Information processing device

Abstract

An information processing device is provided that extracts evaluation results from various perspectives for predetermined items related to an assignment, and makes it possible to propose a proposed answer based on the extracted evaluation results. [Solution] The information processing device of the present invention comprises a predetermined information creation unit that acquires the results of an analysis of the emotions of the text creator regarding the information on evaluation items contained in each piece of text data based on the content of each piece of text data related to the purpose of use, creates relationship information indicating the relationship between the evaluation items and information on emotions for each piece of text data based on the analysis results, and combines the multiple relationship information created for each piece of text data into one piece to create predetermined information that associates the multiple pieces of text data, and a specific information extraction unit that extracts specific information regarding predetermined evaluation items that are associated with information on positive emotions and information on negative emotions from the predetermined information.

Description

[Technical Field]

[0001] The present invention relates to an information processing device. [Background technology]

[0002] Conventionally, for example, companies and local governments have used social listening, utilizing surveys and social networking services (SNS), as a means of improving their own organizations (e.g., their own companies or local communities). In particular, social listening can capture the real voices of many people. Also, conventionally, information processing devices have been proposed that display charts summarizing users' subjective evaluations of content based on the users' speech intentions (see, for example, Patent Document 1). The information processing device disclosed in Patent Document 1 may generate a comparison chart in which comparison items for which multiple users have conflicting subjective evaluations are highlighted using a different color, allowing users to intuitively grasp the comparison items for which the subjective evaluations are conflicting. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2020 / 202719 Summary of the Invention [Problem to be solved by the invention]

[0004] While social listening using SNS can collect people's honest opinions, the large amount of collected data poses challenges in aggregating and extracting areas for improvement and proposing improvement proposals. The technology disclosed in Patent Document 1 primarily involves users listing predetermined items (predetermined evaluation items) related to the problem (purpose of use), and does not automatically extract various evaluation perspectives from large amounts of data. Patent Document 1 also does not consider, for example, proposing solutions (improvement proposals) tailored to the problem based on the evaluation results for the predetermined items.

[0005] The present invention has been made in consideration of the above-mentioned circumstances, and an object of the present invention is to provide an information processing device that extracts evaluation results from various perspectives for predetermined items related to a problem (objective), and makes it possible to propose solutions (improvement proposals, project proposals, strategy proposals, etc.) based on the extracted evaluation results. [Means for solving the problem]

[0006] In order to solve the above problems, an information processing device of the present invention includes a predetermined information creation unit and a specific information extraction unit. The predetermined information creation unit acquires an analysis result of the text creator's emotions regarding information on evaluation items included in each text data based on the content of each of the multiple text data related to the purpose of use. The predetermined information creation unit also creates relationship information indicating the relationship between the evaluation items and information related to emotions for each text data based on the analysis result. The predetermined information creation unit further combines the multiple relationship information created for each of the multiple text data into one to create predetermined information that associates the multiple text data. The specific information extraction unit then extracts specific information related to the predetermined evaluation items associated with information related to positive emotions and information related to negative emotions from the predetermined information. [Effects of the Invention]

[0007] According to the information processing device of the present invention having the above-described configuration, evaluation results from various perspectives can be extracted for specified items (specified evaluation items) related to the problem (purpose of use), and response proposals (improvement proposals, project proposals, strategy proposals, etc.) can be proposed based on the extracted evaluation results. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic configuration diagram of an answer suggestion support system according to a first embodiment of the present invention. [Figure 2] 1 is a hardware configuration diagram of an emotion analysis device and a proposal support device according to a first embodiment of the present invention. [Figure 3]2 is a diagram showing a configuration of input information of the proposal support device according to the first embodiment of the present invention. FIG. [Figure 4] 1 is a diagram showing positive posting examples and negative posting examples to be analyzed by a proposal support device according to a first embodiment of the present invention. FIG. [Figure 5] 1 is a diagram showing an outline of a processing operation of a proposal support device according to a first embodiment of the present invention. [Figure 6] FIG. 2 is a diagram for explaining the content of the creation process of an emotion graph and a text association graph by the proposal support device according to the first embodiment of the present invention. [Figure 7] FIG. 2 is a diagram showing an example of a text association graph created by the proposal support device according to the first embodiment of the present invention. [Figure 8] FIG. 2 is a diagram showing an example of an emotional conflict graph extracted from a text association graph created by the suggestion support device according to the first embodiment of the present invention. [Figure 9] FIG. 2 is a diagram showing an example of a proposal example derived by the proposal support device according to the first embodiment of the present invention. [Figure 10] 1 is a functional block configuration diagram of an answer suggestion support system according to a first embodiment of the present invention. [Figure 11] 3 is a flowchart showing the procedure of a proposal support process performed in the answer proposal support system according to the first embodiment of the present invention. [Figure 12] 5 is a flowchart showing the procedure of an emotion conflict graph extraction process performed in the answer suggestion support system according to the first embodiment of the present invention. [Figure 13] 5 is a flowchart showing the procedure of a process for identifying emotional conflict parts performed in the answer suggestion support system according to the first embodiment of the present invention. [Figure 14] 5 is a flowchart showing the procedure of an emotion conflict graph extraction process performed in the answer suggestion support system according to the first embodiment of the present invention. [Figure 15] FIG. 10 is a functional block configuration diagram of an answer suggestion support system according to a second embodiment of the present invention. [Figure 16]FIG. 10 is a diagram showing an example of a display mode of proposal support information by the proposal support device according to the second embodiment of the present invention. [Figure 17] 10 is a flowchart showing the procedure of a proposal support process performed in an answer proposal support system according to a second embodiment of the present invention. [Figure 18] FIG. 10 is a functional block configuration diagram of an answer suggestion support system according to a third embodiment of the present invention. [Figure 19] FIG. 10 is a diagram showing an example of a prompt used in a proposal support device according to a third embodiment of the present invention, and a result of executing the prompt (an example response). [Figure 20] FIG. 11 is a diagram showing another example of a prompt used in the proposal support device according to the third embodiment of the present invention, and the execution result (reply example) of that prompt. [Figure 21] 10 is a flowchart showing the procedure of a proposal support process performed in an answer proposal support system according to a third embodiment of the present invention. [Figure 22] FIG. 10 is a functional block configuration diagram of an answer suggestion support system according to a fourth embodiment of the present invention. [Figure 23] 10 is a flowchart showing the procedure of a proposal support process performed in an answer proposal support system according to a fourth embodiment of the present invention. [Figure 24] FIG. 10 is a diagram showing an example of a display mode of proposal support information by the proposal support device according to Modification 1 of the present invention. [Figure 25] FIG. 10 is a diagram showing an example of a display mode of proposal support information by a proposal support device according to Modification 2 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0009] A proposal support device and an answer proposal support system including the same according to various embodiments of the present invention will be specifically described below with reference to the drawings. Note that the proposal support device according to the various embodiments described below is an information processing device that supports the creation of answer proposals by utilizing the results of an analysis of people's emotions regarding predetermined items related to a task, which are analyzed based on various text data such as SNS (Social Networking Service) data.

[0010] 1. First embodiment [Configuration of the answer suggestion support system] 1 is a diagram showing the configuration of an answer proposal support system 1 according to the first embodiment. As shown in FIG.

[0011] Within the answer proposal support system 1, the sentiment analysis device 10 is connected to a proposal support device 20. The sentiment analysis device 10 is also connected via a network or the like to a text medium 30 provided outside the answer proposal support system 1. The text medium 30 includes, for example, social media such as SNS and word-of-mouth information, mass media such as newspapers, and media for business information such as questionnaires.

[0012] The emotion analysis device 10 and the proposal support device 20 can each be configured as an information processing device (arithmetic processing device) such as a computer device equipped with an arithmetic function and a communication function. Fig. 2 is a block diagram showing an example of the hardware configuration of a computer device 40 that can be used as the emotion analysis device 10 and the proposal support device 20. Note that the computer device 40 shown in Fig. 2 can also be used as the proposal support devices 21 to 23, which will be described in second to fourth embodiments, respectively, below.

[0013] The computer device 40 includes a CPU (Central Processing Unit) 41, a ROM (Read Only Memory) 42, and a RAM (Random Access Memory) 43, all connected to a bus line 48. The computer device 40 also includes a network I / F (Interface) 44, an operation unit 45, a display unit 46, and a non-volatile storage 47, all connected to the bus line 48. Although not shown in Fig. 2, the computer device 40 also includes various interfaces used when executing input / output processing of various data (various information) between the computer device 40 and external devices.

[0014] The CPU 41 reads out from the ROM 42 to the RAM 43 and executes the program code of software for realizing various processing functions provided in each of the emotion analysis device 10 and the proposal support device 20. At this time, various data such as variables and parameters generated during the calculation process are temporarily written to the RAM 43.

[0015] The network I / F 44 is configured by, for example, a network interface card (NIC) or the like, and transmits and receives various data to and from each device connected via wireless communication.

[0016] The operation unit 45 is configured with, for example, keys and buttons, and generates an operation signal according to the operation content input by an operator (such as a user of the answer proposal support system 1), and supplies the operation signal to the CPU 41. The display unit 46 is configured with, for example, a liquid crystal panel, and displays characters, images, and the like on the screen. The display unit 46 may also be configured with a touch panel, in which case the display unit 46 and the operation unit 45 are configured integrally.

[0017] The nonvolatile storage 47 can be configured, for example, with a hard disk drive (HDD), a solid state drive (SSD), a flexible disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, or a nonvolatile memory. The nonvolatile storage 47 stores an operating system (OS), various parameters, and various programs for causing the computer device 40 to function as the emotion analysis device 10 and the proposal support device 20. Note that information (data) such as programs, tables, and files that realize the functions of the emotion analysis device 10 and the proposal support device 20 may be stored in a recording medium other than the ROM 42 or the nonvolatile storage 47, such as an IC card, an SD card, or a DVD.

[0018] In this embodiment, an example will be described in which the emotion analysis device 10 and the proposal support device 20 are configured as separate entities, but the present invention is not limited to this. For example, the emotion analysis device 10 (emotion analysis device unit) and the proposal support device 20 may be configured as a single information processing device (answer proposal support device). In this case, too, the hardware configuration of the information processing device (answer proposal support device) will be the configuration shown in FIG. 2. When the emotion analysis device 10 and the proposal support device 20 are configured as a single information processing device (answer proposal support device), the hardware of both devices is integrated into one, thereby making it possible to make the answer proposal support system 1 more compact and less expensive.

[0019] [Overview of the emotion analysis device] The emotion analysis device 10 transmits a query (inquiry) corresponding to a request input by a user of the answer proposal support system 1 to the text medium 30, and acquires various text data corresponding to the transmitted query from the text medium 30.

[0020] For example, when the purpose of use of the answer proposal support system 1 is to propose regional revitalization, the sentiment analysis device 10 transmits various information (conditions) such as the name of the city, town, or village in the region targeted for regional revitalization and facility type information (for example, park, parking lot, etc.) as a query to the text medium 30. Then, the sentiment analysis device 10 acquires various text data including the information transmitted as the query from the text medium 30.

[0021] Furthermore, the sentiment analysis device 10 analyzes the sentiment of the text creator (contributor) regarding the terms of the evaluation items (related terms described below) included in each piece of acquired text data. At this time, the sentiment analysis device 10 classifies the sentiment of the text creator regarding the evaluation items for each piece of text data into three types: "positive," "negative," and "neutral."

[0022] For example, if the purpose of using the answer proposal support system 1 is to propose regional revitalization, the emotion analysis device 10 analyzes whether the text creator's emotion toward evaluation items (terms) such as parks and parking lots included in the text data is "positive," "negative," or "neutral."

[0023] Then, the sentiment analysis device 10 outputs the sentiment analysis results of each piece of text data to the proposal support device 20. Note that as a sentiment analysis technique for text data by the sentiment analysis device 10, for example, an analysis technique using AI (Artificial Intelligence) technology that is already in practical use can be used.

[0024] [Configuration of input information for the proposal support device] Next, we will explain the structure of the emotion analysis results (hereinafter simply referred to as "input information") of each piece of text data input from the emotion analysis device 10 to the proposal support device 20. Fig. 3 is a diagram showing the structure of the input information for each piece of text data.

[0025] In this embodiment, as shown in Fig. 3, the proposal support device 20 acquires, as input information, data that is a set of information on types (attributes) called "emotions," "emotion words," "related words," and "domain tags" for each piece of text data. Note that it is also possible to acquire time information, area information, and the like as input information to the proposal support device 20 in addition to the above information. Furthermore, the emotion analysis device 10 can acquire information from text data as natural language processing centered on emotions.

[0026] The "emotion" information is set with information indicating the emotion ("positive," "negative," or "neutral") of the text creator (poster) regarding the "related words" (evaluation items). In this case, one of "positive," "negative," or "neutral" is set as the "emotion" information for each piece of text data. However, if multiple pieces of "emotion" information for a "related word" are included in one piece of text data, one piece of "emotion" information is selected appropriately depending on the content of the text data. Furthermore, attribute information indicating the "emotion" is set in association with the "emotion" information.

