Information processing apparatus, information processing method, and program

Abstract

To allow a person who does not necessarily have sufficient knowledge or experience in development and operation of a machine learning model to appropriately analyze a factor contributing to the decrease in prediction accuracy of the machine learning model.SOLUTION: An information processing apparatus includes: prompt acquisition means which acquires a prompt input by a user; algorithm acquisition means which acquires an analytical algorithm designated in the prompt; and analysis execution means that outputs a question based on the analytical algorithm, executes the analytical algorithm interactively by acquiring an answer of the user to the question, and outputs the analysis result.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] This disclosure relates to the analysis of machine learning models.

Background Art

[0002] MLOps (Machine Learning Operations) is known as a technology that enables continuous and low-cost operation of machine learning models. MLOps is a technology for developing, deploying, and operating machine learning models. In particular, in the actual operation of a system using a machine learning model, it is important to analyze the factors causing a decrease in the prediction accuracy of the machine learning model.

[0003] In order to analyze the factors causing a decrease in prediction accuracy in an actually operating system, sufficient knowledge and experience regarding programming in the development of a machine learning model and maintenance during the operation of the machine learning model are required. Patent Document 1 discloses a method for converting natural language into a programming language.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] One object of the present invention is to enable appropriate analysis of the factors causing a decrease in the prediction accuracy of a machine learning model even for those who do not necessarily have sufficient knowledge and experience regarding the development and operation of the machine learning model.

Means for Solving the Problems

[0006] In one aspect of this disclosure, an information processing apparatus prompt acquisition means for acquiring a prompt input by a user, algorithm acquisition means for acquiring the analysis algorithm specified by the prompt; analysis execution means for interactively executing the analysis algorithm by outputting a question based on the analysis algorithm and acquiring the user's answer to the question, and outputting an analysis result, comprising:

[0007] In another aspect of the present disclosure, an information processing method is executed by a computer, acquiring a prompt input by a user, acquiring the analysis algorithm specified by the prompt, interactively executing the analysis algorithm by outputting a question based on the analysis algorithm and acquiring the user's answer to the question, and outputting an analysis result.

[0008] In still another aspect of the present disclosure, a program causes a computer to execute a process of: acquiring a prompt input by a user, acquiring the analysis algorithm specified by the prompt, interactively executing the analysis algorithm by outputting a question based on the analysis algorithm and acquiring the user's answer to the question, and outputting an analysis result.

Advantages of the Invention

[0009] According to the present disclosure, even a person who does not necessarily have sufficient knowledge and experience regarding the development and operation of a machine learning model can appropriately analyze factors causing a decrease in prediction accuracy by the machine learning model.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Embodiments for Carrying Out the Invention

[0011] Hereinafter, preferred embodiments of the present disclosure will be described with reference to the drawings. <First Embodiment> [Overall Configuration] FIG. 1 shows the overall configuration of an analysis system to which the information processing apparatus according to the present disclosure is applied. When a decrease in prediction accuracy (hereinafter, also referred to as "prediction error") occurs during the operation of prediction processing using a machine learning model, the analysis system 1 analyzes the cause. As shown in the figure, the analysis system 1 includes an information processing apparatus 10 and a terminal apparatus 20. The information processing apparatus 10 and the terminal apparatus 20 can communicate with each other by wire or wirelessly.

[0012] The terminal apparatus 20 is operated by a user who manages and maintains a machine learning model (hereinafter, also referred to as a "prediction model") for performing a predetermined prediction. The predetermined prediction can be various predictions such as, for example, weather prediction, power demand prediction, and sales prediction in a store. In the present embodiment, the user only needs to have certain knowledge and experience regarding the development and maintenance of the machine learning model, and does not have to be an expert with a lot of knowledge and experience.

[0013] When a prediction error occurs in the prediction model during operation, the user analyzes the cause of the prediction error. Specifically, when a prediction error occurs, the user operates the terminal device 20 and sends a prompt instructing the cause analysis of the prediction error to the information processing device 10.

