Information processing device, analysis support method, and analysis support program

Abstract

To facilitate analysis of various subjects. [Solution] The information processing device comprises a reception unit that receives a designation of an analysis target, a prompt generation unit that generates prompts instructing the user to provide the information necessary to generate a report showing the analysis results of the analysis target, a generation control unit that inputs the generated prompts into a generation model to generate a response, and a presentation control unit that presents the report generated based on the response. This information processing device can also support decision-making based on analysis results.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus, an analysis support method, and an analysis support program.

Background Art

[0002] A tool called a digital dashboard (hereinafter simply referred to as a dashboard) that visualizes various types of information related to an analysis target is known. For example, the following Patent Document 1 describes an infrastructure maintenance management support system having a function of graphically visualizing the status of infrastructure and various maintenance-related information as a dashboard.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Although a dashboard is a useful tool for grasping the status and the like of an analysis target, there is a problem that displaying the dashboard requires the effort of collecting various types of information related to the status of the analysis target. Such a problem is not limited to dashboards, and is a problem that occurs with reports of any format that indicate analysis results of an analysis target.

[0005] The present disclosure has been made in view of the above problems, and an exemplary object thereof is to provide a technology that facilitates analysis of various analysis targets.

Means for Solving the Problems

[0006] An information processing device relating to an exemplary aspect of this disclosure includes: a receiving means for receiving the designation of an object to be analyzed; a prompt generating means for generating a prompt that instructs the user to provide the information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control means for inputting the prompt into a machine learning-trained generation model to generate a response to the input prompt, thereby causing the model to generate a response; and a presentation control means for presenting a report showing the analysis results of the object to be analyzed, which is generated based on the response.

[0007] An example of an analysis support method relating to this disclosure includes, at least one processor: an acceptance process for receiving a designation of an analysis target; a prompt generation process for generating a prompt that instructs the system to provide information necessary to generate a report showing the analysis results of the analysis target; a generation control process for inputting the prompt into a machine learning-trained generation model to generate a response to the input prompt and causing the model to generate a response; and a presentation control process for presenting a report showing the analysis results of the analysis target, generated based on the response.

[0008] An example of an analysis support program relating to this disclosure causes a computer to function as: a receiving means for receiving the designation of an object to be analyzed; a prompt generating means for generating prompts that instruct the computer to provide the information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control means for inputting the prompts into a machine learning-trained generation model to generate answers to the input prompts and causing it to generate answers; and a presentation control means for presenting a report showing the analysis results of the object to be analyzed, which is generated based on the answers. [Effects of the Invention]

[0009] One illustrative aspect of this disclosure demonstrates the effect of providing a technology that facilitates analysis of various analytes. [Brief explanation of the drawing]

[0010] [Figure 1]This is a block diagram showing the configuration of the information processing device related to this disclosure. [Figure 2] This flowchart shows the flow of the analysis support method related to this disclosure. [Figure 3] This is a block diagram showing the configuration of other information processing devices related to this disclosure. [Figure 4] This figure shows an example of a report presentation by the information processing device shown in Figure 3. [Figure 5] This figure shows an example of a display screen that lists reports corresponding to multiple analysis targets. [Figure 6] This figure shows an example of the report display screen. [Figure 7] This figure shows an example of how related current events information is displayed. [Figure 8] Figure 3 is a flowchart showing the processing flow executed by the information processing device. [Figure 9] This is a flowchart illustrating the process for monitoring the subject of analysis. [Figure 10] This is a block diagram showing the configuration of a computer that functions as an information processing device related to this disclosure. [Modes for carrying out the invention]

[0011] The following are examples of embodiments of the present invention. However, the present invention is not limited to the exemplary embodiments shown below, and various modifications are possible within the scope of the claims. For example, embodiments obtained by appropriately combining some or all of the technologies (things or methods) employed in each of the exemplary embodiments shown below may also be included in the scope of the present invention. Furthermore, embodiments obtained by appropriately omitting some of the technologies employed in each of the exemplary embodiments shown below may also be included in the scope of the present invention. In addition, the effects mentioned in each of the exemplary embodiments shown below are examples of effects that can be expected in that exemplary embodiment and do not define the scope of the present invention. That is, embodiments that do not produce the effects mentioned in each of the exemplary embodiments shown below may also be included in the scope of the present invention.

[0012] (First Exemplary Embodiment) A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is a basic form of each of the exemplary embodiments described later. Note that the applicable scope of each technique adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technique adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in the present disclosure within a range that does not cause any particular technical hindrance. In addition, each technique shown in the drawings referred to for describing this exemplary embodiment can also be adopted in other exemplary embodiments included in the present disclosure within a range that does not cause any particular technical hindrance.

[0013] (Configuration of Information Processing Apparatus 1) The configuration of the information processing apparatus 1 according to this exemplary embodiment will be described with reference to Figure 1. Figure 1 is a block diagram showing the configuration of the information processing apparatus 1. As shown in Figure 1, the information processing apparatus 1 includes an accepting unit 101, a prompt generation unit 102, a generation control unit 103, and a presentation control unit 104.

[0014] The accepting unit 101 accepts a designation of an analysis target. What is set as the "analysis target" is arbitrary. For example, a product, a service, a person, an organization, a region, an industry, a field, or the like can be set as the analysis target. The accepting unit 101 may accept, for example, a direct designation such as the name of a product or the like, or may accept an indirect designation such as a travel destination popular among people in their 20s, a healthcare-related company, or the like.

[0015] Note that the "analysis" referred to herein includes, in addition to the narrow-sense "analysis" of examining the analysis target by dividing it into its constituent elements, "information collection" for collecting information related to the analysis target, "investigation" for examining the analysis target without dividing it into its constituent elements, and the like. Therefore, "analysis" in the present specification can also be read as "information collection" or "investigation".

[0016] The prompt generation unit 102 generates a prompt that instructs the system to provide the information necessary to generate a report showing the analysis results of a specified target of analysis. Here, "prompt" is an instruction sentence that is input to the generation model described later. Therefore, "prompt" can be read as "instruction sentence". For example, the prompt generation unit 102 generates a natural language sentence as the prompt that includes information indicating the specified target of analysis and instructs the system to output the information necessary to generate a report showing the analysis results of the target of analysis. Such a prompt can be generated, for example, by incorporating information indicating the target of analysis into a predetermined template.

[0017] The generation control unit 103 inputs the prompt generated by the prompt generation unit 102 to a generative model that has been machine-learned to generate a response to the input prompt, and causes the generative model to generate a response. The "generative model" may be any model that has been machine-learned to be capable of outputting a response corresponding to an input prompt. Further, which generative model to use may be determined according to, for example, the data format of the response to be generated. For example, when both the input prompt and the response to be output are natural language text data, a language model trained on natural language may be applied as the above generative model. Here, training on natural language means, in more detail, learning the sequence of constituent elements (such as words) in a natural language sentence, and the sequence of sentences in a text. Examples of such a language model include BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately), and the like. Further, for example, it is also possible to cause the generative model to generate images such as graphs that are constituent elements of a report. In this case, a generative model that has been machine-learned to generate an image corresponding to the content of an input prompt may be used. Further, the generative model may be a fine-tuned version of a general-purpose generative model, so that information necessary for generating a report indicating the analysis result of a specified analysis target can be generated with high accuracy. In fine-tuning, data composed of pairs of a prompt and a response to be generated for that prompt may be used as training data.

[0018] The above generative model may be stored in the information processing apparatus 1, or may be stored in another apparatus. In the latter case, the generation control unit 103 may transmit the prompt to the other apparatus to cause it to generate a response, and acquire the generated response from the other apparatus.

[0019] The presentation control unit 104 presents a report showing the analysis results of the subject being analyzed, which is generated based on the responses generated by the generation model in accordance with the control of the generation control unit 103. Here, "presentation" means making the content of the presentation recognizable to the recipient. For example, the presentation control unit 104 can present the report by displaying it on a display device. Alternatively, the presentation control unit 104 can present the report by making it viewable by the recipient (for example, by uploading it to a server accessible to the recipient). Note that the report "generated based on responses" includes reports that use the responses generated by the generation model as they are, as well as reports generated by extracting parts of those responses, processing them as necessary, and then arranging them. When generating a report using responses generated by the generation model, the report generation may be performed by the information processing device 1 or by another device.

[0020] The recipients of the report are at the discretion of the system. For example, the report may be presented to individuals who have been designated as subjects of analysis, or it may be presented to pre-registered recipients. The method of presenting the report is also at the discretion of the system. For example, the presentation control unit 104 may display the report on the terminal device used by the recipient, or it may present the report by printing it out.

