, for example, the prompt generation unit 13 to use the data of the analysis target as material for determining a preprocessing with a high degree of recommendation. For example, when the characteristics of the data of the analysis target acquired by the request processing unit 12 are similar to the characteristics of data of the analysis target that has previously been subjected to "outlier interpolation" as preprocessing, the prompt generation unit 13 generates a prompt so that a sentence recommending "outlier interpolation" as preprocessing is more likely to be generated. Note that the characteristics of the data may be statistics such as variance and average.

For example, the prompt generating unit 13 writes in element (2) of the prompt, as shown below, that the pre-processing specified in the selected canvas data is highly recommended.
(2) Context: Preprocessing not specified (however, "outlier interpolation" is recommended based on the history and data content of the analysis target)

In this case, for example, a sentence in which the sentence "The recommended preprocessing is outlier interpolation" is added to the sentence shown in box 5313 is generated by the language model group 15.

(Modification)
5 is a diagram showing a modification of the information providing system according to the first embodiment. An information providing device 10a of the information providing system 1a shown in FIG.

The level selection unit 17 selects a level using the level selection model 18. The level selection model 18 is a trained machine learning model that estimates and outputs a level as a classification or regression problem based on the input operation content and results.

[Processing flow of the first embodiment]
The flow of processing in the information providing device 10 will be described with reference to Fig. 6. Fig. 6 is a flowchart showing the flow of processing in the information providing system according to the first embodiment.

First, the request processing unit 12 accepts the operation content or inquiry content as a request via the interface unit 11 (step S101). Next, the request processing unit 12 stores the request content and the result in the history DB 14 (step S102).

Here, the prompt generation unit 13 retrieves past request contents and results from the history DB (step S103).

The level selection unit 17 selects a level based on the request content and the result (step S104). The level selection unit 17 may select a level based on user input.

Next, the prompt generation unit 13 generates a prompt based on the request content and result according to the selected level (step S105).

Then, the sentence generation unit 16 inputs the prompt into a language model (large-scale language model) included in the language model group 15 to generate a sentence (step S106). The sentence generation unit 16 outputs the generated sentence via the interface unit 11 (step S107).

[Effects of the First Embodiment]
As described above, the level selection unit 17 selects the user's level regarding data analysis. The prompt generation unit 13 generates a prompt for causing the language model to generate a sentence corresponding to the level, which is a sentence for supporting the use of the data analysis tool, based on a request received from a terminal device on which the data analysis tool is executed and the level selected by the level selection unit 17. The sentence generation unit 16 generates a sentence by inputting the prompt generated by the prompt generation unit 13 into the language model. The interface unit 11 causes the terminal device to display the sentence generated by the sentence generation unit 16.

This allows users to gain knowledge about how to use data analysis tools through texts that correspond to their own level of data analysis, thereby improving the ease of use of data analysis tools.

The sentence generation unit 16 generates a sentence by inputting a prompt into a language model corresponding to the level selected by the level selection unit 17 from among the language models corresponding to each of the multiple levels.

The level selection unit 17 further selects a role. The prompt generation unit 13 generates a prompt to cause the language model to generate sentences in a style appropriate for the role. This allows the style of the sentences provided to be changed according to the user's preferences.

The lower the level selected by the level selection unit 17, the more the prompt generation unit 13 generates a prompt to cause the language model to generate sentences that will strongly evoke emotions in the user. This can increase the motivation of novice users.

The level selection unit 17 selects a level based on the user's past operations in the data analysis tool and the accuracy of data analysis by a process generated based on the past operations. This makes it possible to select a level that is more suitable for the user.

[System configuration, etc.]
In addition, each component of each device shown in the figure is functionally conceptual, and does not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure, and all or a part of it can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc. Furthermore, each processing function performed by each device can be realized in whole or in any part by a CPU and a program analyzed and executed by the CPU, or can be realized as hardware by wired logic. Note that the program may be executed not only by the CPU but also by other processors such as a GPU.

Furthermore, among the processes described in the embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, control procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified.

[program]
In one embodiment, the information providing device 10 can be implemented by installing an information providing program that executes the above information processing as package software or online software on a desired computer. For example, the information providing program can be executed by an information processing device to function as the information providing device 10. The information processing device here includes desktop or notebook personal computers. In addition, the information processing device also includes mobile communication terminals such as tablet terminals, smartphones, mobile phones and PHS (Personal Handy-phone Systems), and slate terminals such as PDAs (Personal Digital Assistants).

The information providing device 10 can also be implemented as a server that provides the above-mentioned information processing services to a client, the terminal device 20 used by the user. For example, the server is implemented as a server device that provides an information processing service that takes a request as input and outputs text. In this case, the server may be implemented as a web server, or as a cloud that provides the above-mentioned information processing services by outsourcing.

