Information processing device, analysis method, and analysis program

Abstract

To enable various analysis objects to be efficiently analyzed.SOLUTION: An information processing device comprises: a data acquisition unit for acquiring feature data representing the feature of an analysis object, and a phrase acquisition unit for causing a language model having been trained by machine learning to generate, using the feature data, an appealing word or phrase representing the point for the analysis object to appeal to a prescribed subject, or causing the language model to select one that suits the feature data among candidate appealing words and phrases. According to this information processing device, it is made possible to assist a user in making decisions as well.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, an analysis method, and an analysis program. [Background technology]

[0002] When planning a new product, trend analysis is often performed on existing hit products. Generally, trend analysis relies on the experience and intuition of the analyst, but there are also known technologies for automating trend analysis. One example is the SNS (Social Networking Service) image trend analysis system described in Patent Document 1. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2020 / 035934 Summary of the Invention [Problem to be solved by the invention]

[0004] The SNS image trend analysis system described in Patent Document 1 extracts images tagged with specific hashtags from SNS and analyzes SNS trends based on the relationship between the components of the images and the hashtags that co-occur with those images. However, this SNS image trend analysis system has room for improvement in various ways, such as being unable to analyze images that do not have hashtags tagged.

[0005] The present disclosure has been made in view of the above-mentioned problems, and an exemplary purpose thereof is to provide a technique that enables efficient analysis of various analysis targets. [Means for solving the problem]

[0006] An information processing device according to an exemplary aspect of the present disclosure includes a data acquisition means for acquiring feature data representing the characteristics of an object of analysis, and a phrase acquisition means for causing a machine-learned language model to generate appealing phrases representing points that the object of analysis appeals to a predetermined target audience using the feature data, or for causing the language model to select from among the candidate appealing phrases those that match the feature data.

[0007] An analysis method according to an exemplary aspect of the present disclosure includes a data acquisition process in which at least one processor acquires feature data representing the characteristics of an object of analysis, and a phrase acquisition process in which the processor uses the feature data to generate appealing phrases representing points that the object of analysis appeals to a predetermined target audience, or to select from among the candidate appealing phrases the language model selects one that matches the feature data.

[0008] An analysis program according to an exemplary aspect of the present disclosure causes a computer to function as a data acquisition means for acquiring feature data representing the characteristics of an object of analysis, and as a phrase acquisition means for using the feature data to generate appealing phrases representing the points that the object of analysis appeals to a specified target audience in a machine-learned language model, or for selecting from among the candidate appealing phrases those that match the feature data in the language model. [Effects of the Invention]

[0009] According to an exemplary aspect of the present disclosure, an exemplary effect is achieved in that various analysis targets can be efficiently analyzed. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a block diagram illustrating a configuration of an information processing device according to the present disclosure. [Figure 2] FIG. 1 is a flow chart showing the flow of an analysis method according to the present disclosure. [Figure 3] FIG. 10 is a block diagram showing a configuration of another information processing device according to the present disclosure. [Figure 4]FIG. 10 is a diagram showing an example of acquiring feature phrases and appeal phrases. [Figure 5] FIG. 10 is a diagram illustrating an example of a graph showing the analysis results. [Figure 6] 4 is a flowchart showing the flow of processing executed by the information processing device shown in FIG. 3. [Figure 7] FIG. 1 is a block diagram illustrating a configuration of a computer that functions as an information processing device according to the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

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

[0012] First Exemplary Embodiment A first exemplary embodiment, which is an example of an embodiment of the present invention, will be described in detail with reference to the drawings. This exemplary embodiment is a basic form for each of the exemplary embodiments described below. Note that the scope of application of each technique employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technique employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical obstacles arise. Furthermore, each technique shown in the drawings referenced to explain this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical obstacles arise.

[0013] (Configuration of information processing device 1) The configuration of an information processing device 1 according to this exemplary embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the configuration of the information processing device 1. As shown in Fig. 1, the information processing device 1 includes a data acquisition unit 101 and a phrase acquisition unit 102.

[0014] The data acquisition unit 101 acquires feature data that represents the features of the analysis target. Here, the "analysis target" is the target of analysis by the information processing device 1. The analysis target of the information processing device 1 is arbitrary. For example, the information processing device 1 can also analyze arbitrary products and arbitrary services. Furthermore, for example, the information processing device 1 can also analyze arbitrary regions, arbitrary people, arbitrary organizations, and arbitrary groups.