[0014] The information processing device 10 interprets the prompt input using the natural language model and performs an analysis of the cause of the prediction error. Note that a "prompt" refers to an instruction to a generative AI (Artificial Intelligence) including a natural language model. The information processing device 10 interactively performs an analysis of the cause of the prediction error using a pre-prepared analysis algorithm. Specifically, the information processing device 10 asks the user questions about variables and the prediction model according to the analysis algorithm and obtains the user's answers. Then, the information processing device 10 analyzes the cause of the prediction error based on the user's answers, generates an analysis result, and outputs it to the terminal device 20. In this way, the user can analyze the cause of the prediction error by inputting in natural language to the information processing device 10.

[0015] [Natural Language Model] Here, the natural language model will be explained. The natural language model is one that has learned the relationships between words in a sentence and is a model that generates related strings related to the target string from the target string. By using a natural language model that has learned sentences and texts in various contexts, it is possible to generate related strings with appropriate content related to the target string. For example, the case of using a natural language model in question-and-answer will be explained. In this case, the natural language model receives an input of a question "What kind of country is Japan?" as the target string and generates strings such as "Japan is an island country in the Northern Hemisphere..." as an answer to the question.

[0016] The learning method of the natural language model is not particularly limited. As an example, it may be one that is learned to output at least one sentence including the input string. For a specific example, the natural language model can be GPT (Generative Pre-Training) that outputs a sentence including the input string by predicting a highly probable string following the input string. In addition to this, as the natural language model, for example, T5 (Text-to-Text Transfer Transformer), BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately), etc. can also be used. In the present embodiment, a large language model may be used as the natural language model. Also, the natural language model may be one that can access the Internet or other specialized knowledge bases to obtain information.

[0017] [Hardware Configuration] (Information Processing Apparatus) FIG. 2 is a block diagram showing the hardware configuration of the information processing apparatus 10. As shown in the figure, the information processing apparatus 10 includes a processor 11, an interface (IF) 12, a ROM (Read Only Memory) 13, a RAM (Random Access Memory) 14, a database (DB) 15, and a recording medium 16. Each component is connected to each other through, for example, a bus 18.

[0018] The processor 11 is a computer such as a CPU (Central Processing Unit), and controls the entire information processing apparatus 10 by executing a pre-prepared program. Specifically, as the processor 11, a CPU, GPU (Graphics Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Point number Processing Unit), PPU (Physics Processing Unit), TPU (TensorProcessingUnit), quantum processor, microcontroller, or a combination thereof can be used.

[0019] Also, the processor 11 loads the program stored in the ROM 13 or the recording medium 16 into the RAM 14 and executes each process coded in the program. The processor 11 functions as part or all of the information processing apparatus 10. The processor 11 executes a factor analysis process described later.

[0020] The IF 12 transmits and receives data to and from an external device. Specifically, the information processing apparatus 10 receives, through the IF 12, a prompt for instructing factor analysis and an answer to a question output to the user from the terminal device 20.

[0021] The ROM 13 stores various programs executed by the processor 11. The RAM 14 is used as a work memory during the execution of various processes by the processor 11.

[0022] The DB 15 stores a plurality of analysis algorithms used for factor analysis.

[0023] The recording medium 16 is a non-volatile and non-temporary recording medium such as a disk-shaped recording medium or a semiconductor memory. The recording medium 16 may be configured to be detachable from the information processing apparatus 10. The recording medium 16 stores various programs executed by the processor 11.

[0024] In addition to the above, the information processing apparatus 10 may include a display device such as a liquid crystal display and an input device such as a keyboard or a mouse. These display device and input device are used, for example, by the administrator of the information processing apparatus 10 or the like.

[0025] (Terminal device) FIG. 3 is a block diagram showing the hardware configuration of the terminal device 20. The terminal device 20 is, for example, a PC or a tablet terminal. As shown in the figure, the terminal device 20 includes a processor 21, an IF 22, a ROM 23, a RAM 24, an input unit 26, and a display unit 27. Each component is connected to each other, for example, through a bus 28.