[0027] "Emotion word" information (information related to emotions) contains information about expressions (wordings) included in the text data that are used to determine the type of "emotion" ("positive," "negative," or "neutral"). "Emotion word" information is also set in association with attribute information that indicates the "emotion word." In this embodiment, only one piece of "emotion word" information is set for each piece of text data. However, if one piece of text data contains multiple pieces of "emotion word" information (wordings) for each piece of "emotion" information, one piece of "emotion word" information is selected appropriately depending on the content of the text data.

[0028] The "relative word" information is set to a term (phrase) that indicates the target of the "emotion" information (for example, a subject or an object) included in the text data. The "relative word" information is also set in association with attribute information that indicates the "relative word." In this embodiment, n (1 or more) pieces of information are set as the "relative word" information for each piece of text data.

[0029] The category of the content of the text data is set as the "field tag" information. Also, attribute information indicating the "field tag" is set in association with the "field tag" information. In this embodiment, m (1 or more) pieces of information are set as the "field tag" information for each text data.

[0030] In this embodiment, each piece of input information is linked to corresponding text data, and the proposal support device 20 can read the corresponding text data from the input information. Note that the text data linked to the input information may be acquired by the proposal support device 20 together with the input information, or may be provided outside the proposal support device 20. In the latter case, the proposal support device 20 is configured to specify (select) the input information, thereby reading the corresponding text data from outside.

[0031] Here, an example of input information for each text data will be described with reference to Figures 4A and 4B. The examples shown in Figures 4A and 4B are text data acquired by the sentiment analysis device 10 when the answer proposal support system 1 is used to propose regional revitalization for KN City. Figure 4A is an example of a positive post (text data Pc) about "parking lots" (related words) in KN City, and Figure 4B is an example of a negative post (text data Nc) about "parking lots" (related words) in KN City.

[0032] Specifically, examples of positive posts regarding "parking lots" in KN City include the following: "E Castle in KN City is a national historic site. E Castle was renovated in 1583. E Park, where E Castle is located, is fully equipped with parking and toilets. Castle E is surrounded by double and triple moats, and although the area of ​​the enclosure is small, it has a long, narrow building with foundation stones, and roof tiles have been excavated." (See Figure 4A)

[0033] Further examples of negative posts regarding "parking lots" in KN City include the following: "Information about the bird watching event at T Park in KN City has been published in the city's public relations magazine. T Park is the park where I always take walks. Parking is limited, so walking is recommended. T Park is a 10-minute walk from CY Station. (See Figure 4B)

[0034] In the input information (sentiment analysis results) corresponding to the positive post example shown in Figure 4A, the "emotion" information is "positive," and the "emotion word" information is the phrase "complete" (see the bold underlined part in Figure 4A). Also, in the input information corresponding to the positive post example shown in Figure 4A, the "related word" information is the terms "parking lot" and "toilet" (see the part surrounded by dashed lines in Figure 4A), and the "field category" information is "history" and "Japanese history."

[0035] On the other hand, in the input information (sentiment analysis result) corresponding to the negative post example shown in Figure 4B, the "sentiment" information is "negative," and the "sentiment word" information is the phrase "the parking lot has few spaces" (see the bold underlined part in Figure 4B). Also, in the input information corresponding to the negative post example shown in Figure 4B, the "related word" information is the terms "parking lot" and "number of spaces" (see the part surrounded by dashed lines in Figure 4B), and the "field category" information is "railway."

[0036] [Outline of the processing operation of the proposal support device] Next, an overview of the processing operation of the proposal support device 20 will be described. Fig. 5 is a diagram showing an overview of the processing operation of the proposal support device 20. The proposal support device 20 outputs information for supporting the creation of answer proposals (improvement proposals, project proposals, strategy proposals, etc.) according to the purpose of use of the answer proposal support system 1, based on the emotion analysis results of each piece of text data input from the emotion analysis device 10.

[0037] First, the proposal support device 20 stores the input information (emotion analysis results) acquired from the emotion analysis device 10 (P1), as shown in Fig. 5. At this time, the proposal support device 20 may also acquire and store text data linked to each piece of input information.

[0038] Next, the proposal support device 20 creates a graph (hereinafter referred to as a "text association graph") that associates multiple text data (posts) (P2). In this process, the proposal support device 20 first creates a graph (hereinafter referred to as an "emotion graph") that links (connects) information on "related words," "emotion words," and "domain tags" for each piece of text data, starting from information on "related words." Then, the proposal support device 20 combines multiple emotion graphs (relationship information) created for each piece of text data into one to create a text association graph (predetermined information).

[0039] Next, the proposal support device 20 extracts a graph portion of a predetermined range that includes a location where information on both positive and negative "emotion words" is linked to one "relative word" from the text association graph (P3). Note that hereinafter, the graph portion of the predetermined range extracted in P3 will be referred to as an "emotion conflict graph."

[0040] Then, in this embodiment, the extracted emotional conflict graph (specific information) is referenced to consider proposed answers (improvement proposals, project proposals, strategy proposals, etc.) (P4), and the results of the consideration (proposed answers) are output as output information.

[0041] Here, the processing contents of the above P2 to P4 performed by the proposal support device 20 after the input information storage processing (P1 in FIG. 5) will be explained in more detail.

[0042] (1) Text association graph creation process (P2) In the process of creating a text association graph, first, the proposal support device 20 creates an emotion graph for each piece of text data, starting from the node of information on "related words" included in the input information and connecting the nodes of information on "emotion words" and "domain tags" in that order with edges. Then, the proposal support device 20 creates a text association graph that associates multiple emotion graphs by combining nodes of information on common "related words" into a single node between the multiple emotion graphs created for each piece of text data.

[0043] 6A to 6C are diagrams showing examples of creating text association graphs. FIGS. 6A to 6C show examples of creating text association graphs between positive examples of posts about "parking" in KN city shown in FIG. 4A and negative examples of posts about "parking" in KN city shown in FIG. 4B. FIG. 6A is an emotion graph corresponding to the positive examples of posts about "parking" in KN city shown in FIG. 4A, and FIG. 6B is an emotion graph corresponding to the negative examples of posts about "parking" in KN city shown in FIG. 4B. Furthermore, FIG. 6C is a text association graph created by combining the emotion graph shown in FIG. 6A and the emotion graph shown in FIG. 6B.

[0044] In the input information corresponding to the positive post example shown in Figure 4A, the "related word" information is the terms "parking lot" and "toilet," the "emotion word" information is the phrase "complete," and the "domain category" information is "history" and "Japanese history." Therefore, for the post example shown in Figure 4A, an emotion graph is created in which the nodes for "parking lot" and "toilet" are individually connected to the node for "complete" by edges, and the node for "complete" is individually connected to the nodes for "history" and "Japanese history" by edges, as shown in Figure 6A.

[0045] On the other hand, in the input information corresponding to the negative post example shown in Figure 4B, the "related word" information is the terms "parking lot" and "number of spaces," the "emotion word" information is the phrase "the parking lot has few spaces," and the "field category" information is "railway." Therefore, for the post example shown in Figure 4B, an emotion graph is created in which the nodes for "parking lot" and "number of spaces" are individually connected by edges to the node for "the parking lot has few spaces," and the node for "the parking lot has few spaces" is connected by an edge to the node for "railway," as shown in Figure 6B.

[0046] Since the emotion graph shown in FIG. 6A and the emotion graph shown in FIG. 6B share "parking lot" as "related word" information, combining the "parking lot" nodes in both emotion graphs creates the text association graph shown in FIG. 6C. In the text association graph shown in FIG. 6C, the "parking lot" node is individually connected by edges not only to the node for "complete," which is information on the positive "emotion word," but also to the node for "the parking lot has few spaces," which is information on the negative "emotion word." In other words, in the text association graph shown in FIG. 6C, a structural portion is generated in which information on both positive and negative "emotion words" is associated with information on one "related word." Hereinafter, a structural portion having such a relationship will be referred to as an "emotion conflict structural portion."

[0047] Fig. 7 shows an example of the configuration of a final text association graph obtained when a series of processes from creating the emotion graph to creating the text association graph described above is performed on all input information (text data). The example shown in Fig. 7 is an example of the configuration of a text association graph created when the answer proposal support system 1 is used to propose regional revitalization for KN City, for example.

[0048] Note that the configuration example of the text association graph shown in Fig. 7 is a text association graph created taking into consideration text data other than the example posts (text data) shown in Fig. 4A and Fig. 4B. Also, in order to simplify the explanation, Fig. 7 does not show the entire text association graph, but only a portion of the graph near the node for "parking lot" (related word).

[0049] In the text association graph shown in Figure 7, the node for "parking" (a specified evaluation item) is connected by edges not only to the node for "complete," which is information about a positive "sentiment word," and the node for "limited parking spaces," which is information about a negative "sentiment word," but also to the node for "no parking," which is information about a negative "sentiment word." Also, in the text association graph shown in Figure 7, the node for "no parking" is connected by edges to the node for "location" (a related word), and is also individually connected by edges to the nodes for "postal service" and "civil engineering and construction machinery" (field tags).

[0050] In addition, in the text association graph shown in Figure 7, the nodes for "park" and "toilet" (relative word) are individually connected by edges to the node for "beautiful," which is information about a positive "emotion word," and the "beautiful" node is connected by an edge to the node for "travel" (domain tag). Furthermore, in the text association graph shown in Figure 7, the node for "train" (relative word) is connected by an edge to the node for "many trains," which is information about a positive "emotion word," and the node for "many trains" is connected by an edge to the node for "railway" (domain tag).

[0051] Note that the "related words" information for "location," "park," and "train," as well as the "emotional words" information for "no parking lot," "beautiful," and "many trains" shown in Figure 7 are phrases included in text data other than the example posts shown in Figures 4A and 4B. Also, the "domain tags" information for "post," "civil engineering / construction machinery," and "railroad" are categories of content in text data other than the example posts shown in Figures 4A and 4B.

[0052] (2) Extraction of emotional conflict graph (P3) In the emotional conflict graph extraction process, the proposal support device 20 first identifies an emotional conflict structure in the text association graph, in which nodes for both positive and negative "emotion word" information are connected to a node for information on one "related word." Next, the proposal support device 20 identifies a graph structure (hereinafter referred to as a "peripheral structure") surrounding the emotional conflict structure, including nodes for "domain tags" and other "related word" information connected by edges to the nodes for information on each "emotion word" in the emotional conflict structure. The proposal support device 20 then extracts (cuts out) from the text association graph, as an emotional conflict graph, a graph portion consisting of each node included in the identified emotional conflict structure and the peripheral structure, and the edges connecting corresponding nodes.

[0053] FIG. 8 is a diagram showing an example of an emotional conflict graph extracted from the text association graph shown in FIG.

[0054] In the example shown in FIG. 7, the proposal support device 20 first identifies, as an emotional conflict structure, a graph portion consisting of the node for "parking lot" (relative word), the nodes for "complete," "no parking lot," and "few parking spaces" (emotion words), and the edges connecting these. The proposal support device 20 also identifies, as a peripheral structure, a graph portion consisting of the nodes for "toilet" (relative word), "history," and "Japanese history" (domain tag) connected to the node for "complete," the nodes for "location" (relative word), "postal service" and "civil engineering / construction machinery" (domain tag) connected to the node for "no parking lot," the nodes for "number of spaces" (relative word) and "railway" (domain tag) connected to the node for "few parking spaces," and the edges connecting these nodes. The proposal support device 20 then extracts (cuts out) the graph portion within the range of the identified emotional conflict structure and the peripheral structure from the text association graph shown in FIG. 7 as an emotional conflict graph, resulting in an emotional conflict graph for "parking lot" (relative word) shown in FIG. 8.

[0055] Within the range of the text association graph shown in FIG. 7, the only emotional conflict graph extracted is the emotional conflict graph for "parking lot" (related term) shown in FIG. 8. However, if there is a location within the range of the text association graph not shown in FIG. 7 that becomes an emotional conflict graph for a node of information on another "related term," that location is also extracted as an emotional conflict graph for that other "related term." Furthermore, information related to the extracted emotional conflict graph may be displayed as proposal support information on the display unit of the proposal support device 20 (display unit 46 in FIG. 4). Details of this configuration will be described in the second embodiment below.

[0056] (3) Review of answer proposals (P4) In this embodiment, an emotional conflict graph is extracted from the text association graph, and then, based on the information of the extracted emotional conflict graph, a review process is performed to consider answer proposals (improvement proposals, project proposals, strategy proposals, etc.) according to the purpose of use of the answer proposal support system 1.

[0057] In the process of examining answer proposals based on the information in the emotional conflict graph, not only the information in the emotional conflict graph is referenced (scrutinized), but also the contents of the multiple text data linked to the emotional conflict graph, i.e., the multiple text data linked to each of the multiple pieces of input information included in the emotional conflict graph. Then, using the positive opinions about the "related words" included in the emotional conflict graph, an answer proposal is examined to resolve the negative opinions about the "related words."

[0058] In this embodiment, the following three examination methods (A) to (C) can be applied as a method for examining a proposed answer based on the extracted information of the emotional conflict graph.

