Information processing device and information processing method

Abstract

This invention provides an information processing device and method that can evaluate events from various perspectives. [Solution] The information processing device 1 includes an event information acquisition unit 131 that acquires event result information showing the content of an event and the results of the event; a perspective reception unit 132 that receives perspectives for analyzing an event from a user terminal; a hypothesis acquisition unit 133 that generates a first prompt to cause a large-scale language model to generate hypotheses corresponding to the received perspectives regarding the content of the event and the results of the event shown in the event result information, inputs the generated first prompt to the large-scale language model, and acquires the hypotheses generated by the large-scale language model; a verification result acquisition unit 135 that generates a second prompt to cause the large-scale language model to verify the hypotheses based on the event result information, inputs the generated second prompt to the large-scale language model, and acquires verification result information generated by the large-scale language model; and an output unit 136 that outputs the verification result information to a user terminal.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus and an information processing method.

Background Art

[0002] Conventionally, techniques for assisting planners who plan events are known. For example, Patent Document 1 discloses a system that collects personal information of visitors who attended an event and evaluates the effect on the implementation of the event based on the collected personal information.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the conventional technology, the effect on a plurality of predetermined items is evaluated, and there is a problem that verification based on viewpoints different from the plurality of items cannot be performed from various aspects corresponding to the implemented event.

[0005] Therefore, the present invention has been made in view of these points, and an object thereof is to enable an event to be evaluated from various viewpoints.

Means for Solving the Problems

[0006] An information processing device according to a first aspect of the present invention includes: an event information acquisition unit that acquires event result information indicating the content of an event that has been carried out and the results of the event; a viewpoint reception unit that receives viewpoints for analyzing the event from a user terminal used by the user; a hypothesis acquisition unit that generates a first prompt to cause a large-scale language model to generate a hypothesis corresponding to the viewpoint received by the viewpoint reception unit regarding the content of the event and the results of the event indicated by the event result information, inputs the generated first prompt to the large-scale language model, and acquires the hypothesis generated by the large-scale language model; a verification result acquisition unit that generates a second prompt to cause the large-scale language model to verify the hypothesis based on the event result information, inputs the generated second prompt to the large-scale language model, and acquires verification result information indicating the results of the verification generated by the large-scale language model; and an output unit that outputs the verification result information to the user terminal.

[0007] The hypothesis acquisition unit further includes a hypothesis reception unit that generates a first prompt causing a large-scale language model to generate a plurality of hypotheses corresponding to the viewpoints received by the viewpoint reception unit, inputs the generated first prompt to the large-scale language model, acquires the plurality of hypotheses generated by the large-scale language model, displays the plurality of hypotheses on the user terminal, and accepts the selection of a hypothesis from the user terminal. The verification result acquisition unit may generate a second prompt causing the large-scale language model to verify the hypothesis selected by the hypothesis reception unit based on the event result information.

[0008] The hypothesis receiving unit may prioritize displaying hypotheses on the user terminal that match or are similar to hypotheses previously selected by the user of the user terminal, among the multiple hypotheses. The viewpoint reception unit may display multiple viewpoints on the user terminal, and among the multiple viewpoints, it may prioritize displaying viewpoints previously selected by the user of the user terminal on the user terminal, and accept selections from among the multiple viewpoints.

[0009] The aforementioned perspective is an indicator that shows the performance of an event. The event information acquisition unit acquires similar event result information, which is event result information corresponding to similar events that are similar to the aforementioned event and have been held. The perspective reception unit may, from among the multiple indicators, preferentially display on the user terminal multiple perspectives that have indicator values ​​where the difference between the similar event indicated by the similar event result information and the aforementioned event indicated by the event result information is relatively large, and accept selection from among the multiple perspectives.

[0010] The event information acquisition unit may acquire similar event result information, which corresponds to similar events that are similar to the event and have been carried out. The hypothesis acquisition unit may generate the first prompt, which generates a number of hypotheses that focus on the differences between the content of the similar events indicated by the similar event result information and the content of the event indicated by the event result information.

[0011] The hypothesis acquisition unit may generate the first prompt which generates hypotheses corresponding to the relatively important evaluation axes among a plurality of evaluation axes for evaluating the event and the viewpoints received by the viewpoint reception unit.