[0026] The processor 21 is a computer such as a CPU, and controls the entire terminal device 20 by executing a program prepared in advance. Note that the processor 21 may be a GPU, an FPGA, a DSP, an ASIC, or the like.

[0027] The IF 22 transmits and receives data to and from an external device. Specifically, the terminal device 20 transmits, through the IF 22, the prompts created by the user and the user's answers to questions to the information processing apparatus 10. Further, the terminal device 20 receives, through the IF 22, questions and analysis results from the information processing apparatus 10 to the user.

[0028] The ROM 23 stores various programs executed by the processor 21. The RAM 24 is used as a working memory during the execution of various processes by the processor 21.

[0029] The recording medium 25 is a non-volatile and non-transitory recording medium such as a disk-shaped recording medium or a semiconductor memory. The recording medium 25 may be configured to be detachable from the terminal device 20. The recording medium 25 stores various programs executed by the processor 21.

[0030] The input unit 26 is an input device such as a keyboard, a mouse, or a touch panel. The display unit 27 is a display or the like that performs display based on the control of the processor 21.

[0031] [Functional Configuration] FIG. 4 is a block diagram showing the functional configuration of the information processing apparatus 10 according to the first embodiment. Functionally, the information processing apparatus 10 includes a communication unit 111, an analysis unit 112, and an analysis algorithm DB 113.

[0032] The communication unit 111 is constituted by the IF 12 shown in FIG. 2, receives prompts and answers to questions from the terminal device 20, and transmits questions and analysis results to the terminal device 20. The communication unit 111 outputs the prompts and answers received from the terminal device 20 to the analysis unit 112.

[0033] The analysis algorithm DB 113 is constituted by the DB 15 shown in FIG. 2 and stores a plurality of analysis algorithms prepared in advance. Each analysis algorithm is an algorithm for analyzing the cause of prediction errors occurring in the prediction model during operation, and is prepared for each prediction model, for example. For the same prediction model, a plurality of analysis algorithms for performing analysis from different viewpoints may be prepared. Although details will be described later, each analysis algorithm is described in natural language.

[0034] The analysis unit 112 is constituted by the processor 11 shown in FIG. 2 and analyzes the cause of prediction errors using the analysis algorithm. The analysis unit 112 is constituted using a natural language model that can interpret prompts described in natural language.

[0035] When analyzing the causes of prediction errors, the analysis unit 112 acquires the prompts created by the user operating the terminal device 20. Although details will be described later, the prompts include information specifying the analysis algorithm to be used for the analysis and an instruction to perform a cause analysis of prediction errors. The analysis unit 112 acquires the specified analysis algorithm from the analysis algorithm DB 113 according to the prompt. The analysis algorithm is described in natural language, and the analysis unit 112 interprets and executes the analysis algorithm using a natural language model.

[0036] During the execution of the analysis algorithm, the analysis unit 112 asks the user necessary questions. Specifically, the analysis unit 112 generates questions corresponding to the conditional branches included in the analysis algorithm and transmits them to the terminal device 20 via the communication unit 111. The user operates the terminal device 20 to send an answer to the question, and the analysis unit 112 acquires the answer via the communication unit 111. Then, the analysis unit 112 makes a determination in the conditional branch based on the obtained answer and continues the execution of the analysis algorithm. In this way, the analysis unit 112 repeats asking questions to the user and acquiring answers for each conditional branch included in the analysis algorithm, and executes the analysis algorithm interactively. When an analysis result is obtained according to the analysis algorithm, the analysis unit 112 transmits the analysis result to the terminal device 20. Note that the analysis unit 112 is an example of a prompt acquisition means, an algorithm acquisition means, and an analysis execution means.

[0037] As described above, in this embodiment, the information processing apparatus 10 executes an analysis process interactively by repeating questions and answers in natural language according to a pre-prepared analysis algorithm. Therefore, a user who analyzes prediction errors can proceed with the analysis along the analysis algorithm simply by answering the questions transmitted from the information processing apparatus 10. For this reason, if an appropriate analysis algorithm is prepared, even a user without advanced knowledge or experience regarding the analysis of prediction errors can perform an appropriate analysis.