[0015] Furthermore, the "feature data" may be any data that represents the characteristics of the analysis target, and the data format is arbitrary. For example, the feature data may be text data such as a phrase (a single word or a phrase consisting of multiple words) or a document that represents the characteristics of the target, or an image that represents the characteristics of the analysis target. The feature data may also be a graph or numerical data that represents the characteristics of the analysis target, or a combination of these data.

[0016] The phrase acquisition unit 102 uses the feature data acquired by the data acquisition unit 101 to generate, in a machine-learned language model, appealing phrases that express the appealing points of the analysis target to a predetermined target audience. Here, the "appealing phrase" may be a single word or a phrase consisting of multiple words. Furthermore, the "appealing points" are points that appeal to the target audience, and can also be rephrased as the values ​​of the target audience who chooses the analysis target, or the points that the target audience perceives as valuable. Therefore, the "appealing phrases" can also be rephrased as words that express values, keywords that express values, key phrases that express values, etc.

[0017] The "predetermined target" will depend on the subject of analysis. For example, if the subject of analysis is a product or service, the consumer of that product or service will be the predetermined target. For example, if the subject of analysis is a tourist destination, the visitors of that tourist destination will be the predetermined target.

[0018] Furthermore, the "language model" may be one that has been machine-learned so as to generate appealing phrases that express the appealing points of an analysis target having characteristics represented in the feature data to the target. For example, if the feature data is text data, the language model may be a model that has been machine-learned to learn the arrangement of components (such as words) in a sentence or the arrangement of sentences in a piece of writing. For example, if the feature data is image data, the language model may be a model that has been machine-learned to learn the relationship between the image data and phrases that express the appealing points of the target represented by the image data to the target.

[0019] Furthermore, the phrase acquiring unit 102 may acquire appeal phrases by having a machine-learned language model select, from among the appeal phrase candidates, those that match the feature data acquired by the data acquiring unit 101. In this case, the phrase acquiring unit 102 may input the appeal phrase candidates and the feature data acquired by the data acquiring unit 101 to the language model, and output candidates that match the feature data.

[0020] As described above, the information processing device 1 according to this exemplary embodiment is configured to include a data acquisition unit 101 that acquires feature data that represents the features of the analysis target, and a phrase acquisition unit 102 that acquires appealing phrases by either using the feature data acquired by the data acquisition unit 101 to generate appealing phrases that represent the points that the analysis target appeals to a predetermined target audience in a machine-learned language model, or by having the machine-learned language model select from among candidate appealing phrases those that match the feature data acquired by the data acquisition unit 101.

[0021] Understanding the appealing points of an analysis target to a predetermined target audience is important in analyzing the analysis target, but not everyone can easily grasp such points. In this regard, with the above configuration, a user of the information processing device 1 can acquire appealing phrases that represent the appealing points of the analysis target to a predetermined target audience simply by inputting feature data that represents the characteristics of the analysis target into the information processing device 1 and having the information processing device 1 acquire the feature data. Therefore, with the above configuration, it is possible to obtain the effect of efficiently analyzing various analysis targets.

[0022] Furthermore, because appealing phrases are phrases that express the appealing points of the analysis target to a predetermined target audience, they can be said to be useful when a user makes a decision regarding the analysis target. Therefore, the above configuration can also support the user's decision-making. For example, when using the information processing device 1 for product development, it is possible to analyze existing products related to the product to be developed or existing hit products and obtain appealing phrases that express the appealing points of the analysis target to consumers. Then, the user can refer to the obtained appealing phrases and determine the elements to be incorporated into the product to be developed.

[0023] Furthermore, the configuration in which the machine-learned language model selects, from among the candidate appeal phrases, those that match the feature data acquired by the data acquisition unit 101 has the advantage of preventing the analysis results from becoming complicated. That is, when appeal phrases are repeatedly acquired for the same analysis target, or when appeal phrases are acquired for each of multiple analysis targets, it is expected that the total number of acquired appeal phrases will diverge, causing the analysis results to become complicated. In this regard, when appeal phrases are selected from the candidates, the total number of acquired appeal phrases is less likely to diverge, preventing the analysis results from becoming complicated.