[0012] The event information acquisition unit acquires event result information including evaluation information showing the evaluation content for each of the multiple evaluation axes of the implemented event, and the hypothesis acquisition unit may identify relatively important evaluation axes based on the evaluation content for each of the multiple evaluation axes shown in the event result information, and generate the first prompt which generates hypotheses corresponding to the identified evaluation axes and the viewpoints received by the viewpoint reception unit.

[0013] The aforementioned large-scale language model may be a model configured to allow referencing the content of past events, which are events held in the past, past event result information showing the performance of those past events, and hypotheses that were tested for those past events.

[0014] The aforementioned large-scale language model may be a model that has been pre-trained using past events, which are events held in the past, past event result information showing the performance of those past events, and hypotheses that were tested for those past events as training data.

[0015] A second aspect of the present invention relates to an information processing method comprising: a step of obtaining event result information, which is performed by a computer, showing the content of an event and the results of the event; a step of receiving a viewpoint for analyzing the event from a user terminal used by a user; a step of generating a first prompt that causes a large-scale language model to generate a hypothesis corresponding to the received viewpoint regarding the content of the event and the results of the event shown in the event result information; a step of inputting the generated first prompt into the large-scale language model and obtaining the hypothesis generated by the large-scale language model; a step of generating a second prompt that causes the large-scale language model to verify the hypothesis based on the event result information; a step of inputting the generated second prompt into the large-scale language model and obtaining verification result information showing the results of the verification generated by the large-scale language model; and a step of outputting the verification result information to the user terminal. [Effects of the Invention]

[0016] The present invention has the effect of enabling events to be evaluated from various perspectives. [Brief explanation of the drawing]

[0017] [Figure 1] This is a diagram illustrating the overview of an information processing device. [Figure 2] This diagram shows the functional configuration of an information processing device. [Figure 3] This figure shows an example of an analysis support page. [Figure 4] This figure shows an example of the first prompt. [Figure 5] This figure shows an example where multiple hypotheses are displayed on the analysis support page. [Figure 6]It is a diagram showing an example of the second prompt. [Figure 7] It is a diagram showing an example in which verification result information is displayed on an analysis support page. [Figure 8] It is a sequence diagram showing the processing flow in an information processing apparatus, a large language model, and a user terminal.

Mode for Carrying Out the Invention

[0018] [Overview of Information Processing Apparatus 1] FIG. 1 is a diagram showing an overview of an information processing apparatus 1. The information processing apparatus 1 is communicably connected to a large language model (LLM) 2 and a user terminal 3 used by a user who analyzes an event, and is a computer for assisting the user in analyzing an event.

[0019] The large language model 2 is, for example, a generative AI (Artificial Intelligence) such as ChatGPT (registered trademark) or Claude (registered trademark), and outputs an answer corresponding to the instruction indicated by the prompt in response to receiving the prompt as instruction information. The user terminal 3 is, for example, a computer such as a personal computer or a tablet.

[0020] Before the analysis by the user is performed, the information processing apparatus 1 acquires event result information indicating the content of the event that has been carried out and the performance of the event (1 in FIG. 1). Then, the information processing apparatus 1 receives a perspective for analyzing the event from the user terminal 3 (2 in FIG. 1).

[0021] The information processing apparatus 1 generates a first prompt for causing the large language model 2 to generate a hypothesis corresponding to the received perspective with respect to the content of the event and the performance of the event indicated by the event result information (3 in FIG. 1). The information processing apparatus 1 inputs the generated first prompt into the large language model 2 (4 in FIG. 1) and acquires the hypothesis generated by the large language model 2 (5 in FIG. 1).

[0022] When the information processing device 1 acquires a hypothesis, it generates a second prompt (in Figure 1, (6)) that prompts the large-scale language model 2 to verify the hypothesis based on the event result information. The information processing device 1 inputs the generated second prompt to the large-scale language model 2 (in Figure 1, (7)) and acquires verification result information that shows the verification results generated by the large-scale language model 2 (in Figure 1, (8)). The information processing device 1 then outputs the acquired verification result information to the user terminal 3 (in Figure 1, (9)).

[0023] Here, the large-scale language model 2 is a model that incorporates, for example, a search-enhanced generation (RAG) function and is configured to reference past event result information, which shows the content and performance of past events, hypotheses that were tested for those past events, and the results of testing those hypotheses based on the past event result information. As a result, the large-scale language model 2 can appropriately test hypotheses based on event result information, based on the hypotheses that were tested for past events and the results of testing those hypotheses, in response to the perspective received from the user.