[0038] [Analysis Example] Next, an example of the cause analysis of prediction errors by the analysis unit 112 will be described. FIG. 5 is a flowchart showing an example of an analysis algorithm (hereinafter referred to as "analysis algorithm AG"). In this example, the analysis algorithm AG analyzes the cause of prediction errors (hereinafter referred to as "prediction error cause") from the perspective of whether the cause of the prediction error lies in the explanatory variable X, the target variable Y, or the prediction model M itself used by the prediction model M. Then, as an analysis result, the analysis algorithm AG outputs the location where the prediction error cause exists (hereinafter referred to as "cause location"), the cause, and the countermeasure.

[0039] First, the analysis algorithm AG analyzes whether the explanatory variable X is a cause of prediction errors. Specifically, when the explanatory variable X satisfies condition A1 (step S21: Yes), the analysis algorithm AG determines that the explanatory variable X is not a cause of prediction errors (step S22). Also, when the explanatory variable X does not satisfy condition A1 but satisfies condition A2 (step S23: Yes), the analysis algorithm AG determines that the explanatory variable X is not a cause of prediction errors (step S22). On the other hand, when the explanatory variable does not satisfy either condition A1 or condition A2 (step S23: No), the analysis algorithm AG determines that the explanatory variable X is a cause of prediction errors and outputs the analysis result "cause location: explanatory variable X, cause: B1, countermeasure: C1" (step S24).

[0040] When the explanatory variable X is not a cause of prediction errors, the analysis algorithm AG analyzes whether the target variable Y is a cause of prediction errors. Specifically, when the target variable Y satisfies condition A3 (step S25: Yes), the analysis algorithm AG determines that the target variable Y is not a cause of prediction errors (step S26). Also, when the target variable Y does not satisfy condition A3 but satisfies condition A4 (step S27: Yes), the analysis algorithm AG determines that the target variable Y is not a cause of prediction errors (step S26). On the other hand, when the target variable Y does not satisfy either condition A3 or condition A4 (step S27: No), the analysis algorithm AG determines that the target variable Y is a cause of prediction errors and outputs the analysis result "cause location: target variable Y, cause: B2, countermeasure: C2" (step S28).

[0041] When the target variable Y is not a factor, the analysis algorithm AG analyzes whether the prediction model M is a factor for prediction errors. Specifically, when the prediction model M does not satisfy condition A5 (step S29: No), the analysis algorithm AG determines that the prediction model M is a factor for prediction errors and outputs the analysis result "Factor location: Model M, Factor: B3, Countermeasure: C3" (step S30). On the other hand, when the prediction model M satisfies condition A5 (step S29: Yes), the analysis algorithm AG determines that the cause of the prediction error is unknown and outputs the analysis result of "unknown cause" (step S31).

[0042] Here, for the sake of convenience of explanation, the content of the analysis algorithm AG has been explained by the flowchart in FIG. 5. However, in reality, the analysis algorithm AG may be described in a programming language, may be described by IF-THEN rules, or may be described in other forms. And based on the analysis algorithm AG described in any form, an analysis algorithm AG described in natural language is generated and stored in the analysis algorithm DB113. FIG. 6 shows an example of the analysis algorithm AG described in natural language.

[0043] In FIG. 6, in the item "#Overview", it is described that the analysis algorithm AG performs a factor analysis of prediction errors in the order of explanatory variables, target variables, and prediction models, and outputs the factor location, factor, and countermeasure as analysis results.

[0044] In the item "#Explanatory variable analysis", the content of steps S21 to S24 in FIG. 5 is described as the analysis of explanatory variables. In the item "#Target variable analysis", the content of steps S25 to S28 in FIG. 5 is described as the analysis of target variables. Also, in the item "#Model analysis", the content of steps S29 to S31 in FIG. 5 is described as the analysis of prediction models.