[0021] As described above, the information processing device 1 according to this exemplary embodiment employs a configuration comprising: a reception unit 101 that receives the designation of an analysis target; a prompt generation unit 102 that generates prompts instructing the user to provide the information necessary to generate a report showing the analysis results of the analysis target; a generation control unit 103 that inputs the generated prompts into a machine learning-trained generation model to generate answers to the input prompts and causes it to generate answers; and a presentation control unit 104 that presents a report showing the analysis results of the analysis target, which is generated based on the answers.

[0022] With the above configuration, a report showing the analysis results can be generated simply by specifying the object of analysis. Therefore, the above configuration has the effect of making it possible to simplify the analysis of various objects of analysis. Furthermore, the information processing device 1 can also support decision-making based on the analysis results.

[0023] While it might be possible to use market analysis results or other analytical findings from research companies if the goal is simply to obtain analysis results for various subjects, such analyses are typically conducted only once a year at most, making it difficult to obtain results that reflect the latest information. In this respect, using the Information Processing Device 1 makes it possible to obtain comprehensive information on any subject of analysis at any time, grasp the latest trends, and gain business insights (e.g., business insights).

[0024] (Analysis support program) The functions of the information processing device 1 described above can also be implemented by a program. In this exemplary embodiment, the analysis support program causes the computer to function as: a receiving means for receiving the specification of an analysis target; a prompt generating means for generating prompts that instruct the user to provide the information necessary to generate a report showing the analysis results of the analysis target; a generation control means for inputting the prompts into a machine learning-trained generation model to generate answers to the input prompts and causing the model to generate answers; and a presentation control means for presenting a report showing the analysis results of the analysis target, which is generated based on the answers. This analysis support program has the effect of making it possible to facilitate analysis of various analysis targets.

[0025] (Analysis support method flow) The flow of the analysis support method according to this exemplary embodiment will be explained with reference to Figure 2. Figure 2 is a flowchart showing the flow of the analysis support method. Note that the entity executing each step in this analysis support method may be a processor provided in the information processing device 1, a processor provided in another device, or the entity executing each step may be a processor provided in a different device.

[0026] In S1 (reception processing), at least one processor receives the specification of the object to be analyzed.

[0027] In S2 (prompt generation process), at least one processor generates a prompt instructing the system to provide the information necessary to generate a report showing the analysis results of the analysis target specified in S1.

[0028] In S3 (Generation Control Process), at least one processor inputs the prompts generated in S2 into a machine learning-trained generative model to generate answers to the input prompts, causing the model to produce the answers.

[0029] In S4 (Presentation Control Processing), at least one processor presents a report showing the analysis results of the subject, which was generated based on the responses generated in S3.

[0030] As described above, the analysis support method according to this exemplary embodiment employs a configuration that includes: a reception process for receiving the designation of an analysis target; a prompt generation process for generating a prompt that instructs the user to provide the information necessary to generate a report showing the analysis results of the analysis target; a generation control process for inputting the prompt into a machine learning-trained generation model to generate an answer to the input prompt and causing the model to generate an answer; and a presentation control process for presenting a report showing the analysis results of the analysis target, which is generated based on the answer. Therefore, the analysis support method according to this embodiment has the effect of making it possible to facilitate analysis of various analysis targets.

[0031] [Second exemplary embodiment] A second exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same function as those described in the above-described exemplary embodiment are denoted by the same reference numerals, and their descriptions are omitted as appropriate. The scope of application of each technology adopted in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology adopted in this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise. Furthermore, each technology shown in the drawings referenced to describe this exemplary embodiment can also be adopted in other exemplary embodiments included in this disclosure, to the extent that no particular technical problems arise.

[0032] (Configuration of Information Processing Device 1A) The configuration of the information processing device 1A according to this exemplary embodiment will be described with reference to Figure 3. Figure 3 is a block diagram showing the configuration of the information processing device 1A. The information processing device 1A is a device equipped with the function of providing an analysis support service that presents a report showing the analysis results of a specified analysis target. The information processing device 1A may be a local device used by individual users, or it may be a server that provides analysis support services to multiple users.

[0033] As shown in the figure, the information processing device 1A includes a control unit 10A that controls all parts of the information processing device 1A, and a storage unit 11A that stores various data used by the information processing device 1A. The information processing device 1A also includes a communication unit 12A for the information processing device 1A to communicate with other devices, an input unit 13A that receives input to the information processing device 1A, and an output unit 14A for the information processing device 1A to output data. The control unit 10A includes a reception unit 101A, a prompt generation unit 102A, a generation control unit 103A, a presentation control unit 104A, a search unit 105A, a verification unit 106A, a report generation unit 107A, a time-series element detection unit 108A, and a transition information generation unit 109A. Details of the time-series element detection unit 108A and the transition information generation unit 109A will be explained later in the section "About the generation of transition information".

[0034] The reception unit 101A accepts the designation of the object to be analyzed, similar to the reception unit 101 in Exemplary Embodiment 1. As explained in Exemplary Embodiment 1, the object to be analyzed is arbitrary. For example, the reception unit 101A may accept input of a sentence expressing the object to be analyzed in natural language. In addition to the object to be analyzed, the reception unit 101A may also accept the designation of things to be known about the object to be analyzed (in other words, analysis items), or the designation of items to be excluded from the analysis.

[0035] The prompt generation unit 102A generates prompts that instruct the user to provide the information necessary to generate a report showing the analysis results of a specified object of analysis, similar to the prompt generation unit 102 in Exemplary Embodiment 1. More specifically, the prompt generation unit 102A generates prompts that instruct the user to provide the components of a report necessary to generate a report showing the analysis results of a specified object of analysis. For example, the prompt generation unit 102A may generate a prompt that instructs the user to provide the analysis result of one analysis item as one component of the report. Furthermore, if analysis items or items to be excluded from analysis are specified, the prompt generation unit 102A generates prompts that also include these specifications.

[0036] The generation control unit 103A, similar to the generation control unit 103 in Exemplary Embodiment 1, inputs the prompt generated by the prompt generation unit 102A into a machine learning-trained generation model to generate an answer to the input prompt, causing it to generate an answer. As described in Exemplary Embodiment 1, various models can be used as the generation model.

[0037] Furthermore, the generation control unit 103A may generate multiple responses using multiple generation models. In this case, the prompt generation unit 102A generates prompts corresponding to each of the multiple generation models. For example, the prompt generation unit 102A may generate a first prompt instructing the system to provide information necessary for the analysis of a specified analysis item, and a second prompt instructing the system to generate a graph showing the analysis results of the analysis item based on that information. In this case, the generation control unit 103A inputs the first prompt into the first generation model to provide the information necessary for the analysis, and inputs the information contained in that response and the second prompt into the second generation model (which has been trained to generate graphs) to generate a graph. Also, if the prompt generation unit 102A generates multiple prompts (for example, if it generates prompts corresponding to each of the specified analysis items), the generation control unit 103A may sequentially input these multiple prompts into a single generation model to sequentially generate responses. In the following, an example will be described in which a language model that outputs responses in natural language to prompts written in natural language is used as the generation model.

[0038] The presentation control unit 104A, similar to the presentation control unit 104 in Exemplary Embodiment 1, presents a report showing the analysis results of the subject of analysis, generated based on the responses generated by the generation model in accordance with the control of the generation control unit 103A. In this exemplary embodiment, an example is described in which the presentation control unit 104A displays the analysis results on a dashboard (more precisely, a digital dashboard), that is, an example of presenting a dashboard-format report. Furthermore, in this exemplary embodiment, an example is described in which the presentation control unit 104A presents a report (which can also be called a dashboard) generated by the report generation unit 107A.

[0039] The search unit 105A searches for information related to the specified analysis target. The information obtained by the search unit 105A is used to generate responses by the generative model. This allows information that the generative model has not learned to be reflected in the responses it generates. Details of the search performed by the search unit 105A are explained in the "About Search" section below.

[0040] The verification unit 106A verifies whether the content of the response is factual by comparing the response generated by the generation model under the control of the generation control unit 103A with verification data whose content has been confirmed to be factual. The verification method is arbitrary. For example, the verification unit 106A may perform the above verification using a language model that has been trained to output whether or not two input sentences have the same content.