[0024] In this way, the user of user terminal 3 can input various perspectives they wish to investigate into the information processing device 1, and then check hypotheses regarding the perspectives for analyzing the event corresponding to those various perspectives, as well as the results of verifying those hypotheses, and evaluate the event. Therefore, the information processing device 1 can enable the user to evaluate the event from various perspectives.

[0025] [Functional configuration of the information processing device 1] Next, the functional configuration of the information processing device 1 will be described. Figure 2 is a diagram showing the functional configuration of the information processing device 1. The information processing device 1 includes a communication unit 11, a storage unit 12, and a control unit 13.

[0026] The communication unit 11 is a communication interface for sending and receiving data with the large-scale language model 2, user terminal 3, etc., via a communication network such as a mobile phone line or the internet. The memory unit 12 is a storage medium for storing various types of data, and includes ROM (Read Only Memory), RAM (Random Access Memory), hard disk, and SSD (Solid State Drive). The memory unit 12 stores programs to be executed by the control unit 13. For example, the memory unit 12 stores programs that cause the control unit 13 to function as an event information acquisition unit 131, a viewpoint reception unit 132, a hypothesis acquisition unit 133, a hypothesis reception unit 134, a verification result acquisition unit 135, and an output unit 136.

[0027] The control unit 13 is, for example, a CPU (Central Processing Unit). By executing a program stored in the memory unit 12, the control unit 13 functions as an event information acquisition unit 131, a viewpoint reception unit 132, a hypothesis acquisition unit 133, a hypothesis reception unit 134, a verification result acquisition unit 135, and an output unit 136.

[0028] The event information acquisition unit 131 acquires event result information that shows the content and performance of the target event, which is an event that has been held and is the target event for analysis. For example, the event information acquisition unit 131 accesses a web page or the like that shows the content and performance of the target event and acquires information that shows the content of the target event, such as the date and time of the event, the location of the event, the theme, scale, purpose, target audience, advertising media, frequency of advertising, and the quality of services provided.

[0029] Furthermore, the event information acquisition unit 131 acquires information that shows the performance of the target event, including values ​​for indicators such as event sales, costs, revenue, number of attendees, congestion levels by time of day, average stay time of event participants, satisfaction level, number of exhibitors, and repeat rate. These indicators are also used as evaluation axes for evaluating the event.

[0030] Furthermore, the event information acquisition unit 131 acquires event result information, including evaluation information that shows the evaluation content for each of the multiple evaluation axes of the target event. As mentioned above, the multiple evaluation axes are information that shows actual performance details such as event sales, costs, revenue, number of attendees, congestion level by time of day, average stay time of event participants, satisfaction level, number of exhibitors, and repeat rate. The target event is evaluated for each of the multiple evaluation axes based on, for example, the results of comparison with the performance details of past events of the same scale that have been held in the past, or the aggregated results of responses to questionnaires about the target event received from participants, etc. Then, a score is determined for the target event as evaluation information that shows the evaluation content for each of the multiple evaluation axes.

[0031] The event information acquisition unit 131 may, for example, access a webpage or the like that displays the score of the target event and acquire evaluation information for the target event, including the score as an evaluation content for each of the multiple evaluation axes. The event information acquisition unit 131 may also use the large-scale language model 2 to acquire evaluation information for the target event, including the score as an evaluation content for each of the multiple evaluation axes.

[0032] Furthermore, the event information acquisition unit 131 acquires similar event result information, which corresponds to similar events that have been held and are similar to the target event. Similar events are, for example, events that match or are similar to the target event in at least one of the following aspects: the date and time of the event, the location of the event, the theme, scale, purpose, and target audience.

[0033] For example, the event information acquisition unit 131 accesses a webpage or the like that shows the content and results of similar events and acquires similar event result information. The event information acquisition unit 131 may also acquire evaluation information that includes scores for each of the multiple evaluation axes of similar events.

[0034] The event information acquisition unit 131 acquires event result information and similar event result information from web pages, etc., but is not limited to this. The event information acquisition unit 131 may also acquire event result information and similar event result information from the user terminal 3.