Figure 7 is a diagram showing an example of the configuration of a computer that executes an information provision program. The computer 1000 has, for example, a memory 1010 and a CPU 1020. The computer 1000 also has a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. Each of these components is connected by a bus 1080.

The memory 1010 includes a read only memory (ROM) 1011 and a random access memory (RAM) 1012. The ROM 1011 stores a boot program such as a basic input output system (BIOS). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

The hard disk drive 1090 stores, for example, an OS (Operating System) 1091, an application program 1092, a program module 1093, and program data 1094. That is, the programs that define each process of the information providing device 10 are implemented as program modules 1093 in which computer-executable code is written. The program modules 1093 are stored, for example, in the hard disk drive 1090. For example, a program module 1093 for executing processes similar to the functional configuration of the information providing device 10 is stored in the hard disk drive 1090. The hard disk drive 1090 may be replaced by an SSD (Solid State Drive).

The setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in memory 1010 or hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or program data 1094 stored in memory 1010 or hard disk drive 1090 into RAM 1012 as necessary, and executes the processing of the above-described embodiment.

The program module 1093 and the program data 1094 may not necessarily be stored in the hard disk drive 1090, but may be stored in a removable storage medium, for example, and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and the program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or wide area network (WAN)). The program module 1093 and the program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

REFERENCE SIGNS LIST 1 Information providing system 10 Information providing device 11 Interface unit 12 Request processing unit 13 Prompt generating unit 14 History DB
15 Language model group 16 Sentence generation unit 53 Chat screen 511, 512, 513, 514 Card 521, 522 Area 532 Icon 533 Input field 5311, 5312, 5313, 5314 Box

Claims (6)

a level selection unit for selecting a level of data analysis by a user;
a prompt generating unit that generates a prompt for causing a language model to generate a sentence according to the level selected by the level selecting unit, the prompt being a request received from a terminal device on which a data analysis tool that generates a series of processes related to data analysis is executed, the request being an operation content that specifies an algorithm for analyzing data, and a sentence indicating an operation to be performed after the operation content, based on the level selected by the level selecting unit; and
a sentence generation unit that generates a sentence by inputting the prompt generated by the prompt generation unit to the language model;
a display control unit that displays the sentence generated by the sentence generation unit on the terminal device;
having
The information providing device is characterized in that the level selection unit selects a level based on the operation content previously specified by the user in the data analysis tool, the success or failure of data analysis processing by a process generated in accordance with the operation content, and an accuracy index . 2. The information providing device according to claim 1, wherein the sentence generating unit generates a sentence by inputting the prompt into a language model corresponding to a level selected by the level selecting unit, among language models corresponding to each of a plurality of levels. The level selection unit further selects a role;
The information providing device according to claim 1 , wherein the prompt generating unit generates a prompt for causing the language model to generate a sentence in a writing style corresponding to the role. 2. The information providing device according to claim 1, wherein the prompt generating unit generates a prompt for causing the language model to generate a sentence that more strongly stirs the user's emotions as the level selected by the level selecting unit becomes lower. 1. A computer-implemented method for providing information, comprising:
a level selection step of selecting a level for the user's data analysis;
a prompt generating step of generating a prompt for causing a language model to generate a sentence according to the level selected by the level selection step, the prompt being a request received from a terminal device on which a data analysis tool for generating a series of processes related to data analysis is executed, the request being an operation content specifying an algorithm for analyzing data, and a sentence indicating an operation to be performed after the operation content , based on the level selected by the level selection step;
a sentence generating step of generating a sentence by inputting the prompt generated by the prompt generating step into the language model;
a display control step of displaying the sentence generated by the sentence generating step on the terminal device;
Including,
The level selection step selects a level based on the operation content previously specified by the user in the data analysis tool, whether or not a data analysis process is executed by a process generated according to the operation content, and an accuracy index.
13. An information providing method comprising: A level selection step of selecting a level for the user's data analysis;
a prompt generating step of generating a prompt for causing a language model to generate a sentence according to the level selected by the level selection step, the prompt being a request received from a terminal device on which a data analysis tool for generating a series of processes related to data analysis is executed, the request being an operation content specifying an algorithm for analyzing data, and a sentence indicating an operation to be performed after the operation content , based on the level selected by the level selection step;
a sentence generation step of generating a sentence by inputting the prompt generated by the prompt generation step into the language model;
a display control step of displaying the sentence generated by the sentence generating step on the terminal device;
on the computer ,
The level selection step selects a level based on the operation content previously specified by the user in the data analysis tool, whether or not data analysis processing is executed by a process generated according to the operation content, and an accuracy index.
23. An information providing program comprising:

Images