[0024] (Analysis Program) The functions of the information processing device 1 described above can also be realized by a program. The analysis program according to this exemplary embodiment causes a computer to function as a data acquisition unit that acquires feature data representing the characteristics of an analysis target, and as a phrase acquisition unit that uses the feature data to generate appealing phrases representing the points that the analysis target appeals to a predetermined target audience in a machine-learned language model, or to select from candidate appealing phrases that match the feature data in the language model. This analysis program enables efficient analysis of various analysis targets.

[0025] (Analysis method flow) The flow of the analysis method according to this exemplary embodiment will be described with reference to Fig. 2. Fig. 2 is a flow diagram showing the flow of the analysis method. Note that the execution entity of each step in this analysis method may be a processor provided in the information processing device 1, or a processor provided in another device, or each step may be executed by a processor provided in a different device.

[0026] In S1 (data acquisition process), at least one processor acquires feature data representing the features of the analysis target.

[0027] In S2 (phrase acquisition process), at least one processor uses the feature data acquired in S1 to cause a machine-learned language model to generate appealing phrases that express the points that the analysis subject appeals to a predetermined target audience. Alternatively, in S2, at least one processor causes the machine-learned language model to select, from among the candidate appealing phrases, those that match the feature data acquired in S1.

[0028] As described above, the analysis method according to this exemplary embodiment includes a data acquisition process in which at least one processor acquires feature data representing the characteristics of the analysis target, and a phrase acquisition process in which at least one processor uses the acquired feature data to generate appealing phrases representing the points that the analysis target appeals to a predetermined target audience, or a phrase acquisition process in which at least one processor selects, from candidate appealing phrases, those that match the acquired feature data, using the machine-learned language model. Therefore, the analysis method according to this embodiment enables efficient analysis of various analysis targets.

[0029] Second Exemplary Embodiment A second exemplary embodiment, which is one example of an embodiment of the present invention, will be described in detail with reference to the drawings. Components having the same functions as those described in the above exemplary embodiment will be assigned the same reference numerals, and their description will be omitted as appropriate. The scope of application of each technology employed in this exemplary embodiment is not limited to this exemplary embodiment. That is, each technology employed in this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs. Furthermore, each technology shown in each drawing referenced to explain this exemplary embodiment can also be employed in other exemplary embodiments included in the present disclosure, to the extent that no particular technical hindrance occurs.

[0030] (Configuration of information processing device 1A) The configuration of an information processing device 1A according to this exemplary embodiment will be described with reference to FIG. 3. FIG. 3 is a block diagram showing the configuration of the information processing device 1A. The information processing device 1A is a device having a function of supporting analysis of a predetermined analysis target. The analysis target is arbitrary, as in the first exemplary embodiment. The information processing device 1A may be a device whose main function is analysis support, or may be a general-purpose device having other functions as well. Furthermore, the information processing device 1A may be a stationary device or a portable device.

[0031] As shown in the figure, the information processing device 1A includes a control unit 10A that controls all the units of the information processing device 1A, and a storage unit 11A that stores various data used by the information processing device 1A. The information processing device 1A also includes a communication unit 12A that enables the information processing device 1A to communicate with other devices, an input unit 13A that accepts input to the information processing device 1A, and an output unit 14A that enables the information processing device 1A to output data. The control unit 10A includes a data acquisition unit 101A, a phrase acquisition unit 102A, a classification unit 103A, and a display control unit 104A. The storage unit 11A also stores a language model 111A. The classification unit 103A will be described later in the section "Classification of appeal phrases."

[0032] The data acquisition unit 101A acquires feature data representing the features of the analysis target, similar to the data acquisition unit 101 in the first exemplary embodiment. Below, an example will be described in which the data acquisition unit 101A acquires, as feature data, feature words and phrases that are included in a document that describes the analysis target and that represent the features of the analysis target. Note that a method for acquiring feature words and phrases will be described later in the section "Method for acquiring feature words and phrases."

[0033] The phrase acquiring unit 102A, like the phrase acquiring unit 102 in the first exemplary embodiment, acquires appealing phrases using the feature data acquired by the data acquiring unit 101A. Specifically, the phrase acquiring unit 102A causes the language model 111A to generate appealing phrases that represent points that the analysis target appeals to a predetermined target audience. Alternatively, the phrase acquiring unit 102A causes the language model 111A to select, from among the appealing phrase candidates, those that match the feature data acquired by the data acquiring unit 101A. Note that a more detailed method for acquiring appealing phrases will be described later in the section "Method for Acquiring Appealing Phrases."