[0041] The report generation unit 107A generates a report showing the analysis results of the subject of analysis based on the responses generated by the generation model under the control of the generation control unit 103A. For example, the report generation unit 107A may generate the report by arranging each component included in the responses generated by the generation model onto the report format. Alternatively, the report generation unit 107A may generate report components (e.g., graphs, etc.) from the responses generated by the generation model, and then generate the report by arranging these components onto the report format.

[0042] As described above, the information processing device 1A includes a reception unit 101A that receives the designation of an analysis target, a prompt generation unit 102A that generates prompts instructing the user to provide the information necessary to generate a report showing the analysis results of the analysis target, a generation control unit 103A that inputs the generated prompts into a machine learning-trained generation model to generate answers to the input prompts and generates answers, and a presentation control unit 104A that presents a report showing the analysis results of the analysis target generated based on the answers. Therefore, similar to the information processing device 1, the effect is obtained that analysis of various analysis targets can be made easier.

[0043] Furthermore, as described above, the information processing device 1A includes a verification unit 106A that verifies whether the content of the response is factual by comparing the response generated by the generation model under the control of the generation control unit 103A with verification data whose content has been confirmed to be factual. The presentation control unit 104A then presents a report generated based on the response that the verification unit 106A has determined to be factual. As a result, in addition to the effects of the information processing device 1 described in Exemplary Embodiment 1, the possibility of presenting a report with content that is not factual can be reduced.

[0044] Furthermore, as described above, the prompt generation unit 102A generates prompts that instruct the user to provide the components necessary to generate a report showing the analysis results of a specified analysis target. The information processing device 1A also includes a report generation unit 107A that generates a report by arranging the components included in the response generated when the generation control unit 103A inputs the generated prompts into a generation model. The presentation control unit 104A then presents the report generated by the report generation unit 107A. As a result, in addition to the effects of the information processing device 1 described in Exemplary Embodiment 1, the information processing device 1A alone can generate and present a report.

[0045] It should be noted that report generation does not necessarily have to be performed by the information processing device 1A. For example, the presentation control unit 104A can present a report by using a service that accepts input of information necessary for report generation and generates a report using that information. In this case, the presentation control unit 104A can simply send the response generated by the generation model, or information extracted from that response, to the service to generate the report. Furthermore, the process of displaying the dashboard may also be performed by a device other than the information processing device 1A (for example, a server that provides information presentation services via a dashboard). In this case, the presentation control unit 104A presents the dashboard to the user by sending various data necessary for displaying the dashboard to the device that displays the dashboard.

[0046] (Regarding search) As described above, the search unit 105A searches for information related to the specified analysis target. For example, the search unit 105A may perform a search on a database that records current events information and detect relevant current events information related to the analysis target. This makes it possible to present a report that reflects current events. Here, "current events information" refers to current events, that is, information about events happening in society at any given time. For example, news articles published in newspapers or online media (which may also be published as videos), data obtained from interviews for newspapers, news commentary articles, and texts and images posted on SNS (Social Networking Service), etc., are examples of current events information.

[0047] The information detected by the search unit 105A (for example, the relevant current events information mentioned above) is used to generate responses using the generative model. Therefore, from the perspective of enabling the generative model to generate highly reliable responses, it is desirable to search databases that contain highly reliable information. For example, databases containing current events information such as interview data and article data generated by media outlets that actually conduct interviews and disseminate current events information, such as newspaper companies, may be used as search targets.

[0048] The search targets of the search unit 105A are not limited to the examples described above. For example, the search unit 105A may search databases containing specialized and reliable information, such as professional journals in fields related to the content of the prompt. Alternatively, the search unit 105A may search databases containing laws and regulations, for example. In addition, the search unit 105A may search databases of research companies that conduct credit checks on companies. Furthermore, the search unit 105A may detect information related to the prompt (which may be current events information or other information) by crawling the internet.

[0049] Furthermore, the search method is also arbitrary. For example, the search unit 105A may extract keywords contained in the prompt and perform a keyword search using those keywords. Alternatively, for example, the search unit 105A may convert part or all of the prompt into a feature vector and detect information related to the prompt by performing a vector search using the feature vector. Moreover, for example, the search unit 105A may use both keyword search and vector search in combination.

[0050] As described above, the information processing device 1A includes a search unit 105A that detects relevant current events information related to the analysis target from a database that records current events information. The generation control unit 103A is configured to input a prompt containing the detected relevant current events information into a generation model to generate a response. Therefore, in addition to the effects achieved by the information processing device 1, the information processing device 1A can also provide the effect of presenting a report that reflects relevant current events information related to the analysis target from the current events information recorded in the database. Note that there may be one database or multiple databases to be searched.

[0051] (Regarding the generation of transition information) The time-series element detection unit 108A detects multiple related current events information items that are about the same subject but have different time series from among the multiple related current events information items detected by the search unit 105A.

[0052] The method for detecting multiple related current events information with different time series for the same subject is arbitrary. For example, the time series element detection unit 108A may first detect multiple related current events information for the same subject, and then detect related current events information with different time series from among the detected multiple related current events information. One method for detecting multiple related current events information for the same subject is to detect related current events information containing the same keyword. Alternatively, for example, one can convert each related current events information into a feature vector and detect related current events information whose feature vectors are similar (the similarity of the feature vectors is above a threshold, or the similarity of the feature vectors falls within a predetermined number of items). After detecting multiple related current events information for the same subject in this way, a description of the time series (year, month, day, etc.) can be extracted from each related current events information, and multiple related current events information with different time series can be detected based on the extracted description.

[0053] The transition information generation unit 109A generates transition information showing the time-series changes of the target based on multiple related current events information detected by the time-series element detection unit 108A. The generated transition information is included in the report and presented by the presentation control unit 104A.

[0054] Transition information can be any information that allows recognition of the time-series changes of the subject shown in the relevant current events information. For example, the transition information generation unit 109A may extract numerical values ​​(e.g., sales figures) related to the subject from each relevant current events information and generate a sequence of numbers arranged in chronological order as transition information. Alternatively, for example, the transition information generation unit 109A may generate numerical values ​​(e.g., rate of change) or graphs obtained from the above sequence as transition information.

[0055] Thus, the information processing device 1A includes a time-series element detection unit 108A that detects multiple related current events information items that are about the same subject but have different time series from among multiple related current events information items detected by the search unit 105A, and a transition information generation unit 109A that generates transition information showing the time-series changes of the subject based on the multiple related current events information items detected by the time-series element detection unit 108A. The presentation control unit 104A then presents a report including the generated transition information. Therefore, in addition to the effects performed by the information processing device 1, the information processing device 1A can provide useful information such as the time-series changes of matters shown in the related current events information related to the object of analysis.

[0056] For example, if the target of analysis is a specific product, the time-series element detection unit 108A may detect multiple related current events information with different time series from among the multiple related current events information detected by the search unit 105A, all of which describe the sales of that product. In this case, the transition information generation unit 109A can generate transition information showing the time-series changes in the sales of the product.

[0057] As described above, the prompt generation unit 102A may generate prompts instructing the user to provide the components of a report necessary to generate a report showing the analysis results of a specified target. In this case, multiple components of the report may be extracted from the responses generated by the generation model. The time series element detection unit 108A may detect multiple components that describe the same target but have different time series from such multiple components. The transition information generation unit 109A may then generate transition information showing the series changes of the target.

[0058] For example, suppose the responses generated by the generative model include sentences describing the market size of the target market but with different time series (e.g., a sentence describing the market size in 2023 and a sentence describing the market size in 2024). In this case, the time series element detection unit 108A detects these sentences. Then, the transition information generation unit 109A generates transition information showing the time series changes in the market size of the target market. In this case, the transition information generation unit 109A may, for example, input each detected sentence into a language model to generate transition information showing the time series changes in the market size of the target market.

[0059] (Example of a report) Figure 4 shows an example of a report presented by the information processing device 1A. In the example in Figure 4, the user of the information processing device 1A inputs input data A1, which specifies the analysis target, into the information processing device 1A. Input data A1 indicates that the analysis target is "healthcare-related products." Input data A1 also indicates that the analysis items are "market size," "market overview," and "major companies," and that "health foods" are to be excluded.

[0060] Input data A1 may be, for example, natural language text. In this case, the user may input the object to be analyzed into the information processing device 1A as text or as voice input. In the latter case, the text generated by having the information processing device 1A or another device perform speech recognition on the input voice can be obtained as input data A1.