[0035] The perspective reception unit 132 receives perspectives for analyzing events from the user terminal 3 used by the user. For example, when the perspective reception unit 132 receives a request from the user terminal 3 to access an analysis support page that provides support for the analysis of a target event, it displays the analysis support page on the user terminal 3.

[0036] Here, the perspective may be an indicator that shows the performance content of the target event. The perspective reception unit 132 displays an analysis support page on the user terminal 3 that accepts at least one of multiple perspectives, where multiple indicators showing the performance content of the target event are presented as multiple perspectives. Figure 3 is a diagram showing an example of the analysis support page. As shown in Figure 3, the analysis support page displays a selection field for selecting one perspective from among multiple perspectives. When the user presses the selection field, multiple perspectives are displayed, and the user can select a perspective from among them.

[0037] Here, the perspective reception unit 132 may, on the analysis support page, prioritize displaying perspectives previously selected by the user of the user terminal 3 from among multiple perspectives, and accept the selection of a perspective from among multiple perspectives. In this way, the information processing device 1 can easily allow the user to select perspectives that the user is likely to analyze and that the user considers important.

[0038] Furthermore, the perspective reception unit 132 may prioritize displaying on the user terminal 3, as multiple perspectives, multiple indicators that have relatively large differences between the similar events indicated by the similar event result information and the events indicated by the event result information, and accept the selection of a perspective from among the multiple perspectives. In this way, the information processing device 1 can allow the user to select a perspective that differs significantly from similar events and makes it easier to formulate hypotheses.

[0039] The hypothesis acquisition unit 133 generates a first prompt that causes the large-scale language model to generate hypotheses corresponding to the perspectives selected by the perspective reception unit 132 regarding the event content and event results indicated by the event result information. The hypothesis acquisition unit 133 generates a first prompt that causes the large-scale language model 2 to generate multiple hypotheses corresponding to the perspectives selected by the perspective reception unit 132.

[0040] For example, on the analysis support page, a generate button labeled "Generate Hypotheses" is displayed. When the generate button shown on the analysis support page is pressed, the hypothesis acquisition unit 133 generates a first prompt that instructs the large-scale language model 2 to generate multiple hypotheses corresponding to the viewpoint selected by the viewpoint reception unit 132. Figure 4 shows an example of the first prompt. As shown in Figure 4, it can be seen that the system is instructed to generate multiple hypotheses based on the selected viewpoint, based on the event content and the event's performance. The hypothesis acquisition unit 133 then inputs the generated first prompt into the large-scale language model 2 and acquires the multiple hypotheses generated by the large-scale language model 2.

[0041] The hypothesis acquisition unit 133 may generate a first prompt that generates hypotheses corresponding to the relatively important evaluation axes among the multiple evaluation axes for evaluating the event and the viewpoints received by the viewpoint reception unit 132. For example, the hypothesis acquisition unit 133 identifies the relatively important evaluation axes based on the evaluation content for each of the multiple evaluation axes indicated by the event result information.

[0042] For example, the hypothesis acquisition unit 133 calculates the difference between the score of the evaluation content for each of the multiple evaluation axes indicated by the event result information and the score of the evaluation content for each of the multiple evaluation axes indicated by the similar event result information, and identifies the evaluation axis corresponding to the evaluation content with a large difference as a relatively important evaluation axis. Alternatively, the hypothesis acquisition unit 133 may identify a relatively important evaluation axis by identifying an evaluation axis for an evaluation content with a relatively low evaluation content, or by accepting the selection of a relatively important evaluation axis from the user terminal 3. The hypothesis acquisition unit 133 generates a first prompt that generates a hypothesis corresponding to the identified evaluation axis and the viewpoint received by the viewpoint reception unit 132.

[0043] Furthermore, the hypothesis acquisition unit 133 may generate a first prompt that generates multiple hypotheses focusing on the differences between the content of similar events indicated by the similar event result information and the content of events indicated by the event result information. For example, the hypothesis acquisition unit 133 may generate a first prompt that includes the content of the target event, the content of similar events, and an instruction sentence that instructs the extraction of differences between these contents, and generates multiple hypotheses focusing on the extracted differences. In this way, the large-scale language model 2 can be made to focus on the differences in event content and generate multiple hypotheses.