[0034] The language model 111A is a machine-learned model that has been machine-trained to generate appealing phrases that represent points that an analysis target having characteristics expressed in feature data will appeal to a target user. More specifically, the language model 111A is a language model that has been machine-trained to determine the arrangement of components (such as words) in a sentence and the arrangement of sentences in a piece of writing.

[0035] The information processing device 1A does not necessarily need to include the language model 111A, and may use a language model 111A stored in a device external to the information processing device 1A. In this case, the phrase acquisition unit 102A instructs the external device including the language model 111A to generate or select an appeal phrase, and acquires from the device the appeal phrase that the external device has generated or selected using the language model 111A.

[0036] The display control unit 104A causes a display device to display information for supporting analysis related to the analysis target. For example, the display control unit 104A may cause the display device to display the appeal phrases acquired by the phrase acquisition unit 102A. Furthermore, as will be described in detail later, the display control unit 104A may cause the appeal phrases to be displayed in the form of a graph. Note that the display device may be included in the information processing device 1A, or may be an external device to the information processing device 1A. For example, when the output unit 14A is a display device, the display control unit 104A may cause the output unit 14A to display and output the analysis results.

[0037] As described above, the information processing device 1A according to this exemplary embodiment is configured to include a data acquisition unit 101A that acquires feature data representing the features of an analysis target, and a phrase acquisition unit 102A that acquires appealing phrases by using the feature data acquired by the data acquisition unit 101A to cause a machine-learned language model 111A to generate appealing phrases representing points that the analysis target appeals to a predetermined target, or by having the machine-learned language model 111A select from candidate appealing phrases those that match the feature data acquired by the data acquisition unit 101A. This provides the effect of enabling efficient analysis of various analysis targets.

[0038] The information processing device 1A can be used, for example, to discover and interpret consumer values ​​and to support the creation of product concepts. Therefore, the information processing device 1A can be suitably used, for example, for information analysis when planning new products or renewing existing products. For example, by using the information processing device 1A, it is possible to discover other companies' products that share values ​​with one's own products. Furthermore, by using the information processing device 1A, it is also possible to discover other values ​​possessed by the discovered products of other companies. Such discoveries can provide hints for devising new products that can compete with other companies' products. In this way, the appealing phrases generated or selected by the information processing device 1A may support decision-making for product development.

[0039] (How to obtain characteristic phrases) As described above, the data acquiring unit 101A acquires, as feature data, feature words and phrases that are included in a document describing the analysis target and that indicate the characteristics of the analysis target. For example, the data acquiring unit 101A may identify feature words and phrases that indicate the characteristics of the analysis target by performing morphological analysis on the document describing the analysis target, and acquire the identified feature words and phrases as feature data. Note that the document describing the analysis target is arbitrary, and may be, for example, a document posted on a website introducing the analysis target, or a document containing user reviews of the analysis target.

[0040] Furthermore, the data acquiring unit 101A may cause the language model 111A to extract characteristic words and phrases from documents that explain the analysis target. In this case, the data acquiring unit 101A generates a prompt that instructs the language model 111A to extract characteristic words and phrases that indicate characteristics of the analysis target from the documents that explain the analysis target, and inputs the generated prompt and the documents to the language model 111A. As a result, characteristic words and phrases are output from the language model 111A.

[0041] The prompt may be generated by, for example, entering the name of the analysis target in the {analysis target} portion of a template that reads, "Please extract phrases that indicate the characteristics of {analysis target} from the input document." Such a template may be stored in advance in the storage unit 11A or the like. The same applies to the other templates described below.

[0042] The data acquiring unit 101A may directly use the words and phrases extracted from the document explaining the analysis target as characteristic words and phrases (characteristic data), or may present the extracted words and phrases to the user of the information processing device 1A and allow the user to select characteristic words and phrases from the presented words and phrases. For example, the data acquiring unit 101A may display the extracted words and phrases on a display device via the display control unit 104A and allow the user to select characteristic words and phrases from the displayed words and phrases.

[0043] As described above, the data acquiring unit 101A may acquire, as feature data, feature words and phrases that indicate the characteristics of the analysis target and are included in a document describing the analysis target. In this case, the phrase acquiring unit 102A generates appeal phrases by inputting the feature words and phrases into the language model 111A. Alternatively, the phrase acquiring unit 102A inputs the acquired feature words and phrase candidates into the language model 111A, thereby selecting, from the input candidates, those that match the input feature words and phrases as appeal phrases. This achieves, in addition to the effects achieved by the information processing device 1, the effect of being able to acquire appeal phrases from a document describing the analysis target.