[0061] The reception unit 101A of the information processing device 1A acquires input data A1 and passes the acquired input data A1 to the prompt generation unit 102A. The prompt generation unit 102A then generates a prompt instructing the user to provide the information necessary to generate a report showing the analysis results of the specified analysis target, "healthcare-related products". In the example in Figure 4, prompt A2 is generated, which includes the analysis target, analysis items, and items to exclude extracted from input data A1, and instructs the user to provide the analysis results, which are the information necessary to generate the report. The prompt generation unit 102A can generate prompt A2, for example, by incorporating the information extracted from input data A1 into a pre-prepared template.

[0062] Prompt A2 instructs the user to organize the analysis results by analysis item and answer in order of priority. In this way, the prompt generation unit 102A may generate prompts that instruct the user to determine the priority of the components of the report. In the example in Figure 4, the priority is determined by outputting the analysis results for each analysis item, which are components of the report, in order of priority. However, the system may also output a numerical value indicating the priority determination result or a priority category (for example, priority "high", "medium", "low", etc.). Furthermore, "priority" can be rephrased as importance, impact, attention level, etc.

[0063] Furthermore, Prompt A2 may include information indicating the purpose of the response. For example, if a graph is to be generated based on the response to the prompt and the generated graph is to be included in the report, a prompt can be used that instructs the user to provide the information necessary to generate the graph. Prompt A2 may also include information indicating the purpose of the analysis. For example, by including keywords such as "market analysis" or "market research" in Prompt A2, it becomes possible to generate responses suitable for "market analysis" or "market research." In addition, Prompt A2 may also include information indicating the target audience for the report. This allows for, for example, the creation of different reports for general employees and management on the same subject of analysis.

[0064] In input data A1, the user only needs to specify the target of analysis; specifying analysis items or items to exclude is not mandatory. In the analysis support service provided by the information processing device 1A, even if the user does not specify the analysis items, the prompt generation unit 102A generates appropriate prompts according to the target of analysis, making it possible to present a report that includes the analysis results of analysis items corresponding to the target of analysis. For example, the prompt generation unit 102A may generate a prompt that instructs the output of analysis results for multiple analysis items in order to generate a report "for market analysis". This makes it possible to present a report that includes the analysis results of analysis items necessary for market analysis, even without specifying the analysis items.

[0065] Next, the search unit 105A, based on prompt A2, retrieves relevant current events information related to the analysis target from DB1, a database that records current events information. For example, the search unit 105A may convert prompt A2 into a feature vector and use the resulting feature vector to perform a vector search on DB1 to retrieve relevant current events information. In this case, it is desirable to record the feature vector of each current event in DB1, associated with that information. In the example in Figure 4, the relevant current events information A3 includes newspaper data, interview data, and the opinions of commentators (all related to "healthcare-related products" described as the analysis target in prompt A2). In addition to such data, DB1 may also contain other information, such as IR (Inventor Relations) materials, materials showing the results of previous analyses or surveys, and press releases and news releases.

[0066] Next, the generation control unit 103A inputs the prompt A2, including the relevant current events information A3, into the generation model M1 to generate the response. In the example in Figure 4, response A4 is generated, which arranges the analysis results of each analysis item specified in prompt A2 in order of priority. Furthermore, the content of response A4 is based on the relevant current events information A3. For example, the explanation of market size in response A4 is generated based on the description of the market size of the target market in newspaper data, etc., included in the relevant current events information A3.

[0067] Although not shown in Figure 4, the response A4 generated by the generation model M1 is verified by the verification unit 106A to determine whether its content is factual or not. The verification unit 106A may use the data recorded in DB1, or more specifically, the relevant current events information A3 used to generate response A4, as verification data whose content has been confirmed to be factual for this verification. This makes it possible to detect if a response A4 is generated whose content differs from the relevant current events information A3 (i.e., is not factual), and to prevent the presentation of a report with factually incorrect content.

[0068] For example, by having the prompt generation unit 102A generate a prompt that instructs the system to output the answer along with the basis for that answer, it is possible to output the referenced relevant current events information as the basis for an answer generated by referencing relevant current events information. In this case, the verification unit 106A can verify whether the content of the answer is factual by comparing the generated answer with the relevant current events information that serves as its basis.

[0069] Next, the report generation unit 107A generates a report using the answer A4, which the verification unit 106A has determined to be factual. In the example in Figure 4, all the items in answer A4 are determined to be factual, and a report A5 is generated that includes all of those items. If the verification unit 106A determines that some of the items in answer A4 are not factual, the report generation unit 107A generates a report using the remaining parts of answer A4 excluding the undetermined part.

[0070] In the example shown in Figure 4, the report generation unit 107A generates report A5 by arranging the analysis results of each analysis item, which are components of the report shown in response A4, onto a predetermined report format. At this time, the report generation unit 107A may arrange the analysis results shown in response A4 directly onto the report format, or it may acquire an image showing the analysis results shown in response A4 and arrange the acquired image in place of the analysis results. Alternatively, the report generation unit 107A may arrange the acquired image together with the analysis results.

[0071] The image showing the analysis results may, for example, be the transition information generated by the transition information generation unit 109A or an image generated using the transition information. Alternatively, the image showing the analysis results may be an image extracted from relevant current events information referenced during the generation of the analysis results. Furthermore, the image showing the analysis results may be generated by inputting the analysis results into a generative model trained on machine learning to generate images corresponding to the content of the input prompt.

[0072] Report A5 presents the analysis results for "Market Size," "Market Overview," and "Major Companies." The placement of these analysis results corresponds to the priority ranking determined by the generative model M1. Specifically, "Market Size," which is the first output in response A4, has the highest priority, and therefore the analysis results for "Market Size" are placed in the most prominent upper left corner of Report A5. The analysis results for "Market Overview," which is the second output in response A4 and has the second highest priority, are placed in the upper right corner of Report A5, and the analysis results for "Major Companies," which is the third output in response A4 and has the third highest priority, are placed below the analysis results for "Market Size" and "Market Overview."

[0073] As described above, the prompt generation unit 102A may generate prompts that instruct the user to answer with the components of a report necessary to generate a report showing the analysis results of the subject being analyzed, and may also generate prompts that instruct the user to determine the priority of the components. The report generation unit 107A may then generate a report in which components with higher priority are prioritized. This provides the effect of making it easier for the user to recognize high-priority analysis results, in addition to the effects of the information processing device 1.

[0074] Here, "prioritizing placement" means placing components with higher priority in a way that is more easily recognizable to the user than components with lower priority. For example, the report format may be divided into multiple areas in advance, and each area may be ranked in order of how easily it catches the user's eye. In this case, the report generation unit 107A should place the components in order of priority, from highest to lowest, in the higher-ranked areas. Alternatively, the report generation unit 107A may place the components from top to bottom in order of priority, or place only high-priority items, or similar configurations.

[0075] (Specifying multiple analysis targets) The reception unit 101A can accept the designation of multiple analysis targets at once or sequentially. In this case, a report is generated for each analysis target. If multiple reports are generated, the display control unit 104A may display a list of the generated reports for each analysis target and then display the details of the report selected by the user from the displayed list.

[0076] Figure 5 shows an example of a display screen that lists reports corresponding to multiple analysis targets. In the example screen B1 shown in Figure 5, a reduced version of the report generated for the analysis target "XXX" is displayed as selection item B111. In addition, reduced versions of the reports generated for the analysis targets "ABC" and "XYZ" are displayed as selection items B112 and B113. By selecting one of the selection items B111 to B113, the user can display the report corresponding to that selection (the normal-sized report, not a reduced version).

[0077] Furthermore, the update mark B12 is displayed in association with the selection item B111. The update mark B12 is a display object that indicates that the report content has been updated. As will be explained in more detail later, the information processing device 1A can also monitor the analysis targets. By displaying a list of reports for each analysis target being monitored, or the selection items corresponding to each report, and notifying the user with the update mark B12 when a report's content is updated, it becomes possible to easily monitor multiple analysis targets.

[0078] Furthermore, the presentation control unit 104A may present information related to the content of each report along with the report. For example, the presentation control unit 104A may extract words contained in each report to generate a word cloud and present the generated word cloud along with the report. This allows users to recognize keywords of interest in each report before reading the report. A word cloud is a display of each word contained in the target document, with font sizes corresponding to their frequency of appearance. The word cloud may be generated by the presentation control unit 104A or by another device other than the information processing device 1A.

[0079] (Example of report display) Figure 6 shows an example of a report display screen. In the example screen B2 shown in Figure 6, the updated location list display button B21 is displayed. As described above, the information processing device 1A can monitor the analysis target, and the report for the analysis target being monitored is updated as needed. The updated location list display button B21 is a display object for displaying a list of updated locations in the report. When the operation to display the updated location list button B21 is performed, the display control unit 104A displays a list of updated locations in the report.