[0045] The user generates a prompt including the specification of the analysis algorithm AG and an instruction to perform cause analysis, and inputs it to the information processing apparatus 10. FIG. 7 shows an example of the prompt input by the user and the message output by the information processing apparatus 10. First, the prompt 41 created by the user is input to the information processing apparatus 10. The prompt 41 includes the specification of the analysis algorithm AG to be used in the analysis and an instruction to perform cause analysis of prediction errors. Further, the prompt 41 instructs to output one question for each conditional branch of the analysis algorithm AG, and to output the cause location, cause, and countermeasure as the analysis result.

[0046] The analysis unit 112 of the information processing apparatus 10 interprets the prompt 41 using a natural language model and acquires the specified analysis algorithm AG from the analysis algorithm DB 113. The analysis algorithm AG is described in natural language as shown in FIG. 6, and the analysis unit 112 interprets and executes the analysis algorithm AG using a natural language model. Thereby, the analysis unit 112 executes the analysis according to the procedure of the analysis algorithm shown in FIG. 5.

[0047] Specifically, based on the conditional branch of step S21 in FIG. 5, the analysis unit 112 generates a question 42 asking whether the explanatory variable X satisfies the condition A1, and transmits it to the terminal device 20. The user transmits an answer (prompt) 43 to the question to the information processing apparatus 10. Next, based on the answer 43, the analysis unit 112 determines that the explanatory variable X is not a cause of prediction error as shown in step S22. Next, based on the conditional branch of step S25, the analysis unit 112 generates a question 44 asking whether the target variable Y satisfies the condition A3, and transmits it to the terminal device 20.

[0048] In this way, the analysis unit 112 asks the user questions according to the analysis algorithm AG and proceeds with the analysis based on the answers to the questions. When receiving the answer 45 to the question based on the conditional branch of step S29, the analysis unit 112 outputs the analysis result 46 according to step S30 and ends the analysis.

[0049] In this way, in the present embodiment, the user can perform factor analysis of prediction errors by interacting with the information processing apparatus 10 using natural language. Therefore, even a user who does not have advanced knowledge or experience regarding the machine learning model can appropriately perform factor analysis.

[0050] [Analysis Process] Next, the flow of the analysis process for performing the above analysis will be described. FIG. 8 is a flowchart of the analysis process. This process is realized by the processor 11 shown in FIG. 2 executing a pre-prepared program and operating as an element shown in FIG. 3.

[0051] First, the information processing apparatus 10 reads a specified analysis algorithm from the analysis algorithm DB 113 in accordance with the prompt generated by the user (step S51). Next, the information processing apparatus 10 executes the analysis algorithm (step S52). If there is a conditional branch in the analysis algorithm (step S53: Yes), the information processing apparatus 10 generates a question corresponding to the condition and transmits it to the terminal device 20, and continues the analysis based on the user's answer to the question (step S54). In this way, the information processing apparatus 10 asks the user questions for each conditional branch and continues the analysis. Then, when the analysis is completed according to the analysis algorithm (step S55: Yes), the analysis result is output to the terminal device 20 (step S56). Then, the process ends.

[0052] [Modification Example] Next, a modification example of the above embodiment will be described. The following modification examples can be applied in appropriate combinations. [Modification Example 1] In the above analysis example, although the analysis unit 112 asks the user about all the conditional branches in the analysis algorithm, if information regarding a specific conditional branch has already been obtained, the question regarding that conditional branch can be omitted. For example, if the description "The explanatory variable X satisfies the conditions A1 and A2." is included in the prompt created by the user at the start of the analysis, the analysis unit 112 can omit the questions corresponding to steps S21 and S23 in FIG. 5 and only ask the question corresponding to step S25. Similarly, if the description "Assume that the explanatory variable X is not a factor for prediction errors." is included in the prompt created by the user at the start of the analysis, the analysis unit 112 can also omit the questions corresponding to steps S21 and S23 in FIG. 5 and only ask the question corresponding to step S25.

[0053] Also, when the user re-performs the analysis by changing the answer to a question or the like, the analysis unit 112 can omit the questions regarding the parts that utilize the past answers of the user. For example, if the description "The answer regarding the explanatory variable X is the same as in the previous analysis. Please re-run the analysis." is included in the prompt created by the user at the start of the analysis, the analysis unit 112 can perform the analysis by diverting the previous answer of the user for steps S21 and S23 in FIG. 5.