[0059] (A) Manual methods In the examination method (A), the user of the answer proposal support system 1 examines answer proposals by referring to the emotional conflict graph and the contents of multiple pieces of text data linked to the emotional conflict graph. At this time, the user examines answer proposals by referring not only to this information but also to various related information obtained through surveys (e.g., web surveys, interviews, brainstorming sessions, etc.) according to the purpose of use of the answer proposal support system 1. Therefore, in the examination method (A), examination can be performed not only from the perspective based on the emotional conflict graph, but also from the perspective based on various other related information, making it possible to derive various ideas (answer proposals).

[0060] (B) Methods using generative AI (generative artificial intelligence) technology (Part 1) In method (B), the content information (sentences) of multiple text data linked to the emotional conflict graph are directly input as input parameters to a prompt of the generation AI, the prompt is executed, and a response is obtained from the generation AI. Note that a general-purpose large language model (LLM) or the like can be used as the generation AI, for example.

[0061] Consideration method (B) uses generation AI, so it is easier to conduct the consideration compared to the above-mentioned consideration method (A), which relies on human labor, and it is possible to shorten the time (work time) required to consider answer proposals.

[0062] (C) Method using generative AI technology (part 2) In the study method (C), first, elements related to the task (e.g., phrases such as purpose, related words, targets of related words, emotion words, and common elements) are extracted from sentences in multiple text data linked to the emotion conflict graph. Next, the extracted elements are input as input parameters into a prompt for the generation AI, the prompt is executed, and a response is obtained from the generation AI.

[0063] As a method for extracting elements (wording) from multiple text data, a user may extract elements by carefully examining the contents of the multiple text data, or a predetermined extraction program (algorithm) may be used to automatically extract the necessary elements from multiple text data.

[0064] Because examination method (C) uses a generation AI, it is easier to conduct examinations than the above examination method (A), which uses humans, and the time required to consider answer proposals (work time) can be shortened.In addition, examination method (C) extracts elements from sentences of multiple text data linked to an emotional conflict graph and inputs them into the generation AI, so it can reduce the risk of copyright infringement compared to the above examination method (B), which uses sentences from multiple text data as is.

[0065] Furthermore, text data sentences typically contain content other than elements related to the task, so in the above-mentioned examination method (B), which uses sentences from multiple text data as is, there is a possibility that the response content of the generation AI will change due to excessive information in the text data. In contrast, examination method (C) uses only elements related to the task, so a more accurate response can be obtained. However, because examination method (C) requires processing to extract elements from sentences from multiple text data, in terms of ease of implementation and labor (examination time), examination method (B) is superior to examination method (C).

[0066] In this embodiment (first embodiment), of the above-mentioned examination methods (A) to (C), the examination method (A) is adopted. Therefore, the processing performed by the proposal support device 20 of this embodiment is the processing from the input information storage processing (P1) to the emotion conflict graph extraction processing (P3) shown in FIG. 5. Note that, when the above-mentioned examination method (B) or (C) is adopted, the processing performed by the proposal support device 20 is the processing from the input information storage processing (P1) to the answer proposal examination processing (P4) shown in FIG. 5. A configuration employing the above-mentioned examination method (B) or (C), i.e., a configuration of the proposal support device 20 that examines answer proposals using a generation AI, will be described in the third embodiment below.

[0067] 9 is a diagram showing an example of a proposal (example of output information) obtained as output information by the process of examining a proposed answer in this embodiment. The example shown in FIG. 9 is an example of a proposal (example of output information) obtained when examination is performed based on the emotional conflict graph shown in FIG. 8 and the plurality of text data linked thereto (see FIGS. 4A and 4B). However, in this embodiment, in the examination stage, not only the emotional conflict graph shown in FIG. 8 and the contents of the plurality of text data linked thereto but also various other related information (e.g., map information of KN City) according to the purpose of use (regional revitalization) is used.

[0068] As a result, in the example shown in FIG. 9, for example, the following proposal (proposed answer) is obtained. "Create a circular route starting from a facility with plenty of parking. -Parking at E Park and then walking around the KN City area, including T Park. → Sharing parking lots at public facilities encourages visitors to visit multiple facilities.

[0069] [Functional configuration of the answer suggestion support system] Fig. 10 is a functional block diagram of the answer proposal support system 1 (the emotion analysis device 10 and the proposal support device 20) of this embodiment. Note that Fig. 10 only shows functional units related to the proposal support process for answer proposals (improvement proposals, project proposals, strategy proposals, etc.) according to the purpose of use, which is performed by the emotion analysis device 10 and the proposal support device 20.

[0070] 10 , the emotion analysis device 10 functionally comprises a text data acquisition unit 101, a text data storage unit 102, and a text data emotion determination unit 103. The proposal support device 20 functionally comprises an emotion determination result storage unit 201, a text association graph creation unit 202, a text association graph storage unit 203, an emotion conflict graph extraction unit 204, and an emotion conflict information storage unit 205.

[0071] The text data acquisition unit 101 is connected to a text medium 30 provided outside the answer proposal support system 1 via a network or the like. Within the sentiment analysis device 10, the text data acquisition unit 101 is functionally connected to a text data storage unit 102, and the text data storage unit 102 is connected to a text data emotion determination unit 103. The text data emotion determination unit 103 is functionally connected to an emotion determination result storage unit 201 within the proposal support device 20. Furthermore, within the proposal support device 20, the emotion determination result storage unit 201 is functionally connected to a text association graph creation unit 202, and the text association graph creation unit 202 is connected to a text association graph storage unit 203. Furthermore, within the proposal support device 20, the text association graph storage unit 203 is functionally connected to an emotion conflict graph extraction unit 204, and the emotion conflict graph extraction unit 204 is connected to an emotion conflict information storage unit 205.

[0072] In the sentiment analysis device 10, the text data acquisition unit 101 transmits a query to the text media 30 in response to a request (conditions according to the purpose of use) input by a user or the like, and acquires multiple pieces of text data in response to the transmitted query from the text media 30. The text data storage unit 102 stores the multiple pieces of text data acquired by the text data acquisition unit 101. The text data emotion determination unit 103 reads each piece of text data stored in the text data storage unit 102 and analyzes the emotion (positive / negative / neutral) of the text creator (contributor) regarding the information on "related words" included in each piece of text data. The text data emotion determination unit 103 then outputs the result of the emotion analysis performed on all of the text data stored in the text data storage unit 102 as an emotion determination result to the emotion determination result storage unit 201 in the proposal support device 20.

[0073] The processing control of the text data acquisition unit 101 and text data emotion determination unit 103 described above is controlled by software, that is, by a CPU (processor) interpreting and executing programs for realizing their respective functions. The text data storage unit 102 is a memory unit for storing various types of information that are used, referenced, or updated. Therefore, the text data acquisition unit 101 and text data emotion determination unit 103 within the emotion analysis device 10 are functionally included in the CPU of the emotion analysis device 10 (CPU 41 in FIG. 2). The text data storage unit 102 is functionally included in the RAM of the emotion analysis device 10 (RAM 43 in FIG. 2).

[0074] In the proposal support device 20, the emotion determination result storage unit 201 stores the emotion determination results (input information) input from the text data emotion determination unit 103. Note that the input information input from the text data emotion determination unit 103 is data that includes, for each piece of text data, a set of information on "emotion," "emotion word," "related word," and "domain tag," as explained in Figures 3, 4A, and 4B.

[0075] The text association graph creation unit 202 (predetermined information creation unit) first reads the emotion determination results stored in the emotion determination result storage unit 201, i.e., the input information created for each text data, and creates an emotion graph (see FIGS. 6A and 6B) for each text data. Next, the text association graph creation unit 202 creates a text association graph (see FIGS. 6C and 7) based on the emotion graph created for each text data. At this time, the text association graph creation unit 202 creates the text association graph by combining nodes of information on common "related words" into a single node among the multiple emotion graphs created, as described above.

[0076] The text association graph storage unit 203 stores the text association graph created by the text association graph creation unit 202. At this time, the text association graph storage unit 203 may also store an emotion graph for each piece of text data created by the text association graph creation unit 202.

[0077] The emotional conflict graph extraction unit 204 (specific information extraction unit) first reads the text association graph stored in the text association graph storage unit 203 and identifies an emotional conflict structure within the text association graph. Next, the emotional conflict graph extraction unit 204 cuts out (extracts) an emotional conflict graph (see FIG. 8) from the text association graph (see FIG. 7) based on the identified emotional conflict structure. At this time, as described above, the emotional conflict graph extraction unit 204 cuts out, from the text association graph, the graph area of ​​the identified emotional conflict structure and its surrounding structure as an emotional conflict graph. Furthermore, at this time, if there are multiple emotional conflict structures within the text association graph, the emotional conflict graph extraction unit 204 extracts an emotional conflict graph for each emotional conflict structure.

[0078] The emotion conflict graph extraction unit 204 also creates a list of "related word" information corresponding to each extracted emotion conflict graph (hereinafter referred to as an "emotion conflict list"). At this time, the information on each "related word" registered (defined) in the emotion conflict list (list of predetermined evaluation items) is linked to the corresponding emotion conflict graph. In this case, when a user or the like selects information on a predetermined "related word" in the emotion conflict list, the emotion conflict graph extraction unit 204 can read the corresponding predetermined emotion conflict graph.

[0079] The emotional conflict information storage unit 205 stores one or more emotional conflict graphs extracted by the emotional conflict graph extraction unit 204, and an emotional conflict list created by the emotional conflict graph extraction unit 204. In this embodiment, based on the one or more emotional conflict graphs and the emotional conflict list stored in the emotional conflict graph extraction unit 204, a user of the answer proposal support system 1 or the like considers answer proposals according to the purpose (issue).

[0080] The processing control of the above-mentioned text association graph creation unit 202 and emotion conflict graph extraction unit 204 is controlled by software, i.e., by a CPU (processor) interpreting and executing programs for realizing each function. The emotion determination result storage unit 201, text association graph storage unit 203, and emotion conflict information storage unit 205 are memory units for storing various information that is used, referenced, or updated. Therefore, the text association graph creation unit 202 and emotion conflict graph extraction unit 204 in the proposal support device 20 are functionally included in the CPU of the proposal support device 20 (CPU 41 in FIG. 2). The emotion determination result storage unit 201, text association graph storage unit 203, and emotion conflict information storage unit 205 are functionally included in the RAM of the proposal support device 20 (RAM 43 in FIG. 2).

[0081] [Processing flow of the answer suggestion support system] Next, the specific contents of the various processes executed by the sentiment analysis device 10 and the proposal support device 20 (answer proposal support system 1) will be described with reference to the drawings. Note that the control of the various processes by the sentiment analysis device 10 described below is executed on software by each function processing unit (see FIG. 10) included in the CPU (CPU 41 in FIG. 2) of the sentiment analysis device 10. Also, the control of the various processes by the proposal support device 20 described below is executed on software by each function processing unit (see FIG. 10) included in the CPU (CPU 41 in FIG. 2) of the proposal support device 20.

[0082] (Proposal support process flow) 11 is a flowchart showing the steps of a process for supporting the proposal of an answer proposal according to a purpose of use (issue), which is performed by the answer proposal support system 1 (the emotion analysis device 10 and the proposal support device 20) of this embodiment. Note that the proposal support process shown in FIG. 11 is started, for example, when a user or the like performs an input operation of a request (conditions according to the purpose of use) to the answer proposal support system 1.

[0083] First, the text data acquisition unit 101 of the emotion analysis device 10 performs a query transmission process and a corresponding text data acquisition process (S1). In this process, the text data acquisition unit 101 transmits a query corresponding to a request input by a user or the like to the text medium 30, and acquires one or more pieces of text data corresponding to the transmitted query from the text medium 30. In this process, the text data acquisition unit 101 also stores (memorizes) the acquired text data in the text data storage unit 102.

[0084] Next, the text data emotion determination unit 103 of the emotion analysis device 10 performs emotion determination processing on the text data (S2). In this processing, the text data emotion determination unit 103 reads text data from the text data storage unit 102 and analyzes the emotion (positive / negative / neutral) of the text creator (contributor) regarding the information on "related words" included in the text data. In this processing, the text data emotion determination unit 103 also creates data that combines information on "emotions," "emotion words," "related words," and "category tags" obtained by analyzing the read text data. The text data emotion determination unit 103 performs the above-described "emotion" analysis processing and data set creation processing on all text data stored in the text data storage unit 102. In this processing, the text data emotion determination unit 103 also outputs data that combines information on "emotions," "emotion words," "related words," and "category tags" created for each piece of text data as emotion determination results to the emotion determination result storage unit 201 in the proposal support device 20.

[0085] Next, the emotion determination result storage unit 201 of the proposal support device 20 performs a storage process for the emotion determination result (input information) input from the text data emotion determination unit 103 (S3).

[0086] Next, the text association graph creation unit 202 of the proposal support device 20 performs a process of creating a text association graph (S4). In this process, the text association graph creation unit 202 first reads the emotion determination results stored in the emotion determination result storage unit 201, i.e., the input information for each text data, and creates an emotion graph (see FIGS. 6A and 6B) for each text data. The text association graph creation unit 202 then compiles the multiple emotion graphs created for the multiple text data respectively to create a text association graph (see FIGS. 6C and 7). In this process, the text association graph creation unit 202 also stores the created text association graph in the text association graph storage unit 203.