[0080] Furthermore, in example screen B2, the analysis results for each of the multiple analysis items are displayed. Specifically, in example screen B2, the market size analysis results include graph B221 showing the trend of the global market size, graph B222 showing the trend of the domestic market size, and the average annual growth rate of the global and domestic market size, B223. In this way, it is also possible to present multiple analysis results for a single analysis item.

[0081] Graphs B221 and B222 may be generated by the transition information generation unit 109A, or they may be extracted from relevant current events information. The same applies to the average annual growth rate B223. Furthermore, the average annual growth rate B223 can also be calculated from graphs B221 and B222 or the numerical data from which they are derived.

[0082] Furthermore, update mark B23 is displayed in conjunction with graph B222. Update mark B23 indicates that graph B222 has been updated. In this way, if there is an update to any of the analysis items of the analysis target being monitored, the presentation control unit 104A may display update mark B23 on the updated analysis item in the report to clearly indicate that it has been updated.

[0083] Furthermore, in example screen B2, a market overview description B224 is displayed as the result of the market overview analysis. The description B224 may be generated by a generation model based on relevant current events information, for example. Also, in example screen B2, a graph B225 is displayed showing the results of the business opportunity analysis. The graph B225 may be generated by the report generation unit 107A, or it may be extracted from relevant current events information.

[0084] Furthermore, annotation comment B24 is displayed in conjunction with graph B225. Annotation comment B24 is a display object that indicates points of interest in the analysis results. Annotation comment B24 can be generated by a generative model. For example, the prompt generation unit 102A may generate a prompt that instructs the user to respond with the analysis results of the subject being analyzed, and may also generate a prompt that extracts points of particular interest in the analysis results and instructs the user to respond with comments about those points. By inputting such a prompt into the generative model, the generative model can be made to generate a response that includes content to be presented as an annotation comment.

[0085] Furthermore, in example screen B2, the major company lists B226 and B227 and the graph B228 showing the classification results of the major company categories are displayed as the analysis results of major companies. The major company lists B226 and B227 may be extracted from, for example, relevant current events information. The report generation unit 107A can also display the major company lists B226 and B227 by arranging the major companies included in the responses of the generation model according to predetermined criteria (in order of market capitalization for B226, and in order of number of users for B227). Similarly, the graph B228 may be extracted from relevant current events information or generated based on the responses of the generation model.

[0086] Furthermore, screen example B2 includes a display area B229 that displays related news related to the analysis results of each analysis item. For example, the report generation unit 107A may display the relevant current events information used to generate the answers by the generation model as related news in display area B229.

[0087] Furthermore, in example screen B2, the correction request window B25 is displayed overlaid. The correction request window B25 includes an input field B251 and a submit button B252. The user enters the items they wish to correct in the report into input field B251 and operates the submit button B252. As a result, the user's correction request is received by the reception unit 101A, and the report is updated.

[0088] Here, as described above, the generative model used to generate the report may be a language model that has been trained on natural language. In this case, the reception unit 101A may accept natural language feedback about the content of the presented report. When the reception unit 101A receives feedback, the prompt generation unit 102A generates a new prompt that includes the content of the received feedback and instructs the recipient to provide the information necessary to generate a report showing the analysis results of the subject of analysis. Next, the generation control unit 103A inputs the newly generated prompt into the generative model to generate a new response. Then, the presentation control unit 104A presents an updated report based on the new response, that is, a report that reflects the user's feedback. This provides the effect of easily reflecting the user's intentions in the report, in addition to the effects of the information processing device 1.

[0089] For example, suppose a user enters "Please explain in detail how legal compliance is handled in the business opportunity analysis results" into input field B251 and then presses the submit button B252. In this case, the reception unit 101A receives this text as user feedback and passes it on to the prompt generation unit 102A. Next, the prompt generation unit 102A adds the above text to the most recent prompt generated for the report to be corrected (for example, prompt A2 shown in Figure 4) to generate a new prompt. The generation control unit 103A inputs this new prompt into the generation model to generate a new response. This response includes a detailed explanation of legal compliance. Next, the report generation unit 107A updates the report using the newly generated response, and the presentation control unit 104A presents the updated report. The presentation control unit 104A may also notify the user of the updated parts of the report, for example, by displaying the update mark B23 shown in Figure 6.

[0090] Furthermore, user feedback is not limited to requests to supplement the report content, as in the example above. For example, the reception unit 101A can accept instructions to revise the report content, as well as instructions to change the analytical perspective or analytical items, instructions to change the output format, such as replacing written explanations with images that illustrate the explanations, instructions to change the arrangement of each component in the report, and instructions to change the target audience for the report (e.g., for general employees / for management, etc.).

[0091] (Regarding the presentation of relevant current events information that formed the basis of the analysis results) Here, as shown in screen example B2 in Figure 6, if each analysis result and related news, i.e., the related current events information on which each analysis result is based, are displayed in different areas, it may become difficult for the user to recognize the correspondence between each analysis result and the related current events information on which it is based. For this reason, the display control unit 104A may display each analysis result and the related current events information on which it is based in association with each other.

[0092] Furthermore, the display control unit 104A may display relevant current events information used to obtain the analysis results displayed in the display area of ​​the report, in response to a predetermined operation on that display area. This will be explained with reference to Figure 7. Figure 7 shows an example of the display of relevant current events information.

[0093] Screen example B3 in Figure 7 displays a report containing analysis results for each analysis item, such as market size, similar to screen example B2 in Figure 6. Screen example B3 differs from B2 in that screen example B3 does not include the display field B229 which displays related news.

[0094] In the example shown in Figure 7, the cursor Cur1 is positioned over Graph B31, which displays the analysis results for the domestic market size. In the example in Figure 7, when the cursor Cur1 is positioned over Graph B31, a pop-up screen B33 is displayed showing the related news, or current events, that formed the basis of Graph B31. Pop-up screen B33 displays headlines B331 to B333, which show the related current events that formed the basis of Graph B31.

[0095] In other words, in the example shown in Figure 7, the display control unit 104A displays the relevant current events information used to generate graph B31 in response to the operation of moving cursor Cur1 to graph B31 in the report shown in screen example B3. This allows the user to check the underlying relevant current events information with an intuitive and simple operation of moving cursor Cur1 to the analysis result for which they want to check the relevant current events information. Note that the operation to display the relevant current events information is arbitrary and is not limited to the example in Figure 7.

[0096] Furthermore, in the example screen shown in Figure 7, the update mark B32 is displayed in association with graph B31. The update mark B32 is also displayed in association with headline B331, one of the headlines B331 to B333 displayed on the pop-up screen B33. In this way, the presentation control unit 104A may display the relevant current events information that caused the analysis results to be updated in a way that allows it to be distinguished from other relevant current events information. This makes it easier for the user to recognize the relevant current events information that caused the analysis results to be updated.

[0097] (Process flow) The processing flow performed by the information processing device 1A will be explained based on Figure 8. Figure 8 is a flowchart showing the processing flow performed by the information processing device 1A. The flowchart in Figure 8 includes each process of the analysis support method according to this exemplary embodiment.

[0098] In S11 (reception processing), the reception unit 101A receives the specification of the analysis target. For example, the reception unit 101A may receive the specification of the analysis target entered by the user of the analysis support service into the terminal device they operate, via the communication unit 12A. Alternatively, for example, the reception unit 101A may receive the specification of the analysis target via the input unit 13A. In S11, the reception unit 101A may also receive, for example, the specification of analysis items or the specification of items to be excluded from the analysis.

[0099] In S12 (prompt generation process), the prompt generation unit 102A generates a prompt that instructs the system to provide the information necessary to generate a report showing the analysis results of the analysis target specified in S11. For example, the prompt generation unit 102A may generate a prompt that instructs the system to provide the components of the report necessary to generate a report showing the analysis results of the analysis target specified in S11.

[0100] In S13, the search unit 105A searches for multiple pieces of relevant current events information related to the analysis target using the prompt generated in S12. Note that information other than current events information may also be included in the search target in S13. Alternatively, the processing in S13 may be performed before S12. In that case, the prompt generation unit 102A generates a prompt using the relevant current events information detected by the search.

[0101] In S14, the time-series element detection unit 108A detects multiple pieces of related current events information from among the multiple pieces of related current events information detected in S13, which are related to the same subject but have different time series. For example, if the analysis item includes "market size," the time-series element detection unit 108A may detect multiple pieces of related current events information (e.g., news articles) that describe the market size of the analysis target specified in S11 but have different time series. If no related current events information is detected in S14, the process proceeds to S16 without performing the processing in S15.