[0054] (Modification Example 2) The analysis unit 112 may change the expression of the question according to the level of the user's knowledge and experience (hereinafter simply referred to as "knowledge level"). For example, the analysis unit 112 may change the terms and phrases used in the question for a data scientist with less than 3 years of experience and a data scientist with 3 or more years of experience. When the knowledge level of the user is low, the analysis unit 112 may generate questions by reducing the use of technical terms in the questions for the user, using the easier term when there are multiple terms with the same meaning, adding explanations for technical terms, and the like.

[0055] At this time, when the information processing apparatus 10 has information regarding the user's knowledge level, the expression of the question may be changed based on it. Also, when the user designates their knowledge level in the prompt, the expression of the question may be changed accordingly. For example, when the user describes in the prompt "Please ask questions at the knowledge level of a data scientist with less than 3 years of experience.", the information processing apparatus 10 may generate a question corresponding to that knowledge level.

[0056] In addition, when the user responds to the question output by the information processing apparatus 10 with "I don't understand the meaning of the question.", "I don't understand the meaning of the term XX in the question.", etc., the information processing apparatus 10 may regenerate a question with a different expression or regenerate the question using different terms.

[0057] (Modification Example 3) In the above analysis example, the information processing apparatus 10 outputs a question to the user corresponding to the conditional branch in the analysis algorithm, but the questions to the user are not limited to those related to the conditional branch. For example, it may be a question for confirming the type of the analysis algorithm or the preconditions when executing the analysis algorithm.

[0058] (Modification Example 4) In the above embodiment, the user designates the analysis algorithm in the prompt, and the information processing apparatus 10 acquires the designated analysis algorithm from the analysis algorithm DB 113. Instead, the user may include the analysis algorithm described in natural language in the prompt and give an analysis instruction to the information processing apparatus 10. For example, including the description of the analysis algorithm shown in FIG. 6 at the end of the prompt 41 shown in FIG. 7, and setting the first sentence of the prompt 41 as "Please perform an analysis of the prediction error factors using the following analysis algorithm.", etc., to instruct the use of the analysis algorithm in the prompt.

[0059] (Modification Example 5) In the above embodiment, the analysis algorithm described in natural language is stored in the analysis algorithm DB113 shown in FIG. 4, and the analysis unit 112 interprets and executes it using the language model. Instead, when the analysis unit 112 is provided with a language model that can interpret not only natural language but also programming language, the analysis algorithm described in the programming language may be stored in the analysis algorithm DB113, and the analysis unit 112 may interpret and execute it.

[0060] <Second Embodiment> FIG. 9 is a block diagram showing the functional configuration of the information processing apparatus according to the second embodiment. The information processing apparatus 70 includes a prompt acquisition unit 71, an algorithm acquisition unit 72, and an analysis execution unit 73.

[0061] FIG. 10 is a flowchart of the process by the information processing apparatus according to the second embodiment. The prompt acquisition unit 71 acquires the prompt input by the user (step S71). The algorithm acquisition unit 72 acquires the analysis algorithm specified by the prompt (step S72). The analysis execution unit 73 outputs a question based on the analysis algorithm, and acquires the user's answer to the question, thereby interactively executing the analysis algorithm and outputting the analysis result (step S73).

[0062] According to the information processing apparatus 70 of the second embodiment, even a person who does not necessarily have sufficient knowledge and experience regarding the development and operation of the machine learning model can appropriately analyze the factors causing a decrease in the prediction accuracy by the machine learning model.

[0063] Some or all of the above embodiments may be described as follows in the following supplementary notes, but are not limited thereto.

[0064] (Supplementary Note 1) A prompt acquisition unit that acquires a prompt input by a user, An algorithm acquisition unit that acquires the analysis algorithm specified by the prompt, Output questions based on the analysis algorithm, and obtain the user's answers to the questions, thereby interactively executing the analysis algorithm and outputting analysis results, and analysis execution means; An information processing apparatus comprising:

[0065] (Appendix 2) The analysis algorithm is described in natural language, The analysis execution means outputs the question using a language model, and interprets and executes the analysis algorithm using the language model. The information processing apparatus according to Appendix 1.