[0087] Next, the emotion conflict graph extraction unit 204 of the proposal support device 20 performs an emotion conflict graph extraction process (S5). In this process, the emotion conflict graph extraction unit 204 first reads the text association graph (see FIG. 7) stored in the text association graph storage unit 203 and identifies the emotion conflict structure part in the text association graph. In this process, the emotion conflict graph extraction unit 204 also cuts out (extracts) an emotion conflict graph (see FIG. 8) from the text association graph based on the identification result of the emotion conflict structure part. Furthermore, in this process, the emotion conflict graph extraction unit 204 also creates an emotion conflict list. Details of the emotion conflict graph extraction process will be described later with reference to FIG. 12.

[0088] Then, after the processing of S5, in this embodiment, the proposal support device 20 ends the proposal support processing for the answer proposal. Note that, in this embodiment, after the proposal support processing ends, the answer proposal is examined by the examination method (A) described above, and the answer proposal is proposed (output). Specifically, a user or the like of the answer proposal support system 1 carefully examines the contents of the emotion conflict graph and emotion conflict list obtained in the processing of S5, the plurality of text data linked to the emotion conflict graph, and various other related information according to the purpose of use, examines the answer proposal, and outputs the derived answer proposal (see FIG. 9).

[0089] Note that the information such as the emotional conflict graph and corresponding text data extracted in the process of S5, and / or the derived answer proposals may be printed out, for example, or may be displayed on a display unit (display unit 46 in FIG. 2) of the proposal support device 20. The specific configuration of the latter will be described later in the second embodiment and modified example 2.

[0090] (Process flow of emotional conflict graph extraction process) Next, the process of extracting an emotion conflict graph that is called and executed in the process of S5 in the proposal support process (see Fig. 11) will be described with reference to Fig. 12. Fig. 12 is a flowchart showing the steps of the process of extracting an emotion conflict graph.

[0091] First, the emotion conflict graph extraction unit 204 of the proposal support device 20 performs processing to identify the emotion conflict part (S11).

[0092] In the process of S11, the emotion conflict graph extraction unit 204 reads the text association graph (see FIG. 7) stored in the text association graph storage unit 203 and identifies the location of the emotion conflict structure in the text association graph. Specifically, the emotion conflict graph extraction unit 204 identifies a graph portion in the text association graph where nodes of information on both positive "emotion words" and negative "emotion words" are individually connected by edges to a node of information on one "related word." Furthermore, in this process, the emotion conflict graph extraction unit 204 creates a list of information on "related words" corresponding to the identified emotion conflict structure portion, i.e., an emotion conflict list. Note that the process of identifying the emotion conflict location will be described in detail later with reference to FIG. 13.

[0093] Next, the emotion conflict graph extraction unit 204 performs an emotion conflict graph extraction process (S12). In this process, based on the result of identifying the emotion conflict part in the process of S11, the emotion conflict graph extraction unit 204 cuts out (extracts) an emotion conflict graph (see FIG. 8) corresponding to the identification result from the text association graph. Details of the emotion conflict graph extraction process will be explained later with reference to FIG. 14.

[0094] After the process of S12, the proposal support device 20 ends the process of extracting the emotion conflict graph, and returns the process to the proposal support process (FIG. 11) that is the caller of this process.

[0095] (Process flow for identifying emotional conflicts) Next, the process of identifying an emotion conflict part that is called and executed in the process of S11 in the process of extracting an emotion conflict graph (see FIG. 12) will be described with reference to Fig. 13. Fig. 13 is a flowchart showing the steps of the process of identifying an emotion conflict part.

[0096] First, the emotion conflict graph extraction unit 204 of the proposal support device 20 reads the text association graph (see FIG. 7) stored in the text association graph storage unit 203 (S21). Next, the emotion conflict graph extraction unit 204 initializes the emotion conflict list stored in the text association graph storage unit 203 (S22). In this process, the emotion conflict graph extraction unit 204 deletes all information on "related words" registered in the emotion conflict list at the current processing point.

[0097] Next, the emotion conflict graph extraction unit 204 repeatedly executes the following processes of S23 to S26 for all the "related term" information (hereinafter referred to as "related term R") included in the text association graph.

[0098] The emotion conflict graph extraction unit 204 determines whether the number of "emotion word" nodes linked to a predetermined related word R included in the text association graph is N or more (S23). Note that the threshold N for the number of "emotion word" nodes linked to one related word R used in this determination process is the threshold for the number of all "emotion word" nodes (positive, negative, and neutral) linked to the nodes of one related word R. Furthermore, the threshold N for the number of "emotion word" nodes linked to one related word R is a preset value and can be set to any value equal to or greater than 1.

[0099] In S23, if the emotion conflict graph extraction unit 204 determines that the number of nodes of the “emotion word” linked to the predetermined related word R is less than N (if S23 returns a NO determination), the emotion conflict graph extraction unit 204 does not perform the processes of S24 to S26 described below.

[0100] If the determination in S23 is NO and there are other related words R remaining for which the processing of S23 has not been executed, the emotion conflict graph extraction unit 204 returns the processing to S23 after the determination processing of S23 and performs the processing from S23 onwards for the other related words R. Also, if the determination in S23 is NO and there are no other related words R remaining for which the processing of S23 has not been executed, the emotion conflict graph extraction unit 204 ends the processing for identifying emotion conflict parts and returns the processing to the emotion conflict graph extraction processing ( FIG. 12 ) that called this processing.

[0101] On the other hand, if the emotion conflict graph extraction unit 204 determines in S23 that the number of nodes of the "emotion word" linked to the predetermined related word R is N or more (if the determination in S23 is YES), the emotion conflict graph extraction unit 204 determines whether the number of nodes of the positive "emotion word" linked to the predetermined related word R is Np or more (S24).

[0102] In S24, if the emotion conflict graph extraction unit 204 determines that the number of nodes of positive "emotion words" linked to the predetermined related word R is less than Np (if S24 returns a NO determination), the emotion conflict graph extraction unit 204 does not perform the processes of S25 to S26 described below.

[0103] If the determination in S24 is NO and there are other related words R remaining for which the processing of S23 has not been executed, the emotion conflict graph extraction unit 204 returns the processing to S23 after the determination processing of S24 and performs the processing from S23 onwards for the other related words R. If the determination in S24 is NO and there are no other related words R remaining for which the processing of S23 has not been executed, the emotion conflict graph extraction unit 204 ends the processing for identifying emotion conflict parts and returns the processing to the emotion conflict graph extraction processing ( FIG. 12 ) that called this processing.

[0104] On the other hand, if the emotion conflict graph extraction unit 204 determines in S24 that the number of nodes of positive "emotion words" linked to the predetermined related word R is Np or more (if the determination in S24 is YES), the emotion conflict graph extraction unit 204 determines whether the number of nodes of negative "emotion words" linked to the predetermined related word R is Nn or more (S25).

[0105] In S25, if the emotion conflict graph extraction unit 204 determines that the number of nodes of negative "emotion words" linked to the predetermined related word R is less than Nn (if S25 is determined to be NO), the emotion conflict graph extraction unit 204 does not perform the processing of S26 described below.

[0106] If the determination in S25 is NO and there are other related words R remaining for which the processing of S23 has not been executed, the emotion conflict graph extraction unit 204 returns the processing to the processing of S23 after the determination processing of S25, and performs the processing from S23 onwards for the other related words R. Also, if the determination in S25 is NO and there are no other related words R remaining for which the processing of S23 has not been executed, the emotion conflict graph extraction unit 204 ends the processing for identifying emotion conflict parts, and returns the processing to the emotion conflict graph extraction processing ( FIG. 12 ) that called this processing.

[0107] The threshold value Np of the number of nodes of positive "emotion words" linked to one related word R used in the determination process of S24 is a preset value and can be set to any value greater than or equal to 1 and less than or equal to N. The threshold value Nn of the number of nodes of negative "emotion words" linked to one related word R used in the determination process of S25 is also a preset value and can be set to any value greater than or equal to 1 and less than or equal to N. However, the threshold values ​​Np and Nn are set so that the sum of the threshold value Np of the number of nodes of positive "emotion words" and the threshold value Nn of the number of nodes of negative "emotion words" is less than or equal to N.

[0108] On the other hand, if the emotion conflict graph extraction unit 204 determines in S25 that the number of negative "emotion word" nodes linked to the predetermined related term R is Nn or more (if the determination in S25 is YES), the emotion conflict graph extraction unit 204 registers the predetermined related term R currently being processed in an emotion conflict list (S26). Also, in this process, the emotion conflict graph extraction unit 204 stores the emotion conflict list updated by the registration process in the text association graph storage unit 203.

[0109] Then, after the processing of S26, if there are other related terms R remaining for which the processing of S23 described above has not been executed, the emotion conflict graph extraction unit 204 returns the processing to the processing of S23 and executes the processing from S23 onwards for the other related terms R. Furthermore, if there are no other related terms R remaining for which the processing of S23 described above has not been executed after the processing of S26, the emotion conflict graph extraction unit 204 ends the processing for identifying emotion conflict parts and returns the processing to the emotion conflict graph extraction processing ( FIG. 12 ) that called this processing.

[0110] (Process flow for extracting emotional conflict graphs) Next, the emotion conflict graph cut-out process that is called and executed in the process of S12 in the emotion conflict graph extraction process (see FIG. 12) will be described with reference to Fig. 14. Fig. 14 is a flowchart showing the steps of the emotion conflict graph cut-out process.

[0111] First, the emotion conflict graph extraction unit 204 of the proposal support device 20 reads the emotion conflict list stored in the text association graph storage unit 203 (S31).

[0112] Next, the emotion conflict graph extraction unit 204 repeatedly executes the following processes of S32 and S33 for all related terms R registered in the emotion conflict list.

[0113] The emotion conflict graph extraction unit 204 cuts out an emotion conflict graph from the text association graph, starting from a specific related word R registered in the emotion conflict list (S32). In this process, the emotion conflict graph extraction unit 204 cuts out (extracts) from the text association graph an emotion conflict structure part starting from the node of the specific related word R, and the regions (peripheral structure parts) of the nodes of "related words" and "domain tags" linked to the node of the "emotion word" included in the emotion conflict structure part, as an emotion conflict graph. Note that the text association graph referenced in this process is the text association graph that has been read in by the process of S21 in the process of identifying the emotion conflict part described above (see FIG. 13).

[0114] Next, the emotion conflict graph extraction unit 204 stores the extracted emotion conflict graph in the text association graph storage unit 203 (S33).

[0115] Then, after the processing of S33, if there are other related terms R remaining in the emotion conflict list for which the processing of S31 described above has not been performed, the emotion conflict graph extraction unit 204 returns the processing to the processing of S31 and performs the processing from S31 onwards for the other related terms R. On the other hand, if there are no other related terms R remaining in the emotion conflict list for which the processing of S31 described above has not been performed after the processing of S32, the emotion conflict graph extraction unit 204 ends the emotion conflict graph cut-out processing and returns the processing to the emotion conflict graph extraction processing ( FIG. 12 ) that called this processing.

[0116] [Various effects] As described above, in this embodiment, multiple emotion graphs that reflect the emotions of the text creators (contributors) are created from multiple text data acquired according to the purpose of use (assignment), and a text association graph is created that associates these multiple emotion graphs. Then, from the text association graph, graph sections (emotion conflict graphs) where both positive and negative evaluations are given to evaluation items (related words) related to the assignment are identified and extracted.

[0117] In other words, the emotional conflict graph extracted in this embodiment represents the evaluation results of multiple text creators (contributors) from different perspectives on evaluation items (related words) related to the issue. In this embodiment, positive opinions on the evaluation items (related words) included in the emotional conflict graph can be used to consider response proposals to resolve negative opinions on the evaluation items. Therefore, in this embodiment, evaluation results from various perspectives on specific items (evaluation items) related to the purpose of use (issue) can be extracted, and response proposals (improvement proposals, project proposals, strategy proposals, etc.) can be proposed based on the extracted evaluation results.

[0118] Furthermore, in this embodiment, it is possible to consider not only from the perspective of the emotional conflict graph, but also from the perspective of various other related information regarding the purpose of use, so that various ideas (proposed answers) can be derived.

[0119] 2. Second embodiment Next, an answer suggestion support system according to a second embodiment of the present invention will be described. Note that, in the following description of this embodiment, devices and components having the same functions as those in the first embodiment will be denoted by the same reference numerals.

[0120] In this embodiment, a configuration will be described in which a new function has been added to the proposal support device 20 of the first embodiment, in which information on the created emotional conflict graph and emotional conflict list (hereinafter referred to as "emotional conflict information") is displayed on the display unit as proposal support information.

[0121] The configurations of the answer proposal support system 2 of this embodiment, and the emotion analysis device 10 and proposal support device 21 constituting it, are similar to the configurations of the answer proposal support system 1 of the first embodiment described above with reference to Figures 1 and 2, and the emotion analysis device 10 and proposal support device 20 constituting it. Therefore, illustrations and descriptions of the configuration of the answer proposal support system 2 and the hardware configurations of the emotion analysis device 10 and proposal support device 21 will be omitted here.