[0102] In S15, the transition information generation unit 109A generates transition information showing the time-series changes of the target based on the multiple related current events information detected in S14. For example, if multiple related current events information with different time series describing the market size of the target of analysis are detected in S14, the transition information generation unit 109A generates transition information (e.g., a graph, a sequence of numbers, numerical data, or text) showing the time-series changes of the market size of the target of analysis. A generation model may be used to generate the transition information. The generation model may be the same as the one used in S16, or a different model may be used.

[0103] In S16 (generation control processing), the prompt generated in S12 is input to the generation model along with the relevant current events information detected in S13 to generate a response. Note that when performing the processing in S16, the prompt generation unit 102A may regenerate the prompt using the relevant current events information detected in S13.

[0104] In S17, the verification unit 106A verifies whether the content of the response is factual by comparing the response generated in S16 with verification data whose content has been confirmed to be factual. The verification in S17 is performed for each component of the report (specifically, the analysis results of each analysis item) included in the response generated in S16.

[0105] In S18, the report generation unit 107A generates a report by arranging each component included in the response generated in S16 onto the report format. The report generation unit 107A generates the report using the components from the response generated in S16 that were determined to be factual in S17, and does not include components that were determined to be false. The report generation unit 107A also includes the transition information generated in S15 in the report either as is or after processing it, such as by creating a graph.

[0106] In S19 (presentation control processing), the presentation control unit 104A presents a report showing the analysis results of the subject of analysis, which was generated based on the report generated in S18, i.e., the response generated by the control in S16. This completes the processing shown in Figure 8. In S18, the report generation unit 107A may generate a dashboard-style report. In this case, in S19, the presentation control unit 104A displays the analysis results on the dashboard.

[0107] (Monitoring of the subjects of analysis) As described above, the information processing device 1A can also monitor the object of analysis. In this case, the search unit 105A searches for relevant current events information at predetermined search timings. If the search unit 105A detects new relevant current events information, the prompt generation unit 102A generates a new prompt using the new relevant current events information to instruct the system to provide the information necessary to generate a report showing the analysis results of the object of analysis. The generation control unit 103A then inputs the generated new prompt into the generation model to generate a new response, and the presentation control unit 104A presents an updated report based on the generated new response. This provides the effect of easily performing continuous monitoring of the object of analysis, in addition to the effects of the information processing device 1. Furthermore, it also makes it possible to immediately alert on changes in the object of analysis. For example, if the market for a certain product or service is the object of analysis, it is possible to notify the system of changes in that market.

[0108] Furthermore, to ensure that the latest news and other information are reflected in the report, it is desirable to continuously record the latest news and other information in the databases that the search unit 105A searches. It is also effective to include web pages such as news sites that publish the latest news and topics in the search targets of the search unit 105A.

[0109] Figure 9 is a flowchart of the process related to monitoring the subject of analysis. The process in Figure 9 is performed, for example, after the process in Figure 8 has finished executing.

[0110] In S21, the search unit 105A determines whether a predetermined search timing has arrived. If the result in S21 is YES, the process proceeds to S22. If the result in S21 is NO, the process in S21 is repeated after a predetermined time. The search timing can be predetermined. For example, the search timing may be the timing when the database targeted for search by the search unit 105A is updated, or the timing may be when a predetermined period (e.g., 1 day) has elapsed since the previous search.

[0111] In S22, the search unit 105A performs a search for relevant current events information. The search for relevant current events information can be performed in the same manner as in S13 in Figure 8. Even if the search method and search conditions are the same as the previous search, new relevant current events information may be detected if the database etc. that the search unit 105A searches has been updated. Note that for the second and subsequent searches, the search conditions may be changed from the previous search.

[0112] In S23, the search unit 105A determines whether or not it has detected new relevant current events information as a result of the search in S22. New relevant current events information refers to relevant current events information that was not detected in previous searches. If the result in S23 is YES, the process proceeds to S24. On the other hand, if the result in S23 is NO, the process returns to S21.

[0113] In S24, the prompt generation unit 102A generates a new prompt that instructs the system to provide the information necessary to generate a report showing the analysis results of the subject under analysis, using the new relevant current events information detected in the search in S22. For example, the prompt generation unit 102A may generate a prompt that includes the new relevant current events information detected in the search in S22 and instructs the system to provide the information necessary to generate a report showing the analysis results of the subject under analysis, using the new relevant current events information as a reference. Alternatively, for example, the prompt generation unit 102A may generate a prompt that includes the newly detected relevant current events information and instructs the system to update the previous response based on the new relevant current events information. Note that "previous response" refers to the response generated in S16 of Figure 8 if it is the first update of the report, and the response used in the most recent update if it is the second or subsequent update of the report.

[0114] In S25, the generation control unit 103A inputs the new prompt generated in S24 into the generation model to generate a new answer. Subsequently, in S26, the verification unit 106A verifies whether the content of the answer generated in S25 is factual or not.

[0115] In S27, the report generation unit 107A updates the report using the components of the report included in the response generated in S25 that were determined to be factual in S26. It is not necessary to use all of the components that were determined to be factual in S26 for the update; it is sufficient to use only the components whose content has been updated from those that were determined to be factual in S26.

[0116] In S28, the display control unit 104A notifies the user that the report has been updated. The process then returns to S21. The method of notification in S28 is arbitrary. For example, as shown in screen example B1 of Figure 5, the update may be notified by displaying an update mark B12 associated with the updated report. Alternatively, the display control unit 104A may notify the user of the update by means of, for example, email.

[0117] [Variation] In Exemplary Embodiment 2, an example was shown in which a prompt is generated instructing the user to answer with the components of a report showing the analysis results of the subject to be analyzed, and a report is generated by inputting this prompt into the generation model and arranging each component included in the generated answer. However, the method of generating the report is not limited to this example. For example, the prompt generation unit 102A of Exemplary Embodiment 2 may generate a prompt instructing the user to generate a report in which the analysis results for each analysis item of the subject to be analyzed are arranged. In this case, since a report in which each analysis result is arranged is output from the generation model, the presentation control unit 104A only needs to present the output report. Furthermore, by including a sentence instructing the user to arrange the analysis results for each analysis item in order of priority in the prompt, it is also possible to automatically generate a report in which the analysis results with higher priority are arranged to stand out.

[0118] The entities executing each process described in the exemplary embodiments above are arbitrary and not limited to the examples given. For example, a system having the same functions as the information processing devices 1 and 1A can be constructed using multiple devices that can communicate with each other. Furthermore, the entities executing each process shown in the flowcharts in Figures 8 and 9 may be a single device (which can also be called a processor) or multiple devices (which can also be called processors).

[0119] [Examples of implementation using software] Some or all of the functions of the information processing devices 1,1A (hereinafter also referred to as "each of the above devices") may be implemented by hardware such as integrated circuits (IC chips) or by software.

[0120] In the latter case, each of the above devices is implemented, for example, by a computer that executes instructions for a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as computer C) is shown in Figure 10. Figure 10 is a block diagram showing the hardware configuration of computer C, which functions as each of the above devices.

[0121] Computer C comprises at least one processor C1 and at least one memory C2. Memory C2 stores a program P that causes computer C to operate as each of the above-mentioned devices. In computer C, processor C1 reads program P from memory C2 and executes it, thereby realizing each of the above-mentioned devices.

[0122] For processor C1, for example, a CPU (Central Processing Unit), GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Point Number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination thereof can be used. For memory C2, for example, flash memory, HDD (Hard Disk Drive), SSD (Solid State Drive), or a combination thereof can be used.

[0123] Computer C may also be equipped with RAM (Random Access Memory) for loading program P at runtime and for temporarily storing various data. Furthermore, computer C may be equipped with communication interfaces for sending and receiving data with other devices. Additionally, computer C may be equipped with input / output interfaces for connecting input / output devices such as keyboards, mice, displays, and printers.

[0124] Furthermore, program P can be recorded on a non-temporary, tangible recording medium M that is readable by computer C. Such a recording medium M could be, for example, tape, disk, card, semiconductor memory, or programmable logic circuitry. Computer C can acquire program P via such a recording medium M. Program P can also be transmitted via a transmission medium. Such a transmission medium could be, for example, a communication network or broadcast waves. Computer C can also acquire program P via such a transmission medium.