[0066] (Appendix 3) The analysis execution means outputs a question regarding the conditional branch for each conditional branch included in the analysis algorithm. The information processing apparatus according to Appendix 1.

[0067] (Appendix 4) The prompt includes conditional branch related information indicating whether the conditions included in the analysis algorithm are satisfied, The analysis execution means outputs the question for conditional branches other than the conditional branch corresponding to the conditional branch related information among the conditional branches included in the analysis algorithm. The information processing apparatus according to Appendix 3.

[0068] (Appendix 5) The prompt includes information indicating the user's knowledge level, The question output means outputs a question using phrases corresponding to the user's knowledge level. The information processing apparatus according to Appendix 1.

[0069] (Appendix 6) The prompt includes a description indicating the analysis algorithm, The algorithm acquisition means acquires the analysis algorithm included in the prompt. The information processing apparatus according to Appendix 1.

[0070] (Appendix 7) The analysis algorithm is described in a programming language, The analysis execution means is the information processing apparatus according to Appendix 1 that interprets and executes the analysis algorithm using a language model capable of interpreting a programming language.

[0071] (Appendix 8) Executed by a computer, obtains a prompt input by a user, obtains the analysis algorithm specified by the prompt, outputs a question based on the analysis algorithm, and obtains the user's answer to the question, thereby interactively executing the analysis algorithm and outputting an analysis result.

[0072] (Appendix 9) obtains a prompt input by a user, obtains the analysis algorithm specified by the prompt, outputs a question based on the analysis algorithm, and obtains the user's answer to the question, thereby interactively executing the analysis algorithm and outputting an analysis result, and causes a computer to execute the process.

[0073] Although the present disclosure has been described with reference to the embodiments and examples above, the present disclosure is not limited to the above embodiments and examples. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure.

Explanation of Reference Numerals

[0074] 10 Information processing apparatus 111 Communication unit 112 Analysis unit 113 Analysis algorithm DB

Claims

1. A prompt acquisition means for acquiring a prompt input by a user, An algorithm acquisition means for acquiring an analysis algorithm specified by the prompt, An analysis execution means for outputting a question based on the analysis algorithm, and acquiring the user's answer to the question, thereby executing the analysis algorithm interactively and outputting an analysis result, An information processing apparatus comprising the above.

2. The analysis algorithm is described in natural language, The analysis execution means outputs the question using a language model, and interprets and executes the analysis algorithm using the language model. The information processing apparatus according to claim 1.

3. The analysis execution means outputs a question regarding each conditional branch included in the analysis algorithm. The information processing apparatus according to claim 1.

4. The prompt includes conditional branch related information indicating whether a condition included in the analysis algorithm is satisfied, The analysis execution means outputs the question for conditional branches other than the conditional branch corresponding to the conditional branch related information among the conditional branches included in the analysis algorithm. The information processing apparatus according to claim 3.

5. The prompt includes information indicating the knowledge level of the user, The analysis execution means outputs a question using a phrase according to the knowledge level of the user. The information processing apparatus according to claim 1.

6. The prompt includes a description indicating the analysis algorithm, The algorithm acquisition means acquires the analysis algorithm included in the prompt. The information processing apparatus according to claim 1.

7. The analysis algorithm is described in a programming language, The analysis execution means interprets and executes the analysis algorithm using a language model that can interpret the programming language. The information processing apparatus according to claim 1.

8. Executed by a computer, Acquire a prompt input by a user, Acquire the analysis algorithm specified by the prompt, An information processing method for outputting a question based on the analysis algorithm, and acquiring the user's answer to the question, thereby executing the analysis algorithm interactively and outputting an analysis result.

9. Acquire a prompt input by a user, Acquire the analysis algorithm specified by the prompt, A program that causes a computer to execute a process of interactively executing the analysis algorithm and outputting an analysis result by outputting a question based on the analysis algorithm and obtaining the user's answer to the question.

Images