[0122] Furthermore, in the answer proposal support system 2 of this embodiment, the content of the emotion determination result output from the emotion analysis device 10 to the proposal support device 21, i.e., the configuration of the input information of the proposal support device 21, is also the same as that of the first embodiment described above with reference to Fig. 3. Therefore, illustration and description of the configuration of the input information of the proposal support device 21 will be omitted here.

[0123] Furthermore, the processing operations of the emotion analysis device 10 of this embodiment and the contents from the input information acquisition process to the emotion conflict graph extraction process in the proposal support device 21 are the same as those of the first embodiment described above with reference to Figures 4 to 8 and 11 to 14. Therefore, here, explanations of the processing operations of the emotion analysis device 10 and the contents from the input information acquisition process to the emotion conflict graph extraction process in the proposal support device 21 will be omitted.

[0124] [Functional configuration of the answer suggestion support system] Fig. 15 is a functional block configuration diagram of the answer proposal support system 2 (the emotion analysis device 10 and the proposal support device 21) of this embodiment. Note that Fig. 15 only shows functional units related to the proposal support process for answer proposals (improvement proposals, project proposals, strategy proposals, etc.) according to the purpose of use, which is performed by the emotion analysis device 10 and the proposal support device 21.

[0125] 10, the functional block configuration of the proposal support device 21 of this embodiment is the same as the functional block configuration of the proposal support device 20 of the first embodiment, except that an emotional conflict information display control unit 216 is newly added. Note that the other functional components in the proposal support device 21 of this embodiment and the functional components of the emotion analysis device 10 have the same functions as those described in the first embodiment, and therefore a description of these functional components will be omitted here.

[0126] The emotional conflict information display control unit 216 is functionally connected to the emotional conflict information storage unit 205. Although not shown here, the emotional conflict information display control unit 216 is connected to the display unit (display unit 46 in FIG. 2) of the proposal support device 21 and controls the display of emotional conflict information on the display unit. Therefore, the emotional conflict information display control unit 216 is functionally included in the CPU (CPU 41 in FIG. 2) of the proposal support device 21.

[0127] Specifically, the emotional conflict information display control unit 216 (display control unit) reads the emotional conflict list stored in the emotional conflict information storage unit 205, and displays the emotional conflict list on the display screen of the display unit (display unit 46 in FIG. 2) of the proposal support device 21. Furthermore, when one piece of "related word" information in the emotional conflict list displayed on the display unit is selected by a predetermined operation by a user or the like, the emotional conflict information display control unit 216 displays an emotional conflict graph linked to (corresponding to) the selected "related word" on the display unit. Note that the predetermined operation for selecting the "related word" information displayed on the display unit can be, for example, a selection operation using a mouse or the like, or an operation such as touching the display screen.

[0128] [Example of display of emotional conflict information] Fig. 16 is a diagram showing an example of emotional conflict information (emotional conflict list and emotional conflict graph) displayed on the display unit of the proposal support device 21 of this embodiment. The example shown in Fig. 16 is an example of a display mode when the emotional conflict list and emotional conflict graph stored in the emotional conflict information storage unit 205 are displayed on the display unit in the example of extraction (cutout) of the emotional conflict graph described in Figs. 6 to 8.

[0129] In this embodiment, a display screen 50 of the display unit of the proposal support device 21 is provided with a display area 51 that displays an emotion conflict list and a display area 52 that displays an emotion conflict graph, as shown in Fig. 16. In the example shown in Fig. 16, the display area 51 and the display area 52 are arranged in this order in the direction from left to right (horizontal direction) within the display screen 50.

[0130] In the display area 51, various pieces of "related word" information (for example, "parking lot," "genki," "KN store," "CY station," etc.) registered in the emotional conflict list are displayed in a line from top to bottom (vertical direction). The order in which the "related word" information is listed in the display area 51 can be set arbitrarily, and may be, for example, the order in which the emotional conflict points are identified or alphabetical order. Furthermore, for example, the order in which the "related word" information is listed in the display area 51 may be in descending order of priority according to the purpose of use (issue). The configuration in this case will be described in the fourth embodiment below.

[0131] Displayed in the display area 52 is an emotional conflict graph linked to information on a predetermined "related word" selected by a user or the like from the emotional conflict list displayed in the display area 51. In the example shown in Fig. 16, this is an emotional conflict graph that is displayed when a user or the like selects "parking lot" as information on the "related word" from the emotional conflict list displayed in the display area 51. In this case, an emotional conflict graph similar to the emotional conflict graph starting from "parking lot" (related word) described in Fig. 8 is displayed in the display area 52.

[0132] In the example shown in Fig. 8, in order to make it easier for users to recognize (identify) the emotional conflict structure in the emotional conflict graph, the line type of the negative "emotion word" node and the edge connecting the negative "emotion word" node with the "relative word" node are shown with dashed lines. In the example shown in Fig. 8, the line types of the other nodes and edges are shown with solid lines.

[0133] The display mode of nodes and edges in the emotion conflict graph is not limited to the example shown in Fig. 8. For example, the "relative word" and "domain tag" nodes and edges connected to the negative "emotion word" node may be shown with dashed lines. Also, for example, the "relative word" and "domain tag" nodes connected to both the positive "emotion word" node and the negative "emotion word" node may be shown with a line type other than solid and dashed lines.

[0134] Furthermore, to facilitate recognition (identification) of the emotional conflict structure in the text association graph, for example, the line color of the negative "emotion word" node and the edge connecting it to the "relative word" node may be different from the line color of the other nodes and edges. In this case, for example, the line color of the "relative word" and "category tag" nodes and edges connected to the negative "emotion word" node may also be the same as the line color of the negative "emotion word" node. Furthermore, for example, the line color of the "relative word" and "category tag" nodes connected to both the positive "emotion word" node and the negative "emotion word" node may be displayed in a line color other than the line color of the positive "emotion word" node and the negative "emotion word" node.

[0135] Furthermore, to make it easier to recognize (identify) the emotional conflict structure in the text association graph, for example, the line type and color of the node of information on negative "emotion words" and the edge connecting it to the node of information on "related words" may be made different from the line type and color of the other nodes and edges.

[0136] [Processing flow of the answer suggestion support system] Next, a description will be given of the processing flow of the process for supporting the proposal of an answer plan according to the purpose of use (problem), which is performed by the answer proposal support system 2 (sentiment analysis device 10 and proposal support device 21) of this embodiment. FIG. 17 is a flowchart showing the procedure of the process for supporting the proposal of an answer plan according to the purpose of use (problem), which is performed by the answer proposal support system 2 of this embodiment. Note that, also in this embodiment, the control of various processes by the sentiment analysis device 10 and the proposal support device 21 is executed on software by each function processing unit (see FIG. 15) included in the CPU of each device (CPU 41 in FIG. 2).

[0137] The processes of S41 to S45 in the proposal support process of this embodiment shown in FIG. 17 are similar to the processes of S1 to S5 in the proposal support process of the first embodiment (see FIG. 11), and therefore, description of these processes will be omitted here.

[0138] After the process of S45, the emotion conflict information display control unit 216 of the proposal support device 21 performs a display process of the emotion conflict information (S46).

[0139] In this process, the emotion conflict information display control unit 216 reads the emotion conflict list created (updated) in the process of S45 from the emotion conflict information storage unit 205, and controls the display unit to display the read emotion conflict list. Also, in this process, the emotion conflict information display control unit 216 reads, from the emotion conflict information storage unit 205, an emotion conflict graph linked to information on a predetermined "related term" selected by a user or the like from information on various "related terms" registered in the emotion conflict list. Then, the emotion conflict information display control unit 216 controls the display unit to display the read emotion conflict graph. At this time, the emotion conflict information display control unit 216 displays the read emotion conflict list and emotion conflict graph on the display screen 50 of the display unit, for example, in the display format shown in FIG.

[0140] Note that, when the display of the emotional conflict information is started, and no "related word" information has been selected by the user or the like from the various "related word" information registered in the emotional conflict list, for example, the emotional conflict graph may not be displayed on the display screen 50. Also, when no "related word" information has been selected, the emotional conflict graph linked to a specific "related word" in the emotional conflict list, for example, the "related word" information registered at the top of the emotional conflict list, may be displayed on the display screen 50.

[0141] Then, after processing S46, the proposal support device 21 ends the proposal support process of this embodiment. Note that in this embodiment, after processing S46, the user or the like of the answer proposal support system 2 carefully examines the emotional conflict information (emotion conflict graph and emotional conflict list) displayed on the display unit, the multiple pieces of text data linked to the emotional conflict graph, and various other related information according to the purpose of use, considers answer proposals, and outputs the derived answer proposals.

[0142] [Various effects] As described above, the proposal support device 21 of this embodiment has the same functional components as the proposal support device 20 of the first embodiment. Therefore, this embodiment can achieve the same effects as the first embodiment.

[0143] Furthermore, this embodiment has a function for displaying emotional conflict information (an emotional conflict graph and an emotional conflict list). Therefore, evaluation items (related words) that have been evaluated both positively and negatively, and the contents of those evaluations, can be easily checked visually, which makes it easier to consider proposed answers (improvement proposals, project proposals, strategy proposals, etc.) and reduces the time required for consideration.

[0144] 3. Third embodiment Next, an answer suggestion support system according to a third embodiment of the present invention will be described. In the following, in this embodiment, devices and components having the same functions as those in the first embodiment will be described using the same reference numerals.

[0145] In this embodiment, a configuration will be described in which a generation AI is used as a method for examining answer proposals (improvement proposals, project proposals, strategy proposals, etc.) based on an extracted emotion conflict graph and information linked thereto in the proposal support device 20 of the first embodiment. In this embodiment, the above-mentioned examination method (B) or examination method (C) is adopted as a method for using a generation AI. Note that examination method (B) is a method in which multiple pieces of text data linked to an emotion conflict graph, i.e., sentences from two or more pieces of text data corresponding to the emotion conflict graph, are input directly into the generation AI to obtain answer proposals. Furthermore, examination method (C) is a method in which elements (information related to input parameters) are extracted from sentences from multiple pieces of text data linked to the emotion conflict graph and input them into the generation AI to obtain answer proposals.

[0146] As the generation AI that can be employed in this embodiment, for example, a general-purpose LLM or an interactive AI service can be used, but a generation AI developed specifically for the answer proposal support system of the present invention may also be used. Furthermore, in this embodiment, an example is described in which the prompt execution function unit (program) is provided within the proposal support device 22, but the prompt execution function unit may also be provided outside the proposal support device 22 (for example, in the cloud, etc.). However, from the perspective of, for example, information security and efficiency of calculation processing, it is advantageous to provide the prompt execution function unit within the proposal support device 20, as in this embodiment.

[0147] The configurations of the answer proposal support system 3 of this embodiment, and the sentiment analysis device 10 and proposal support device 22 that constitute it, are similar to the configurations of the answer proposal support system 1 of the first embodiment described above with reference to Figures 1 and 2, and the sentiment analysis device 10 and proposal support device 20 that constitute it. Therefore, illustrations and descriptions of the configuration of the answer proposal support system 3 and the hardware configurations of the sentiment analysis device 10 and proposal support device 22 will be omitted here.

[0148] Furthermore, in the answer proposal support system 3 of this embodiment, the configuration of the input information of the proposal support device 22 is also similar to that of the first embodiment described above with reference to Fig. 3. Therefore, illustration and description of the configuration of the input information of the proposal support device 22 will be omitted here.

[0149] Furthermore, the processing operations of the emotion analysis device 10 of this embodiment and the contents from the input information acquisition process to the emotion conflict graph extraction process in the proposal support device 22 are the same as those of the first embodiment described above with reference to Figures 4 to 8 and 11 to 14. Therefore, here, explanations of the processing operations of the emotion analysis device 10 and the contents from the input information acquisition process to the emotion conflict graph extraction process in the proposal support device 22 will be omitted.

[0150] [Functional configuration of the answer suggestion support system] Fig. 18 is a functional block configuration diagram of the answer proposal support system 3 (the emotion analysis device 10 and the proposal support device 22) of this embodiment. Note that Fig. 18 only shows functional units related to the proposal support process for answer proposals (improvement proposals, project proposals, strategy proposals, etc.) according to the purpose of use, which is performed by the emotion analysis device 10 and the proposal support device 22.

[0151] 10, the functional block configuration of the proposal support device 22 of this embodiment is the same as the functional block configuration of the proposal support device 20 of the first embodiment, except that a prompt creation unit 226, a template storage unit 227, a prompt storage unit 228, a proposal examination unit 229, and a proposal storage unit 230 are newly added. Note that the other functional components in the proposal support device 22 of this embodiment and the functional components of the emotion analysis device 10 have the same functions as those described in the first embodiment, and therefore descriptions of these functional components will be omitted here.