[0044] The hypothesis reception unit 134 displays multiple hypotheses on the user terminal 3 and accepts the selection of a hypothesis from the user terminal 3. The hypothesis reception unit 134 sends multiple hypotheses to the user terminal 3, displays these hypotheses on the analysis support page, and accepts the selection of one of the hypotheses from the user terminal 3. Figure 5 shows an example of multiple hypotheses being displayed on the analysis support page. In the analysis support page shown in Figure 5, it can be seen that multiple hypotheses generated by the large-scale language model 2 are displayed as "Hypotheses by AI". Also, at the bottom of the analysis support page shown in Figure 5, it can be seen that a hypothesis selection button is displayed to accept the selection of the hypothesis to be verified from among these hypotheses, and a verification start button is displayed with the words "Verify hypothesis" to start the verification. The hypothesis reception unit 134 accepts the selection of a hypothesis from the user terminal 3 via the hypothesis selection button.

[0045] Here, the hypothesis reception unit 134 may prioritize displaying on the user terminal 3 hypotheses that match or are similar to hypotheses previously selected by the user of the user terminal 3. In this case, the hypothesis reception unit 134 stores information indicating previously selected hypotheses in the storage unit 12. Then, in response to the hypothesis acquisition unit 133 acquiring multiple hypotheses, the hypothesis reception unit 134 identifies the degree of similarity between the multiple hypotheses and the previously selected hypotheses stored in the storage unit 12. The hypothesis reception unit 134 identifies hypotheses that match or are similar to hypotheses previously selected by the hypothesis acquisition unit 133 among the multiple hypotheses, and prioritizes displaying the identified hypotheses on the analysis support page.

[0046] The verification result acquisition unit 135 generates a second prompt that causes the large-scale language model 2 to verify the hypothesis based on the event result information of the target event. Specifically, after the verification result acquisition unit 135 receives a hypothesis selection from the user terminal 3 on the analysis support page, it receives a command to start hypothesis verification from the user terminal 3 in response to the start verification button being pressed. Upon receiving the command to start verification, the verification result acquisition unit 135 generates a second prompt that causes the large-scale language model 2 to verify the hypothesis selected by the hypothesis reception unit 134, based on the event result information of the target event.

[0047] Figure 6 shows an example of a second prompt. As shown in Figure 6, it can be seen that the selected hypothesis is instructed to be verified based on the content of the event and the actual results of the event. The verification result acquisition unit 135 inputs the generated second prompt to the large-scale language model 2 and acquires verification result information showing the verification results generated by the large-scale language model 2.

[0048] The output unit 136 outputs verification result information to the user terminal 3, which shows the results of hypothesis verification obtained from the large-scale language model 2 by the verification result acquisition unit 135. For example, the output unit 136 displays the verification result information on the analysis support page by sending the acquired verification result information to the user terminal 3.

[0049] Figure 7 shows an example of how verification results information is displayed on the analysis support page. Figure 7 shows that verification results information has been added to the analysis support page shown in Figure 5. In the analysis support page shown in Figure 7, it can be confirmed that the verification results information generated by the large-scale language model 2 is displayed as "Hypothesis Verification Results by AI".

[0050] [Processing flow in information processing device 1] Next, we will explain the processing flow in the information processing device 1. Figure 8 is a sequence diagram showing the processing flow in the information processing device 1, the large-scale language model 2, and the user terminal 3.

[0051] First, the event information acquisition unit 131 acquires event result information and similar event result information (S1). Next, the perspective reception unit 132, upon receiving a request from the user terminal 3 to obtain an analysis page, sends an analysis support page to the user terminal 3 (S2, S3) and displays the analysis support page on the user terminal 3. Then, the perspective reception unit 132 accepts the selection of a perspective from the user terminal 3 (S4).

[0052] Next, the hypothesis acquisition unit 133 generates a first prompt (S5) that causes the large-scale language model 2 to generate hypotheses corresponding to the perspectives received in S4 regarding the content of the event and the actual results of the event as indicated by the event result information. The hypothesis acquisition unit 133 inputs the generated first prompt into the large-scale language model 2 (S6) and acquires hypothesis information indicating the multiple hypotheses generated by the large-scale language model 2 (S7).

[0053] Next, the hypothesis reception unit 134 transmits the multiple hypotheses indicated by the hypothesis information acquired in S7 to the user terminal 3 (S8), thereby displaying the multiple hypotheses on the analysis support page. Then, the hypothesis reception unit 134 accepts the selection of a hypothesis from the user terminal 3 (S9).