[0125] Furthermore, each of the above functions of each of the above devices may be implemented by a single processor in a single computer, by multiple processors in a single computer working together, or by multiple processors in each of multiple computers working together. In addition, the programs for implementing each of the above functions in each of the above devices may be stored in a single memory in a single computer, distributed and stored in multiple memories in a single computer, or distributed and stored in multiple memories in each of multiple computers.

[0126] [Additional Notes] This disclosure includes the technologies described in the following appendices. However, the present invention is not limited to the technologies described in the following appendices, and various modifications are possible within the scope of the claims.

[0127] (Note A1) An information processing device comprising: a receiving means for receiving the designation of an object to be analyzed; a prompt generating means for generating a prompt that instructs the user to provide the information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control means for inputting the prompt into a machine learning-trained generation model to generate an answer to the input prompt, thereby causing the model to generate an answer; and a presentation control means for presenting a report showing the analysis results of the object to be analyzed, which is generated based on the answer.

[0128] (Appendix A2) The information processing device described in Appendix A1 comprises a search means for detecting relevant current events information related to the object of analysis from a database for recording current events information, and the generation control means inputs a prompt containing the relevant current events information to the generation model to generate a response.

[0129] (Note A3) The information processing apparatus described in Appendix A2, wherein the search means performs a search for the relevant current events information each time a predetermined search timing occurs, the prompt generation means generates a new prompt instructing the system to provide the information necessary to generate a report showing the analysis results of the subject of analysis when the search means detects new relevant current events information, the generation control means inputs the new prompt into the generation model to generate a new response, and the presentation control means presents an updated report based on the new response.

[0130] (Note A4) An information processing device according to any one of Appendix A1 to A3, comprising a verification means for verifying whether the content of a response is factual by comparing the response generated by the generation model with verification data whose content has been confirmed to be factual, and the presentation control means presents a report generated based on the response whose content has been determined to be factual by the verification means.

[0131] (Note A5) The information processing apparatus described in Appendix A2, wherein the generation model is a language model that has been machine-learned to learn natural language, the receiving means receives natural language feedback on the content of the presented report, the prompt generation means generates a new prompt that instructs the model to provide the information necessary to generate a report that includes the content of the feedback and shows the analysis results of the subject of analysis, the generation control means inputs the new prompt to the generation model to generate a new response, and the presentation control means presents an updated report based on the new response.

[0132] (Note A6) The information processing device according to any one of Appendix A1 to A5, wherein the prompt generation means generates prompts instructing the user to provide the components of a report necessary to generate a report showing the analysis results of the object being analyzed, and the generation control means generates a report by arranging the components included in the generated response when the prompt is input into the generation model, and the presentation control means presents the report generated by the report generation means.

[0133] (Note A7) The information processing apparatus according to Appendix A6, wherein the prompt generating means generates a prompt that instructs the user to answer with the components of a report necessary to generate a report showing the analysis results of the object being analyzed, and the prompt instructs the user to determine the priority of the components, and the report generating means generates a report in which the components with higher priority are preferentially arranged.

[0134] (Note A8) The information processing apparatus described in Appendix A2 or A3 comprises: a time-series element detection means for detecting multiple related current events information concerning the same subject but with different time series from among multiple related current events information detected by the search means; and a transition information generation means for generating transition information indicating the time-series change of the subject based on the multiple related current events information detected by the time-series element detection means, wherein the presentation control means presents the report including the transition information.

[0135] (Note B1) An analysis support method comprising: an acceptance process in which at least one processor accepts the designation of an object to be analyzed; a prompt generation process instructing the user to provide information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control process instructing a machine learning-trained generative model to input the prompt to generate an answer to the input prompt and cause the model to generate an answer; and a presentation control process for presenting a report showing the analysis results of the object to be analyzed, generated based on the answer.

[0136] (Note B2) The analysis support method according to Appendix B1, wherein the at least one processor includes a search process for detecting relevant current events information related to the object of analysis from a database for recording current events information, and in the generation control process, the at least one processor inputs a prompt containing the relevant current events information to the generation model to generate an answer.

[0137] (Note B3) The analysis support method described in Appendix B2, wherein at least one processor performs a search for relevant current events information each time a predetermined search timing occurs, and when new relevant current events information is detected by the search process, generates a new prompt instructing the system to provide the information necessary to generate a report showing the analysis results of the subject of analysis using the new relevant current events information, inputs the new prompt into the generation model to generate a new response, and presents an updated report based on the new response.

[0138] (Note B4) The analysis support method according to any one of Appendix B1 to B3, wherein the at least one processor includes a verification process that verifies whether the content of the response is factual by comparing the response generated by the generation model with verification data whose content has been confirmed to be factual, and in the presentation control process, the at least one processor presents a report generated based on the response whose content has been determined to be factual by the verification process.

[0139] (Note B5) The analysis support method described in Appendix B2, wherein the generative model is a language model that has been trained on natural language, the at least one processor receives natural language feedback on the content of the presented report, generates a new prompt instructing it to provide information necessary to generate a report that includes the content of the feedback and shows the analysis results of the subject of analysis, inputs the new prompt into the generative model to generate a new response, and presents an updated report based on the new response.

[0140] (Note B6) The analysis support method according to any one of Appendix B1 to B5, wherein in the prompt generation process, the at least one processor generates a prompt instructing the system to provide the components of a report necessary to generate a report showing the analysis results of the subject of analysis; the at least one processor generates a report by arranging the components included in the response generated by inputting the prompt into the generation model in the generation control process; and in the presentation control process, the at least one processor presents the report generated by the report generation process.

[0141] (Note B7) The analysis support method according to Appendix B6, wherein in the prompt generation process, the at least one processor generates a prompt instructing the system to respond with the components of a report necessary to generate a report showing the analysis results of the subject of analysis, and the prompt instructing the system to determine the priority of the components, and in the report generation process, the at least one processor generates a report in which the components with higher priority are preferentially arranged.

[0142] (Note B8) The analysis support method according to Appendix B2 or B3, wherein the at least one processor includes a time-series element detection process that detects from among a plurality of related current events information detected by the search process a plurality of related current events information that are about the same subject but have different time series, and a transition information generation process in which the at least one processor generates transition information indicating the time-series change of the subject based on the plurality of related current events information detected in the time-series element detection process, and in the presentation control process, the at least one processor presents the report including the transition information.

[0143] (Note C1) An analysis support program that causes a computer to function as: a receiving means for receiving the designation of an object to be analyzed; a prompt generating means for generating prompts that instruct the computer to provide the information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control means for inputting the prompts into a machine learning-trained generation model to generate answers to the input prompts and causing the computer to generate answers; and a presentation control means for presenting a report showing the analysis results of the object to be analyzed, which is generated based on the answers.

[0144] (Note C2) The analysis support program described in Appendix C1, wherein the computer functions as a search means for detecting relevant current events information related to the object of analysis from a database that records current events information, and the generation control means inputs a prompt containing the relevant current events information into the generation model to generate an answer.

[0145] (Note C3) The analysis support program as described in Appendix C2, wherein the search means performs a search for relevant current events information each time a predetermined search timing occurs, the prompt generation means generates a new prompt instructing the system to provide the information necessary to generate a report showing the analysis results of the subject of analysis when new relevant current events information is detected by the search means, the generation control means inputs the new prompt into the generation model to generate a new response, and the presentation control means presents an updated report based on the new response.

[0146] (Note C4) The analysis support program described in any of Appendix C1 to C3, wherein the computer functions as a verification means for verifying whether the content of a response is factual by comparing the response generated by the generation model with verification data whose content has been confirmed to be factual, and the presentation control means presents a report generated based on the response whose content has been determined to be factual by the verification means.

[0147] (Note C5) The analysis support program described in Appendix C2, wherein the generation model is a language model that has been machine-learned using natural language, the receiving means receives natural language feedback on the content of the presented report, the prompt generation means generates a new prompt instructing the model to provide the information necessary to generate a report that includes the content of the feedback and shows the analysis results of the subject of analysis, the generation control means inputs the new prompt into the generation model to generate a new response, and the presentation control means presents an updated report based on the new response.

[0148] (Appendix C6) The analysis support program described in any of Appendix C1 to C5, wherein the prompt generation means generates prompts instructing the respondent to provide the components of a report necessary to generate a report showing the analysis results of the object being analyzed, the computer functions as a report generation means that generates a report by arranging the components included in the response generated when the generation control means inputs the prompts into the generation model, and the presentation control means presents the report generated by the report generation means.