[0152] Within the proposal support device 22, the prompt creation unit 226 is connected to the emotional conflict information storage unit 205, the template storage unit 227, and the prompt storage unit 228. Furthermore, the prompt storage unit 228 is connected to the proposal examination unit 229, which is connected to the proposal storage unit 230. Note that the prompt creation unit 226 and the proposal examination unit 229 are functionally included in the CPU (CPU 41 in FIG. 2) of the proposal support device 22, and the prompt storage unit 228 and the proposal storage unit 230 are functionally included in the RAM (RAM 43 in FIG. 2) of the proposal support device 22. Furthermore, the template storage unit 227 is functionally included in the ROM (ROM 42 in FIG. 2) or RAM of the proposal support device 22.

[0153] The prompt creation unit 226 reads emotional conflict information (emotion conflict graph and emotional conflict list) from the emotional conflict information storage unit 205, and also reads a prompt template that serves as an input instruction to the generation AI from the template storage unit 227. Furthermore, based on the emotional conflict information (emotion conflict graph and emotional conflict list), the prompt creation unit 226 inputs predetermined data into the input parameters of the prompt template to create a prompt to be executed by the generation AI. The prompt creation unit 226 then stores the created prompt in the prompt storage unit 228.

[0154] In this embodiment, the prompt template differs depending on the type of review method for answer plans using the generation AI (the above-mentioned review method (B) or (C)). Therefore, the prompt creation operation in the prompt creation unit 226 also differs depending on the type of review method for answer plans using the generation AI. Furthermore, in this embodiment, the input operation of input parameters into the prompt template in the prompt creation unit 226 requires some input operations by the user, etc. (such as an input operation for the "analysis purpose" and an operation for selecting "related terms," ​​which will be described later). Therefore, in this case, the prompt creation unit 226 controls the display unit to display information prompting the input operation, an input form for the input parameters, etc. on the display screen 50. Specific examples of prompt templates and the prompt creation operation in the prompt creation unit 226 depending on the type of review method for answer plans using the generation AI will be described later with reference to the drawings.

[0155] The prompt storage unit 228 stores the prompts created by the prompt creation unit 226 .

[0156] The proposal examination unit 229 (answer proposal output unit) reads a prompt from the prompt storage unit 228, inputs the prompt to the generation AI to execute the prompt, and obtains a response from the generation AI. The proposal examination unit 229 may also control the display unit to display the response from the generation AI on the display screen 50.

[0157] In this embodiment, for example, a response may be output from the generation AI as a proposed answer. Furthermore, for example, a user or the like may comprehensively consider a proposed answer by taking into account not only the response obtained using the generation AI but also multiple text data (posts) referenced when creating the prompt and the results of surveys of various related information (e.g., web surveys, interviews, brainstorming sessions, etc.) according to the purpose of use of the answer proposal support system 3. In the latter case, the user or the like operates the operation unit (operation unit 45 in FIG. 2 ) of the proposal support device 22 to input a comprehensive proposed answer to the proposal support device 22. Therefore, in the latter case, after obtaining a response from the generation AI, the proposal examination unit 229 waits until a comprehensive proposed answer is input from the user or the like, and once the comprehensive proposed answer is input, the proposal examination unit 229 stores the response from the generation AI as a proposed answer in the proposal storage unit 230. In the former case, the proposal examination unit 229 stores the response from the generation AI as a proposed answer in the proposal storage unit 230.

[0158] Furthermore, the proposal storage unit 230 stores the answer plan acquired by the proposal examination unit 229. Note that if the answer plan is the comprehensive answer plan described above, the proposal storage unit 230 may store a response from the generation AI in addition to the answer plan.

[0159] [Examples of prompt creation and AI responses] Next, an example of the prompt creation operation in the prompt creation unit 226 and a response from the generation AI according to the methods (B) and (C) of examining answer proposals using the generation AI will be described.

[0160] Here, as in the first and second embodiments, an example of prompt creation will be described in which the answer proposal support system 3 is used for the purpose of proposing regional revitalization for KN City, and "parking lot" is selected as the evaluation item (related word) for problem solving. Therefore, in the prompt creation operation in the prompt creation unit 226, reference is made to, for example, the text data of the positive post example shown in Fig. 4A and the text data of the negative post example shown in Fig. 4B, which are linked to the sentiment conflict graph for "parking lot" (related word).

[0161] (1) A concrete example of prompt creation behavior when using the study method (B) First, a specific example (hereinafter referred to as "specific example 1") of the prompt creation operation in prompt creator 226 that is performed when study method (B) is adopted will be described with reference to FIGS. 19A to 19D.

[0162] Fig. 19A is an example of a prompt template used in the study method (B), and Fig. 19B is a diagram summarizing the contents (elements) of the input parameters of the prompt template. Fig. 19C is an example of a prompt with input parameters already entered, i.e., a prompt (completed prompt) to be input to the generation AI, and Fig. 19D is an example of a response obtained when the prompt shown in Fig. 19C is executed by the generation AI.

[0163] In specific example 1, the following prompt template shown in Fig. 19A is stored in advance in template storage unit 227. Note that [X], [O], [R], [Pc], and [Nc] in the following prompt template are input parameters of the prompt. “[X] analyzes various voices for the purpose of [O]. There were positive and negative opinions about [R]. Positive opinion: "[Pc]" Negative opinion: "[Nc]" Use the positive feedback to improve the negative feedback."

[0164] The information entered into input parameter [X] (elements of the post content) is information about the "analysis subject," and the information entered into input parameter [O] is information about the "analysis purpose." The information about the "analysis subject" is the target of the "analysis purpose." Therefore, in specific example 1, as shown in FIG. 19B, input parameter [X] (analysis subject) is "KN City," and input parameter [O] (analysis purpose) is "regional revitalization."

[0165] The information (element of the post content) input into the input parameter [R] is information on the "relative word," and in specific example 1, the input parameter [R] is "parking lot." The information input into the input parameter [Pc] is information on positive post content for the "relative word," and in specific example 1, the input parameter [Pc] is text data (sentence) of a positive post example about "parking lot" in KN city shown in FIG. 4A. The information input into the input parameter [Nc] is information on negative post content for the "relative word," and in specific example 1, the input parameter [Nc] is text data (sentence) of a negative post example about "parking lot" in KN city shown in FIG. 4B.

[0166] In specific example 1, first, the prompt creation unit 226 reads the prompt template shown in Fig. 19A from the template storage unit 227. The prompt creation unit 226 also acquires the input parameters [X], [O], [R], [Pc], and [Nc] (see Fig. 19B) of the prompt template. The method for acquiring these input parameters is as follows.

[0167] The input parameter [X] (analysis subject) is included in a query made when using the answer proposal support system 3, and so in this embodiment, query information is acquired in advance, and the prompt creation unit 226 acquires the input parameter [X] (analysis subject) by referencing the query information. Note that the present invention is not limited to this, and, for example, the query information may be linked to an emotion conflict graph, and the prompt creation unit 226 may acquire the input parameter [X] (analysis subject) by referencing the query information linked to the emotion conflict graph.

[0168] In this embodiment, the input parameter [O] (analysis purpose) is input by a user or the like by operating the operation unit (operation unit 45 in FIG. 2) of the proposal support device 22 when the prompt is created by the prompt creation unit 226. Note that the present invention is not limited to this, and for example, the input parameter [O] (analysis purpose) information may also be input when using the answer proposal support system 3, and the prompt creation unit 226 may acquire the input parameter [O] by referring to that information.

[0169] Regarding the input parameter [R] (related word), in this embodiment, when the prompt is created by the prompt creation unit 226, the user operates the operation unit (operation unit 45 in Figure 2) of the proposal support device 22 to select and input "parking lot" (related word).

[0170] For the input parameter [Pc] (positive post content), the prompt creation unit 226 acquires text data of positive post examples related to "parking lot" in KN city, for example, as shown in Fig. 4A, which are linked to the sentiment conflict graph of "parking lot" (related term). Also, for the input parameter [Nc] (negative post content), the prompt creation unit 226 acquires text data of negative post examples related to "parking lot" in KN city, for example, as shown in Fig. 4B, which are linked to the sentiment conflict graph of the related term "parking lot."

[0171] Next, the prompt creation unit 226 inputs the input parameters [X], [O], [R], [Pc], and [Nc] shown in Figure 19B into the prompt template shown in Figure 19A, creates the next prompt shown in Figure 19C, and stores the created prompt (completed prompt) in the prompt storage unit 228. "KN City is analyzing various opinions with the aim of revitalizing the local area. There were positive and negative comments about the parking lot. Positive opinion: "The text of the post Pc shown in Figure 4A (omitted)" Negative opinion: "The text of post Nc shown in Figure 4B (omitted)" Use the positive feedback to improve the negative feedback."

[0172] Then, when the proposal examining unit 229 inputs the prompt (completion prompt) shown in FIG. 19C into the generation AI and executes it, the generation AI produces the following example reply shown in FIG. 19D. “To solve the problem of limited parking spaces at T Park, we can use the parking spaces at E Park. Taking advantage of the fact that E Park has a fully equipped parking lot, we can make up for the lack of parking at T Park. Specifically, one possible measure would be to install signs directing people to the parking lot at E Park when the parking lot at T Park is full. In addition, the problem of lack of parking spaces can be solved by running a shuttle bus between E Park and T Park. By adopting these methods, we can improve the convenience of T Park and contribute to revitalizing the local area."

[0173] In specific example 1, at least the completion prompt and the example of a reply from the generation AI shown in FIG. 19C and FIG. 19D, respectively, are displayed on the display unit of the proposal support device 22 (display unit 46 in FIG. 2).

[0174] (2) A concrete example of prompt creation behavior when using method (C) Next, a specific example (hereinafter referred to as "specific example 2") of the prompt creation operation in prompt creator 226 performed when study method (C) is adopted will be described with reference to FIGS. 20A to 20D.

[0175] Fig. 20A is an example of a prompt template used in the study method (C), and Fig. 20B is a diagram summarizing the contents (elements) of the input parameters of the prompt template. Fig. 20C is an example of a prompt with input parameters already entered, i.e., a prompt (completed prompt) to be input to the generation AI, and Fig. 20D is an example of a response obtained when the prompt shown in Fig. 20C is executed by the generation AI.

[0176] In specific example 2, the following prompt template shown in Fig. 20A is stored in advance in the template storage unit 227. Note that [X], [O], [R], [P], [Pt], [N], [Nt], and [Co] in the following prompt template are input parameters of the prompt. “[X] analyzes various voices for the purpose of [O]. Regarding ([R], [Co]), [Pt] is thought to be [P] and [Nt] is thought to be [N]. Please take advantage of the good points of [Pt] and improve the bad points of [Nt]."

[0177] The input parameters [X], [O], and [R] are the same as the input parameters [X], [O], and [R] of specific example 1 described in Figures 19A to 19D, so description of these input parameters will be omitted here.

[0178] The information (elements of the post content) input to the input parameter [P] is information on a positive "emotion word," and in Specific Example 2, the input parameter [P] is "complete." The information input to the input parameter [Pt] is information on the target of the input parameter [P], and in Specific Example 2, the input parameter [Pt] is "T Park." Note that the target of the input parameter [P] here is the item that corresponds to a higher level of the "related word" (input parameter [P]) in the corresponding positive post example. Therefore, in Specific Example 2, "T Park" (see FIG. 4A), which has a "parking lot" (related word) that corresponds to "complete" (a positive emotion word), becomes the input parameter [Pt].

[0179] The information (element of the post content) input to the input parameter [N] is information on a negative "emotion word," and in Specific Example 2, the input parameter [N] is "the parking lot has few spaces." The information input to the input parameter [Nt] is information on the target of the input parameter [N], and in Specific Example 2, the input parameter [Nt] is "E Park." Note that the target of the input parameter [N] here is an item that corresponds to a higher level of the "related word" (input parameter [N]) in the corresponding negative post example. Therefore, in Specific Example 2, "E Park" (see FIG. 4B), which has a "parking lot" (related word) that corresponds to "the parking lot has few spaces" (a negative emotion word), becomes the input parameter [Nt].

[0180] Furthermore, the information (elements of the posted content) input to the input parameter [Co] is information on the common points between the input parameter [Pt] and the input parameter [Nt], and in the specific example 2, the input parameter [Co] is "park."

[0181] In Specific Example 2, first, the prompt creation unit 226 reads the prompt template shown in FIG. 20A from the template storage unit 227. The prompt creation unit 226 also acquires input parameters [X], [O], [R], [P], [Pt], [N], [Nt], and [Co] (see FIG. 20B) of the prompt template. The method for acquiring these input parameters is as follows. Note that the method for acquiring the input parameters [X], [O], and [R] is the same as the method for acquiring the input parameters [X], [O], and [R] described in Specific Example 1, so the method for acquiring these input parameters will not be described here.

[0182] The prompt creation unit 226 acquires the input parameter [P] (positive emotion word) by referring to the node of information on positive "emotion words" linked to the node of "parking lot" in the emotion conflict graph of "parking lot" (related word) (see FIG. 8). Note that the present invention is not limited to this, and the prompt creation unit 226 may acquire the input parameter [P] by referring to, for example, input information of positive post examples linked to the emotion conflict graph of "parking lot" (related word) (see FIGS. 3 and 4A).