[0054] Next, the verification result acquisition unit 135 generates a second prompt that causes the large-scale language model 2 to verify the hypothesis based on the event result information (S10). Then, the verification result acquisition unit 135 inputs the generated second prompt to the large-scale language model 2 (S11) and acquires verification result information that shows the verification results generated by the large-scale language model 2 (S12). The output unit 136 transmits the acquired verification result information to the user terminal 3 (S13) and displays the verification result information on the analysis support page.

[0055] [Example 1] In the above-described embodiment, the large-scale language model 2 is, for example, a model that incorporates a search extension generation function and is configured to allow referencing of past event result information corresponding to past events, hypotheses verified for those past events, and the verification results of those hypotheses based on the past event result information, but is not limited to this.

[0056] Large-scale language model 2 may be a model that has been pre-trained using past event results information, which shows the content and performance of past events, and hypotheses that have been tested for those past events as training data, as well as a model that has been pre-trained using hypotheses and the results of testing those hypotheses based on past event results information as training data.

[0057] In this way, the information processing device 1 can appropriately verify hypotheses based on event result information, based on the hypotheses verified for past events and the verification results of those hypotheses, in relation to the perspective received from the user, similar to how the RAG function allows the large-scale language model 2 to reference past event result information, hypotheses verified for those past events, and the verification results of those hypotheses based on the past event result information.

[0058] [Differentiation 2] In the embodiment described above, the hypothesis receiving unit 134 accepts the selection of one hypothesis, and the verification result acquisition unit 135 generates a second prompt to have the large-scale language model 2 verify the said hypothesis, but it is not limited to this. The hypothesis acquisition unit 133 may cause the large-scale language model 2 to generate one hypothesis, and the hypothesis receiving unit 134 may not accept the selection of a hypothesis, and the verification result acquisition unit 135 may generate a second prompt to have the large-scale language model 2 verify the hypothesis generated by the large-scale language model 2.

[0059] Furthermore, the verification result acquisition unit 135 may generate a second prompt that causes the large-scale language model 2 to verify each of the multiple hypotheses acquired by the hypothesis acquisition unit 133. In this way, the information processing device 1 can output the hypothesis corresponding to the viewpoint selected by the user, and the verification result corresponding to that hypothesis, to the user terminal 3 without accepting any user input, thereby further reducing the user's burden.

[0060] [Effects of this embodiment] As explained above, the information processing device 1 receives perspectives for analyzing the event to be analyzed, generates a first prompt that causes the large-scale language model 2 to generate hypotheses corresponding to the received perspectives regarding the content of the event and the actual performance of the event as indicated by the event result information, inputs the generated first prompt into the large-scale language model 2, and retrieves the hypotheses generated by the large-scale language model 2. Then, the information processing device 1 generates a second prompt that causes the large-scale language model 2 to verify the hypotheses based on the event result information, inputs the generated second prompt into the large-scale language model 2, retrieves verification result information showing the results of the verification generated by the large-scale language model 2, and outputs it to the user terminal 3. In this way, the information processing device 1 can evaluate events from various perspectives.

[0061] Furthermore, this invention will make it possible to contribute to Goal 9 of the United Nations-led Sustainable Development Goals (SDGs), "Build resilient infrastructure, promote inclusive and sustainable industrialization and foster innovation."

[0062] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of its gist. For example, all or part of the apparatus can be configured by functionally or physically distributing and integrating in any unit. Furthermore, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effects of the new embodiments resulting from the combinations are combined with the effects of the original embodiments. [Explanation of symbols]

[0063] 1. Information Processing Device 2. Large-scale language models 3. User terminals 11 Communications Department 12 Storage section 13 Control Unit 131 Event Information Acquisition Department 132 Perspective Reception Department 133 Hypothesis Acquisition Section 134 Hypothesis Reception Department 135 Verification Result Acquisition Unit 136 Output section