[0149] (Note C7) The analysis support program described in Appendix C6, wherein the prompt generation means generates a prompt instructing the respondent to provide the components of a report necessary to generate a report showing the analysis results of the object being analyzed, and the prompt instructs the respondent to determine the priority of the components, and the report generation means generates a report in which the components with higher priority are preferentially arranged.

[0150] (Note C8) The computer functions as a time-series element detection means for detecting multiple related current events information concerning the same subject but with different time series from among multiple related current events information detected by the search means, and a transition information generation means for generating transition information showing the time-series changes of the subject based on the multiple related current events information detected by the time-series element detection means, and the presentation control means presents the report including the transition information, as described in Appendix C2 or C3.

[0151] (Note D1) An information processing device comprising at least one processor, the at least one processor performing: an acceptance process for receiving a designation of an object to be analyzed; a prompt generation process for generating a prompt that instructs the user to provide information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control process for inputting the prompt into a machine learning-trained generative model to generate an answer to the input prompt, causing the model to generate an answer; and a presentation control process for presenting a report showing the analysis results of the object to be analyzed, which is generated based on the answer.

[0152] The information processing device may also include memory. Furthermore, the memory may store a program that causes at least one processor to execute each of the aforementioned processes.

[0153] (Note D2) The information processing apparatus according to Appendix D1, wherein the at least one processor performs a search process to detect relevant current events information related to the object of analysis from a database that records current events information, and in the generation control process, the at least one processor inputs a prompt containing the relevant current events information to the generation model to generate a response.

[0154] (Note D3) The information processing device described in Appendix D2, wherein at least one processor performs a search for relevant current events information each time a predetermined search timing occurs, and when new relevant current events information is detected by the search process, generates a new prompt instructing the system to provide the information necessary to generate a report showing the analysis results of the subject of analysis using the new relevant current events information, inputs the new prompt into the generation model to generate a new response, and presents an updated report based on the new response.

[0155] (Note D4) The information processing device according to any one of the appendices D1 to D3, wherein at least one processor performs a verification process to verify whether the content of the response is factual by comparing the response generated by the generation model with verification data whose content has been confirmed to be factual, and in the presentation control process, the at least one processor presents a report generated based on the response whose content has been determined to be factual by the verification process.

[0156] (Note D5) The information processing device described in Appendix D2, wherein the generative model is a language model that has been trained on natural language, the at least one processor receives natural language feedback on the content of the presented report, generates a new prompt instructing the model to provide the information necessary to generate a report that includes the content of the feedback and shows the analysis results of the subject of analysis, inputs the new prompt into the generative model to generate a new response, and presents an updated report based on the new response.

[0157] (Note D6) An information processing device according to any one of the appendices D1 to D5, wherein in the prompt generation process, the at least one processor generates a prompt instructing the system to provide the components of a report necessary to generate a report showing the analysis results of the object being analyzed; the at least one processor executes a report generation process in which it generates a report by arranging the components included in the response generated by inputting the prompt into the generation model in the generation control process; and in the presentation control process, the at least one processor presents the report generated by the report generation process.

[0158] (Note D7) The information processing apparatus according to Appendix D6, wherein in the prompt generation process, the at least one processor generates a prompt instructing the system to respond with the components of a report necessary to generate a report showing the analysis results of the object being analyzed, and the prompt instructing the system to determine the priority of the components, and in the report generation process, the at least one processor generates a report in which the components with higher priority are preferentially arranged.

[0159] (Note D8) The information processing apparatus according to Appendix D2 or D3, wherein the at least one processor performs a time-series element detection process to detect multiple related current events information concerning the same subject but with different time series from among multiple related current events information detected in the search process, and a transition information generation process to generate transition information indicating the time-series change of the subject based on the multiple related current events information detected in the time-series element detection process, and in the presentation control process, the at least one processor presents the report including the transition information.

[0160] (Note E) A non-temporary recording medium that records an analysis support program that causes a computer to execute: an acceptance process for receiving the specification of an object to be analyzed; a prompt generation process for generating prompts that instruct the computer to provide the information necessary to generate a report showing the analysis results of the object to be analyzed; a generation control process for inputting the prompts into a machine learning-trained generation model to generate answers to the input prompts and causing it to generate answers; and a presentation control process for presenting a report showing the analysis results of the object to be analyzed, which is generated based on the answers. [Explanation of Symbols]

[0161] 1. Information Processing Device 101 Reception area (reception method) 102 Prompt generation unit (prompt generation means) 103 Generation Control Unit (Generation Control Means) 104 Presentation Control Unit (Presentation Control Means) 1A Information Processing Device 101A Reception area (reception method) 102A Prompt generation unit (prompt generation means) 103A Generation Control Unit (Generation Control Means) 104A Presentation Control Unit (Presentation Control Means) 105A Search Unit (Search Means) 106A Verification Unit (Verification Means) 107A Report generation unit (report generation means) 108A Time-series element detection unit (time-series element detection means) 109A Transition information generation unit (transition information generation means) M1 Generative Model D1 Database

Claims

1. A means of receiving requests for the items to be analyzed, A prompt generation means that generates a prompt instructing the user to provide the information necessary to generate a report showing the analysis results of the subject of analysis, A generation control means that inputs the aforementioned prompt into a machine learning-trained generation model to generate an answer to the input prompt, and causes the model to generate an answer. An information processing device comprising: a presentation control means for presenting a report showing the analysis results of the subject of analysis, which is generated based on the aforementioned response.

2. The system includes a search means for detecting relevant current events information related to the subject of analysis from a database that records current events information, The information processing apparatus according to claim 1, wherein the generation control means inputs a prompt including the relevant current events information to the generation model to generate a response.

3. The search means performs the search for the relevant current events information each time a predetermined search timing occurs. When the search means detects new relevant current events information, the prompt generation means generates a new prompt that instructs the user to provide the information necessary to generate a report showing the analysis results of the subject of analysis using the new relevant current events information. The generation control means inputs the new prompt to the generation model to generate a new response. The information processing apparatus according to claim 2, wherein the presentation control means presents an updated report based on the new response.

4. The system includes a verification means for verifying whether the content of a response is factual by comparing the response generated by the generation model with verification data whose content has been confirmed to be factual. The information processing apparatus according to any one of claims 1 to 3, wherein the presentation control means presents a report generated based on a response whose content has been determined to be factual by the verification means.

5. The aforementioned generative model is a language model that has been trained on natural language, The aforementioned receiving means accepts natural language feedback regarding the content of the presented report. The prompt generation means generates a new prompt that includes the content of the feedback and instructs the respondent to provide the information necessary to generate a report showing the analysis results of the subject of analysis. The generation control means inputs the new prompt to the generation model to generate a new response. The information processing apparatus according to claim 2, wherein the presentation control means presents an updated report based on the new response.

6. The prompt generation means generates prompts that instruct the respondent to provide the components of a report necessary to generate a report showing the analysis results of the object being analyzed, The generation control means includes a report generation means that generates a report by arranging each component included in the response generated when the prompt is input to the generation model, The information processing device according to any one of claims 1 to 3, wherein the presentation control means presents the report generated by the report generation means.

7. The prompt generation means generates prompts that instruct the respondent to provide the components of a report necessary to generate a report showing the analysis results of the object being analyzed, and that instruct the respondent to determine the priority of the components. The information processing apparatus according to claim 6, wherein the report generation means generates a report in which the components with higher priority are preferentially arranged.

8. A time-series element detection means detects from among the multiple related current events information detected by the search means multiple related current events information concerning the same subject but with different time series, The system comprises a transition information generation means that generates transition information indicating the time-series changes of the target based on the plurality of related current events information detected by the time-series element detection means, The information processing apparatus according to claim 2 or 3, wherein the presentation control means presents the report including the transition information.

9. At least one processor, The reception process for specifying the target of analysis, A prompt generation process that generates a prompt instructing the user to provide the information necessary to generate a report showing the analysis results of the subject of analysis, A generation control process that inputs the aforementioned prompt into a machine learning-trained generation model to generate an answer to the input prompt, and causes the model to generate an answer. An analysis support method, comprising: a presentation control process that presents a report showing the analysis results of the subject of analysis, generated based on the aforementioned response.

10. Computers, A means of receiving requests for the items to be analyzed. Prompt generation means that generates a prompt instructing the user to provide the information necessary to generate a report showing the analysis results of the subject of analysis, A generation control means that inputs the prompt into a machine learning-trained generation model to generate an answer to the input prompt, and causes the model to generate an answer, and An analysis support program that functions as a presentation control means for presenting a report showing the analysis results of the subject of analysis, generated based on the aforementioned response.

Images