[0183] The prompt creation unit 226 can extract the input parameter [Pt] (the target of the input parameter [P]) by executing a predetermined extraction processing program from text data of positive posting examples (see FIG. 4A) related to "parking lots" in KN City. Furthermore, when the prompt creation unit 226 creates a prompt, the user or the like may input the input parameter [Pt] by operating the operation unit (operation unit 45 in FIG. 2) of the proposal support device 20 while referring to the positive posting examples.

[0184] The prompt creation unit 226 acquires the input parameter [N] (negative emotion word) by referring to the node of information on negative "emotion words" linked to the node of "parking lot" in the emotion conflict graph of "parking lot" (related word) (see FIG. 8). Note that the present invention is not limited to this, and the prompt creation unit 226 may acquire the input parameter [N] by referring to, for example, input information of negative posting examples linked to the emotion conflict graph of "parking lot" (related word) (see FIGS. 3 and 4A).

[0185] The input parameter [Nt] (the target of the input parameter [N]) can be extracted by the prompt creation unit 226, for example, by executing a predetermined extraction processing program from text data of negative posting examples (see FIG. 4B) about "parking lots" in KN City. Also, when the prompt creation unit 226 creates a prompt, the user or the like may input the input parameter [Nt] by operating the operation unit (operation unit 45 in FIG. 2) of the proposal support device 22, referring to negative posting examples.

[0186] The input parameter [Co] (common point between the input parameters [Pt] and [Nt]) can be extracted by the prompt creation unit 226, for example, by executing a specific extraction processing program from positive and negative post examples related to "parking lots" in KN City, as well as from the input parameters [Pt] and [Nt]. Furthermore, the input parameter [Co] may be input by a user or the like operating the operation unit (operation unit 45 in FIG. 2) of the proposal support device 22 when the prompt creation unit 226 creates a prompt.

[0187] Next, the prompt creation unit 226 inputs the input parameters [X], [O], [R], [P], [Pt], [N], [Nt], and [Co] shown in Figure 20B into the prompt template shown in Figure 20A, and creates the next prompt shown in Figure 20C. At this time, if some of the input parameters are input by the user, etc., the prompt creation unit 226 waits until input of some of the input parameters from the user, etc. is complete. The prompt creation unit 226 also stores the created prompt in the prompt storage unit 228. "KN City is analyzing various opinions with the aim of revitalizing the local area. Regarding (parking lots, parks), T Park is thought to be fully equipped, while E Park is thought to have a limited number of parking spaces. Please take advantage of the good points of T Park and improve the bad points of E Park."

[0188] Then, when the proposal examining unit 229 inputs the prompt (completion prompt) shown in FIG. 20C into the generation AI and executes it, the generation AI produces the following example reply shown in FIG. 20D. “To solve the problem of lack of parking spaces in E Park, parking spaces in T Park can be used. T Park has a fully equipped parking lot, so it can make up for the lack of parking at E Park. Specifically, if E Park's parking lot is full, one possible solution would be to install signs directing people to T Park's parking lot. Another solution would be to operate a shuttle bus between T Park and E Park. Implementing these methods would improve the convenience of T Park and contribute to revitalizing the local area.

[0189] In the specific example 2, at least the completion prompt and the example of a reply from the generation AI shown in FIGS. 20C and 20D, respectively, are displayed on the display unit of the proposal support device 22 (display unit 46 in FIG. 2).

[0190] [Processing flow of the answer suggestion support system] Next, a description will be given of the processing flow of the process for supporting the proposal of an answer plan according to the purpose of use (problem), which is performed by the answer proposal support system 3 (sentiment analysis device 10 and proposal support device 22) of this embodiment. FIG. 21 is a flowchart showing the procedure of the process for supporting the proposal of an answer plan according to the purpose of use (problem), which is performed by the answer proposal support system 3 of this embodiment. Note that, also in this embodiment, the control of various processes by the sentiment analysis device 10 and the proposal support device 22 is executed on software by each function processing unit (see FIG. 15) included in the CPU of each device (CPU 41 in FIG. 2).

[0191] The processes of S51 to S55 in the proposal support process of this embodiment shown in FIG. 21 are similar to the processes of S1 to S5 in the proposal support process of the first embodiment (see FIG. 11), and therefore, description of these processes will be omitted here.

[0192] After the process of S55, the prompt creation unit 226 of the proposal support device 22 performs a process of creating a prompt (S56).

[0193] In the processing of S56, the prompt creation unit 226 first reads emotional conflict information (an emotional conflict graph and an emotional conflict list) from the emotional conflict information storage unit 205, and also reads a prompt template from the template storage unit 227. Next, the prompt creation unit 226 inputs predetermined data into the input parameters of the prompt template to create a prompt to be executed by the generation AI. Note that the method by which the prompt creation unit 226 creates a prompt differs depending on the type of method (review method (B) or (C)) for examining a proposed answer using the generation AI, and the specific creation methods are as described above in Specific Example 1 (see FIGS. 19A to 19D) and Specific Example 2 (see FIGS. 20A to 20D). The prompt creation unit 226 then stores the created prompt in the prompt storage unit 228.

[0194] Next, the proposal examination unit 229 of the proposal support device 22 reads the prompt (completion prompt) from the prompt storage unit 228, inputs the prompt to the generation AI, and executes the prompt (S57). In this process, the proposal examination unit 229 also obtains the execution result of the prompt, i.e., a response from the generation AI, and controls the display unit to display the response.

[0195] If the response from the generation AI is used as a proposed answer, the proposal review unit 229 stores the response as a proposed answer in the proposal storage unit 230 in the processing of S57. On the other hand, if, for example, the user etc. comprehensively considers a proposed answer by taking into account not only the response from the generation AI but also multiple pieces of text data referenced when creating the prompt and the results of research on various related information according to the purpose of use, the proposal review unit 229, after obtaining the response from the generation AI in the processing of S57, waits until a comprehensive proposed answer is input by the user etc., and once the comprehensive proposed answer is input, stores the proposed answer in the proposal storage unit 230.

[0196] Then, in this embodiment, after the processing of S57, the proposal support device 22 ends the proposal support processing for the answer plan.

[0197] [Various effects] As described above, the proposal support device 22 of this embodiment has the same functional components as the proposal support device 20 of the first embodiment. Therefore, this embodiment can achieve the same effects as the first embodiment.

[0198] Furthermore, in this embodiment, it is possible to provide support for proposing and examining answer plans using a generation AI. This makes it easier to propose and examine answer plans, and the examination time can be shortened. In particular, as explained in the above specific example 1, when sentences of multiple text data linked to an emotional conflict graph are directly input to the generation AI to examine answer plans (examination method (B)), the process of proposing and examining answer plans becomes easier, and the examination time can be significantly shortened.

[0199] Furthermore, as explained in Specific Example 2 above, when elements are extracted from sentences (posted content) of multiple text data linked to an emotional conflict graph and input into the generation AI to consider answer proposals (examination method (C)), the sentences of multiple text data are not used as is, thereby reducing the risk of copyright infringement. Furthermore, as explained in Specific Example 2 above, when examination method (C) is adopted, only elements related to the issue are used, so a more accurate response can be obtained from the generation AI.

[0200] 4. Fourth Embodiment Next, an answer suggestion support system according to a fourth embodiment of the present invention will be described. Note that, in the following description of this embodiment, components and devices having the same functions as those in the first embodiment will be denoted by the same reference numerals.

[0201] In this embodiment, a configuration will be described in which a new function is added to the proposal support device 20 of the first embodiment to weight the importance (priority) of the information on the "related words" that are the starting point of the extracted emotional conflict graph in terms of the consideration of answer proposals.

[0202] The configurations of the answer proposal support system 4 of this embodiment, and the emotion analysis device 10 and proposal support device 23 that constitute it, are similar to the configurations of the answer proposal support system 1 of the first embodiment described above with reference to Figures 1 and 2, and the emotion analysis device 10 and proposal support device 20 that constitute it. Therefore, illustrations and descriptions of the configuration of the answer proposal support system 4 and the hardware configurations of the emotion analysis device 10 and proposal support device 23 will be omitted here.

[0203] Furthermore, in the answer proposal support system 4 of this embodiment, the configuration of the input information of the proposal support device 23 is also similar to that of the first embodiment described above with reference to Fig. 3. Therefore, illustration and description of the configuration of the input information of the proposal support device 23 will be omitted here.

[0204] Furthermore, the processing operations of the emotion analysis device 10 of this embodiment and the contents from the input information acquisition process to the emotion conflict graph extraction process in the proposal support device 23 are the same as those of the first embodiment described above with reference to Figures 4 to 8 and 11 to 14. Therefore, here, explanations of the processing operations of the emotion analysis device 10 and the contents from the input information acquisition process to the emotion conflict graph extraction process in the proposal support device 23 will be omitted.

[0205] [Functional configuration of the answer suggestion support system] Fig. 22 is a functional block configuration diagram of the answer proposal support system 4 (the emotion analysis device 10 and the proposal support device 23) of this embodiment. Note that Fig. 22 only shows functional units related to the proposal support process for answer proposals (improvement proposals, project proposals, strategy proposals, etc.) according to the purpose of use, which is performed by the emotion analysis device 10 and the proposal support device 23.

[0206] 22 and 10, the functional block configuration of the proposal support device 23 of this embodiment is the same as the functional block configuration of the proposal support device 20 of the first embodiment, except that an emotional conflict information weighting unit 236 is newly added. Note that the other functional components in the proposal support device 23 of this embodiment and the functional components of the emotion analysis device 10 have the same functions as those described in the first embodiment, and therefore descriptions of these functional components will be omitted here.

[0207] As shown in FIG. 22, the emotional conflict information weighting unit 236 is functionally connected to the emotional conflict information storage unit 205 .

[0208] The emotional conflict information weighting unit 236 (importance setting unit) first reads out the emotional conflict list stored in the emotional conflict information storage unit 205. Then, the emotional conflict information weighting unit 236 performs weighting processing (importance setting processing) on ​​the information of each "related word" registered in the emotional conflict list, taking into account the importance (priority) of the information of each "related word" registered in the emotional conflict list for answer proposal consideration.

[0209] The importance of the "related word" information for examining answer proposals, i.e., the importance for the purpose of use, is arbitrarily set in advance, taking into consideration, for example, information from dictionaries and usage history. For example, the importance can be set by taking into consideration whether the "related word" information is a unique term (a term that is frequently used) or whether it is an uncommon term. Also, for example, the importance can be set according to the content (degree of positivity or negativity) of the "emotion word" information linked to the "related word" information in the emotion conflict graph.

[0210] Furthermore, the emotional conflict information weighting unit 236 may change the order of the information on the "related words" according to the importance in the emotional conflict list in which each weighted related word is registered. For example, the emotional conflict information weighting unit 236 may change the order of the information on each "related word" in the emotional conflict list from the top to the bottom in order of importance (priority). Furthermore, the emotional conflict information weighting unit 236 may delete "related words" with low importance from the emotional conflict list based on the importance in the emotional conflict list.

[0211] Then, the emotional conflict information weighting unit 236 stores an emotional conflict list including the information of the "related words" updated by the weighting process in the emotional conflict information storage unit 205. At this time, the emotional conflict information weighting unit 236 may, for example, associate the information of each "related word" in the emotional conflict list with the corresponding information of importance (priority), and store the emotional conflict list in the emotional conflict information storage unit 205. Furthermore, at this time, the emotional conflict information weighting unit 236 may, for example, add information of importance (priority) associated with the information of each "related word" in the emotional conflict list, and store the emotional conflict list with the added information of importance in the emotional conflict information storage unit 205. Note that the emotional conflict information weighting unit 236 may, for example, associate the information of importance with the information of the "related word" in the input information, and store the input information again in the emotion determination result storage unit 201.

[0212] [Processing flow of the answer suggestion support system] Next, a description will be given of the processing flow of the process of supporting the proposal of an answer plan according to the purpose of use (problem) performed by the answer proposal support system 4 (sentiment analysis device 10 and proposal support device 23) of this embodiment. FIG. 23 is a flowchart showing the procedure of the process of supporting the proposal of an answer plan according to the purpose of use (problem) performed by the answer proposal support system 4 of this embodiment. Note that, also in this embodiment, the control of various processes by the sentiment analysis device 10 and the proposal support device 23 is executed on software by each function processing unit (see FIG. 15) included in the CPU of each device (CPU 41 in FIG. 2).

[0213] The processes of S61 to S65 in the proposal support process of this embodiment shown in FIG. 23 are similar to the processes of S1 to S5 in the proposal support process of the first embodiment (see FIG. 11), and therefore, description of these processes will be omitted here.

[0214] After the process of S65, the emotional conflict information weighting unit 236 of the proposal support device 23 performs a weighting process of the emotional conflict information (S66).

[0215] In the process of S66, the emotional conflict information weighting unit 236 first reads the emotional conflict list created (updated) in the process of S65. Then, the emotional conflict information weighting unit 236 weights the information of each "related word" in the emotional conflict list according to the predetermined importance (priority) of the "related word" in the answer plan consideration. For example, the emotional conflict information weighting unit 236 links the information of each "related word" with quantitative information (e.g., a weighting coefficient) or qualitative information (e.g., high, medium, low, etc.) corresponding to the importance. As a result, each emotional conflict graph linked to the information of each "related word" in the emotional conflict list and each emotional conflict graph are also linked to importance (priority) information. In other words, the process of S66 essentially applies weighting processing to the emotional conflict information (e.g., emotional conflict graph and emotional conflict list).