Claims

1. An event information acquisition unit acquires event result information that shows the content of the event that was held and the results of the said event, A perspective receiving unit that receives perspectives for analyzing the aforementioned event from the user terminal used by the user, A hypothesis acquisition unit generates a first prompt that causes a large-scale language model to generate multiple hypotheses corresponding to the perspectives received by the perspective receiving unit, based on the content of the event and the actual results of the event as indicated by the event result information, inputs the generated first prompt into the large-scale language model, and acquires the multiple hypotheses generated by the large-scale language model. A hypothesis receiving unit that, among the multiple hypotheses mentioned above, prioritizes displaying hypotheses on the user terminal that match or are similar to hypotheses previously selected by the user terminal, and accepts the selection of a hypothesis from the user terminal. Based on the event result information, the hypothesis receiving unit generates a second prompt to cause the large-scale language model to verify the hypothesis selected by the hypothesis receiving unit, inputs the generated second prompt to the large-scale language model, and obtains verification result information indicating the verification results generated by the large-scale language model. An output unit that outputs the verification result information to the user terminal, An information processing device having

2. An event information acquisition unit acquires event result information that shows the content of the event that was held and the results of the said event, A perspective receiving unit that displays multiple perspectives for analyzing the event from a user terminal used by the user, and accepts selection from the multiple perspectives, wherein the perspective receiving unit prioritizes displaying on the user terminal the perspective previously selected by the user of the user terminal. A hypothesis acquisition unit generates a first prompt that causes a large-scale language model to generate a hypothesis corresponding to the perspective received by the perspective receiving unit, regarding the content of the event and the actual results of the event as indicated by the event result information, inputs the generated first prompt into the large-scale language model, and acquires the hypothesis generated by the large-scale language model. A verification result acquisition unit generates a second prompt to cause the large-scale language model to verify the hypothesis based on the event result information, inputs the generated second prompt to the large-scale language model, and acquires verification result information indicating the verification result generated by the large-scale language model. An output unit that outputs the verification result information to the user terminal, An information processing device having

3. An event information acquisition unit acquires event result information that shows the content of the event that was held and the results of the said event, and acquires similar event result information that corresponds to similar events that are similar to the said event and have been held. A perspective receiving unit that receives perspectives for analyzing the event from a user terminal used by the user, which are indicators that show the actual content of the event, and which preferentially displays on the user terminal as multiple perspectives multiple indicators among multiple indicators that have a relatively large difference between the similar event shown in the similar event result information and the event shown in the event result information, and accepts selection from among the multiple perspectives, A hypothesis acquisition unit generates a first prompt that causes a large-scale language model to generate a hypothesis corresponding to the perspective received by the perspective receiving unit, regarding the content of the event and the actual results of the event as indicated by the event result information, inputs the generated first prompt into the large-scale language model, and acquires the hypothesis generated by the large-scale language model. A verification result acquisition unit generates a second prompt to cause the large-scale language model to verify the hypothesis based on the event result information, inputs the generated second prompt to the large-scale language model, and acquires verification result information indicating the verification result generated by the large-scale language model. An output unit that outputs the verification result information to the user terminal, An information processing device having

4. An event information acquisition unit acquires event result information that shows the content of the event that was held and the results of the said event, and acquires similar event result information that corresponds to similar events that are similar to the said event and have been held. A perspective receiving unit that receives perspectives for analyzing the aforementioned event from the user terminal used by the user, A hypothesis acquisition unit generates a first prompt that corresponds to the perspectives received by the perspective receiving unit regarding the content of the event indicated by the event result information and the actual results of the event, and that causes a large-scale language model to generate multiple hypotheses focusing on the differences between the content of the similar event indicated by the similar event result information and the content of the event indicated by the event result information, inputs the generated first prompt into the large-scale language model, and acquires the multiple hypotheses generated by the large-scale language model. A verification result acquisition unit generates a second prompt to cause the large-scale language model to verify the multiple hypotheses based on the event result information, inputs the generated second prompt to the large-scale language model, and acquires verification result information indicating the verification results generated by the large-scale language model. An output unit that outputs the verification result information to the user terminal, An information processing device having

5. The hypothesis acquisition unit generates the first prompt which generates hypotheses corresponding to the relatively important evaluation axes among a plurality of evaluation axes for evaluating an event and the viewpoints received by the viewpoint reception unit. The information processing apparatus according to any one of claims 1 to 4.

6. The event information acquisition unit acquires the event result information, which includes evaluation information indicating the evaluation content for each of the multiple evaluation axes of the implemented event. The hypothesis acquisition unit identifies relatively important evaluation axes based on the evaluation content for each of the multiple evaluation axes indicated by the event result information, and generates the first prompt which generates hypotheses corresponding to the identified evaluation axes and the viewpoints received by the viewpoint reception unit. The information processing apparatus according to claim 5.