[0216] In addition, in the processing of S66, the emotional conflict information weighting unit 236 may rearrange the order of information on each "related word" in the emotional conflict list, for example, in descending order of importance, based on the weighted "related word" information. Furthermore, in the processing of S66, the emotional conflict information weighting unit 236 may delete "related words" with low importance from the emotional conflict list based on the importance.

[0217] Next, the emotional conflict information weighting unit 236 re-stores the emotional conflict information whose weighting has been updated in the processing of S66 in the emotional conflict information storage unit 205 (S67). Then, after processing of S67, the proposal support device 23 ends the proposal support processing of this embodiment. Note that in this embodiment, after processing of S67, a user or the like of the answer proposal support system 4 carefully examines the weighted and updated emotional conflict information (emotion conflict graph and emotional conflict list), the multiple pieces of text data linked to the emotional conflict graph, and the contents of various other related information according to the purpose of use, considers answer proposals, and outputs the derived answer proposals.

[0218] [Various effects] As described above, the proposal support device 23 of this embodiment has the same functional components as the proposal support device 20 of the first embodiment. Therefore, this embodiment can achieve the same effects as the first embodiment.

[0219] Furthermore, in this embodiment, in the weighted and updated emotional conflict list, information on various "related words" can be registered, for example, in descending order of importance (priority). In this case, based on the output (displayed) emotional conflict list, users can easily recognize information on "related words" with high importance, which makes it easier to consider proposed answers (improvement proposals, project proposals, strategy proposals, etc.) and reduces the time required for consideration.

[0220] Furthermore, in this embodiment, based on the importance of the "related words," it is possible to delete "related words" with low importance from the emotional conflict list. In this case, since there is no need to consider "related words" with low importance, it is possible to significantly reduce the time required to consider response proposals (improvement proposals, project proposals, strategy proposals, etc.).

[0221] 5. Various Modifications The answer proposal support systems 1 to 4 (sentiment analysis device 10 and proposal support devices 20 to 23) according to various embodiments of the present invention have been described above, but the present invention is not limited to these, and various other modifications are possible without departing from the gist of the present invention as set forth in the claims. For example, the following various modifications can be adopted, and the following various modifications can also achieve the same effects as the various embodiments described above.

[0222] [Variation 1] The display mode of the emotion conflict information (emotion conflict graph and emotion conflict list) on the display unit of the proposal support device 21 in the second embodiment is not limited to the display mode shown in Fig. 16. Fig. 24 is a diagram showing the display mode of the emotion conflict information on the display unit of the proposal support device 21 in Modification 1.

[0223] In Modification 1, as shown in Fig. 24 , similarly to the second embodiment, a display area 51 in which an emotional conflict list is displayed and a display area 52 in which an emotional conflict graph is displayed are provided on a display screen 50 of a display unit of the proposal support device 21. In Modification 1, a search form section 53 for searching for information on "related words" is provided above the display area 51 within the display screen 50. When a user or the like operates the operation unit (operation unit 45 in Fig. 2 ) of the proposal support device 21 to input information on "related words" into the search form section 53, if there is information on the corresponding "related words" among the information group of "related words" registered in the emotional conflict list, an emotional conflict graph linked to the information on the "related words" is displayed in the display area 52.

[0224] Moreover, in the first modification, a check box 51a is provided to the left of each "related term" in the emotional conflict list displayed in the display area 51. When a user or the like operates (by a selection operation or a touch operation) the operation unit (operation unit 45 in FIG. 2) of the proposal support device 21 to select the check box 51a of a predetermined "related term," a check mark is displayed in the check box 51a. Note that if the "related term" input into the search form unit 53 is found in the information group of "related terms" registered in the emotional conflict list, a check mark is also displayed in the check box 51a of the "related term." In either case, an emotional conflict graph linked to the information of the predetermined "related term" ("parking lot" in the example shown in FIG. 24) for which a check mark is displayed is displayed in the display area 52.

[0225] When the display mode of the emotional conflict information (emotional conflict graph and emotional conflict list) of the first modification shown in Fig. 24 is adopted, it becomes easier for the user to select information on "related words," and the information on the currently selected "related word" can be easily recognized. Therefore, in this example, it becomes easier to consider proposed answers (improvement proposals, project proposals, strategy proposals, etc.), and the consideration time can be further reduced.

[0226] [Variation 2] The proposal support device 21 having the display function of the emotional conflict information (emotion conflict graph and emotional conflict list) described in the second embodiment may be added with a function of performing weighting processing of the importance (priority) of the emotional conflict information described in the fourth embodiment for consideration of answer plans (Modification 2). That is, the proposal support device 21 of the second embodiment may be combined with the proposal support device 23 of the fourth embodiment.

[0227] In this configuration, the information on the various "related words" in the emotional conflict list displayed in the display area 51 on the display screen 50 can be displayed, for example, in descending order of importance (priority). Therefore, in this case, the information on the "related words" that are highly important to the consideration of answer proposals can be easily recognized, making it easier to consider answer proposals (improvement proposals, project proposals, strategy proposals, etc.) and further shortening the consideration time.

[0228] [Variation 3] In the proposal support device 22 that examines answer proposals using the generation AI described in the third embodiment, a function for displaying the emotional conflict information (emotional conflict graph and emotional conflict list) described in the second embodiment or the first modification may be added (Modification 3). That is, the proposal support device 22 of the third embodiment may be combined with the proposal support device 21 of the second embodiment or the first modification.

[0229] Furthermore, in Modification 3, various information such as a response from the generation AI and a proposed answer derived through consideration by the user or the like may be displayed as proposal support information on the display screen 50 of the display unit (display unit 46 in FIG. 2) of the proposal support device 22. FIG. 25 is a diagram showing an example of the display of various information such as a response from the generation AI and a proposed answer derived through consideration by the user or the like. Note that the example shown in FIG. 25 is an example in which the display function of Modification 1 (see FIG. 24) is applied to the proposal support device 22 of the third embodiment.

[0230] In the third modification, various information such as a response from the generation AI and a proposed answer derived through consideration by the user or the like is displayed in a display area 52 within the display screen 50. Specifically, in the example shown in FIG. 25, text data of a positive post and a negative post linked to the "related word" information input to the generation AI (the input parameter "R" in FIG. 19B or FIG. 20B) are displayed in the "Evidence Post 1 (Positive)" and "Evidence Post 2 (Negative)" columns in FIG. 25, respectively. Also, in the example shown in FIG. 25, a prompt input to the generation AI (see FIG. 19C or FIG. 20C) is displayed in the "Consultation (Prompt)" column in FIG. 25. Then, a response from the generation AI (see FIG. 19D or FIG. 20D) is displayed in the "Suggestion Example (Response from the Generation AI)" column in FIG. 25.

[0231] Furthermore, for example, if a user comprehensively considers a proposed answer by taking into account not only the response from the generation AI but also multiple pieces of text data referenced when generating the prompt and the results of research into various related information according to the purpose of use, the derived comprehensive proposed answer will be displayed in the "Proposal Consideration Results" column in Figure 25 (within the area surrounded by the dashed line).

[0232] In addition, in Modification 3, before the generation AI is executed, an emotion conflict graph and an emotion conflict list are displayed as proposal support information on the display screen 50 of the display unit (display unit 46 in FIG. 2) of the proposal support device 22, for example, in the display mode explained in FIG. 24. Then, when the generation AI is executed, the display mode of the display screen 50 switches from the display mode shown in FIG. 24 to the display mode shown in FIG. 25.

[0233] [Variation 4] In the proposal support device 22 that uses the generation AI described in the third embodiment or the third modification, a function may be added to perform a weighting process of the importance (priority) of the emotional conflict information described in the fourth embodiment for the consideration of answer proposals (modification 4). That is, the proposal support device 22 of the third embodiment or the third modification may be combined with the proposal support device 23 of the fourth embodiment. With this configuration, in addition to the various effects obtained in the third embodiment, the various effects obtained in the fourth embodiment can also be obtained.

[0234] [Other various modifications] In the various embodiments described above, examples have been described in which the relationships between the information on "related words," "emotion words," and "category tags" in each piece of text data are represented by a graph (emotion graph, text association graph) consisting of nodes and edges. However, the present invention is not limited to this. Any aspect can be adopted as long as it is possible to define the relationships between the information on "related words," "emotion words," and "category tags." For example, instead of a graph, the relationships between the information on "related words," "emotion words," and "category tags" may be defined in the form of a table or the like. However, from the perspective of making it easier to recognize the relationships between the information on "related words," "emotion words," and "category tags," it is advantageous to represent the relationships between the information on "related words," "emotion words," and "category tags" by a graph consisting of nodes and edges, as in the various embodiments described above.

[0235] Furthermore, in the above various embodiments, examples have been described in which the input information for each piece of text data includes information on "emotions," "emotion words," "related words," and "domain tags." However, the present invention is not limited to this. For example, topic information and regional information contained in the text data (posted content) may be included in the input information. In this case, nodes for the topic information and regional information may be included in the emotion graph and text association graph.

[0236] In the various embodiments described above, examples have been described in which text data is input to the emotion analysis device 10 from text media 30, but the present invention is not limited to this, and for example, the data input to the emotion analysis device 10 may be voice data. In this case, the emotion analysis device 10 is provided with a processing function unit that converts the voice data into text data, and the converted text data is acquired by the text data acquisition unit 101.

[0237] In the various embodiments described above, when a plurality of proposed answers are derived, a processing function unit may be provided that sets a priority for each proposed answer with respect to the purpose of use (issue).

[0238] Furthermore, in the various embodiments described above, the various processing function units included in the proposal support devices 20 to 23 are configured as software, but the present invention is not limited to this, and some or all of the various processing functions described above may be configured as hardware. [Explanation of symbols]

[0239] 1,2,3,4...Answer proposal support system, 10...Emotion analysis device, 20,21,22,23...Proposal support device, 30...Text media, 50...Display screen, 51,52...Display area, 51a...Check box, 53...Search form, 101...Text data acquisition unit, 102...Text data storage unit, 103...Text data emotion determination unit, 201...Emotion determination result storage unit, 202...Text association graph creation unit, 203...Text association graph storage unit, 204...Emotion conflict graph extraction unit, 205...Emotion conflict information storage unit, 216...Emotion conflict information display control unit, 226...Prompt creation unit, 227...Template storage unit, 228...Prompt storage unit, 229...Proposal consideration unit, 230...Proposal storage unit, 236...Emotion conflict information weighting unit

Claims

1. a predetermined information creation unit that acquires an analysis result of the feelings of the text creator with respect to the information on the evaluation items included in each piece of text data based on the content of each piece of text data related to the purpose of use, creates relationship information indicating the relationship between the evaluation items and the information on the feelings for each piece of text data based on the analysis result, and combines the plurality of relationship information created for each piece of text data into one piece of predetermined information that associates the plurality of text data; a specific information extraction unit that extracts, from the predetermined information, specific information regarding predetermined evaluation items that are associated with information regarding the positive emotions and information regarding the negative emotions. Information processing device.

2. The system further includes an answer proposal output unit that uses generative artificial intelligence to output an answer proposal for the purpose of use based on the specific information extracted by the specific information extraction unit and two or more text data corresponding to the specific information. The information processing device according to claim 1 .

3. A prompt creation unit that creates a prompt for executing the generative artificial intelligence, The prompt creation unit inputs the two or more pieces of text data corresponding to the specific information into the prompt. The information processing device according to claim 2 .

4. A prompt creation unit that creates a prompt for executing the generative artificial intelligence, The prompt creation unit extracts information regarding input parameters of a prompt for executing the generative artificial intelligence from the two or more pieces of text data corresponding to the specific information, and inputs the extracted information into the prompt. The information processing device according to claim 2 .

5. a display unit that can display the specific information extracted by the specific information extraction unit; a display control unit that controls the display operation of the specific information on the display unit. The information processing device according to claim 1 .

6. The display control unit controls the display unit to display, as the specific information, a graph in which information regarding positive emotions and information regarding negative emotions for the predetermined evaluation item are linked. The information processing device according to claim 5 .

7. The display control unit controls the display unit to display a list of the predetermined evaluation items. The information processing device according to claim 6 .

8. The system further includes an importance setting unit that sets an importance level for the predetermined evaluation item with respect to the purpose of use. The information processing device according to claim 1 .

9. When there are a plurality of types of the predetermined evaluation items, the importance setting unit creates a list of the plurality of types of the predetermined evaluation items registered in an order according to the importance set for each of the plurality of types of the predetermined evaluation items. The information processing device according to claim 8 .

10. The method further includes a sentiment analysis device that analyzes the sentiment of a text creator regarding information on the evaluation items included in each of the plurality of text data based on the content of each of the plurality of text data related to the purpose of use. The information processing device according to claim 1 .

Images