7. The aforementioned large-scale language model is a model configured to reference past events, which are events held in the past, past event results information showing the performance of those past events, and hypotheses that were tested for those past events. The information processing apparatus according to any one of claims 1 to 4.

8. The aforementioned large-scale language model is a model that has been pre-trained using past events (events held in the past), past event results (information indicating the performance of those past events), and hypotheses that were tested for those past events as training data. The information processing apparatus according to any one of claims 1 to 4.

9. A computer executes A step of obtaining event result information that shows the content of the event that was held and the results of the said event, A step of receiving perspectives for analyzing the aforementioned events from the user terminal used by the user, The steps include generating a first prompt that causes a large-scale language model to generate multiple hypotheses corresponding to the received perspectives regarding the content of the event and the actual results of the event as indicated by the event result information, inputting the generated first prompt into the large-scale language model, and obtaining the multiple hypotheses generated by the large-scale language model, The steps include: displaying, on the user terminal, preferentially selecting hypotheses from among the aforementioned multiple hypotheses that match or are similar to hypotheses previously selected by the user terminal; and accepting the selection of a hypothesis from the user terminal; The steps include: generating a second prompt to cause the large-scale language model to verify the hypothesis selected based on the event result information; inputting the generated second prompt into the large-scale language model; and obtaining verification result information indicating the results of the verification generated by the large-scale language model. The steps include outputting the verification result information to the user terminal, An information processing method having

10. A computer executes A step of obtaining event result information that shows the content of the event that was held and the results of the said event, A step of displaying multiple perspectives for analyzing the event from a user terminal used by the user, and accepting a selection from the multiple perspectives, wherein the perspective previously selected by the user of the user terminal is given priority in displaying on the user terminal. The steps include: generating a first prompt that causes a large-scale language model to generate a hypothesis corresponding to the received perspective regarding the content of the event and the actual results of the event as indicated by the event result information; inputting the generated first prompt into the large-scale language model; and obtaining the hypothesis generated by the large-scale language model. The steps include: generating a second prompt to cause the large-scale language model to verify the hypothesis based on the event result information, inputting the generated second prompt into the large-scale language model, and obtaining verification result information indicating the verification result generated by the large-scale language model; The steps include outputting the verification result information to the user terminal, An information processing method having

11. A computer executes The steps include obtaining event result information that shows the content of the event that was held and the results of the said event, and obtaining similar event result information that corresponds to similar events that are similar to the said event and have been held, A step of receiving a perspective for analyzing the event from a user terminal used by the user, which is an indicator showing the actual content of the event, wherein, among a plurality of indicators, a plurality of indicators having an indicator value where the difference between the similar event shown in the similar event result information and the event shown in the event result information is relatively large are displayed on the user terminal as a plurality of perspectives, and the user receives a selection from among the plurality of perspectives. The steps include: generating a first prompt that causes a large-scale language model to generate a hypothesis corresponding to the received perspective regarding the content of the event and the actual results of the event as indicated by the event result information; inputting the generated first prompt into the large-scale language model; and obtaining the hypothesis generated by the large-scale language model. The steps include: generating a second prompt to cause the large-scale language model to verify the hypothesis based on the event result information, inputting the generated second prompt into the large-scale language model, and obtaining verification result information indicating the verification result generated by the large-scale language model; The steps include outputting the verification result information to the user terminal, An information processing method having

12. A computer executes The steps include obtaining event result information that shows the content of the event that was held and the results of the said event, and obtaining similar event result information that corresponds to similar events that are similar to the said event and have been held, A step of receiving perspectives for analyzing the aforementioned events from the user terminal used by the user, The steps include: generating a first prompt that causes a large-scale language model to generate multiple hypotheses corresponding to the received perspective on the content of the event and the actual results of the event as indicated by the event result information, and focusing on the differences between the content of the similar event indicated by the similar event result information and the content of the event indicated by the event result information; inputting the generated first prompt into the large-scale language model; and obtaining the multiple hypotheses generated by the large-scale language model. The steps include: generating a second prompt to cause the large-scale language model to verify the multiple hypotheses based on the event result information, inputting the generated second prompt into the large-scale language model, and obtaining verification result information indicating the results of the verification generated by the large-scale language model; The steps include outputting the verification result information to the user terminal, An information processing method having

Images