[0044] The data acquiring unit 101A can also generate characteristic phrases from image data that indicate the characteristics of the analysis target. For example, the data acquiring unit 101A can generate characteristic phrases by inputting image data that indicates the characteristics of the analysis target into a language model that has learned the relationship between the image data and the phrases that indicate the characteristics of the analysis target shown in the image data through machine learning. By using such a language model, the data acquiring unit 101A can generate characteristic phrases such as "spicy" from image data that shows ramen with bright red soup.

[0045] In this way, the data acquiring unit 101A may generate characteristic words and phrases that indicate the characteristics of the analysis target from an image showing the analysis target and acquire the characteristic words and phrases as feature data. In this case, the word and phrase acquiring unit 102A inputs the acquired characteristic words and phrases into the language model 111A to generate appeal words and phrases. Alternatively, the word and phrase acquiring unit 102A inputs the acquired characteristic words and phrases and candidate appeal words into the language model 111A to select, from the input candidates, those that match the input characteristic words and phrases as appeal words and phrases. This provides the effect of acquiring appeal words and phrases from an image showing the analysis target in addition to the effect provided by the information processing device 1.

[0046] (How to obtain appealing phrases) As described above, the phrase acquiring unit 102A may cause the language model 111A to generate an appeal phrase, or may input candidates for appeal phrases to the language model 111A and cause the language model 111A to select an appeal phrase from the input candidates. Specifically, the phrase acquiring unit 102A can cause the language model 111A to generate or select an appeal phrase by generating a prompt for generating or selecting an appeal phrase and inputting the prompt to the language model 111A.

[0047] A prompt for generating appeal phrases can be generated using a template such as "Please tell me the points that the feature {characteristic phrase} of {analysis target} appeals to consumers." In this case, the phrase acquisition unit 102A can generate a prompt for generating appeal phrases by inputting the name of the analysis target in the {analysis target} part of the template and inputting characteristic phrases in the {characteristic phrase} part.

[0048] Furthermore, the prompt for selecting an appeal phrase can be generated using a template such as, "Please select from the list an appropriate feature {characteristic phrase} in {analysis target} that appeals to consumers." In this case, the phrase acquisition unit 102A can generate a prompt for selecting an appeal phrase by inputting the name of the analysis target in the {analysis target} portion of the template and the characteristic phrase in the {characteristic phrase} portion. Then, the phrase acquisition unit 102A can input a list containing candidate appeal phrases along with this prompt.

[0049] Note that if there is no appropriate phrase among the candidates, a new appeal phrase may be generated. In this case, the phrase acquisition unit 102A may input a prompt to the language model 111A instructing it to generate a new appeal phrase if there is no appropriate phrase among the candidates. For example, the phrase acquisition unit 102A may input a prompt to the language model 111A such as, "Please select from the list an appropriate phrase that represents the point that the feature {characteristic phrase} in {analysis target} appeals to consumers. However, if there is no appropriate phrase, please generate a phrase that indicates the point that the feature {characteristic phrase} in {analysis target} appeals to consumers."

[0050] Furthermore, instead of inputting feature phrases to the language model 111A, the phrase acquisition unit 102A may input documents describing the analysis target as feature data to the language model 111A. In this case, too, the language model 111A can be made to generate or select appeal phrases using the prompts described above.

[0051] Candidates for appeal phrases may be determined in advance. In this case, appeal phrases are selected from the range of predetermined candidates, thereby preventing the appeal phrases from diverging. In particular, when analyzing products and services with various attributes, the appeal points of products and services with different attributes are selected from common candidates, which has the advantage of linking products and services with different attributes with common appeal phrases. Note that "attributes" here refer to the characteristics or properties of the analysis target, such as a product or service. The attributes used as the basis for analyzing the analysis target can be determined as appropriate. For example, the genre or category of the product or service can be used as the "attributes."

[0052] Furthermore, for each analysis target included in a group of analysis targets with common attributes, appeal phrases generated or selected by the language model 111A may be used as candidates, allowing appeal phrases in some groups to express the appeal points of analysis targets in other groups.

[0053] For example, if the analysis target is a product, attributes such as "food," "clothing," and "daily necessities" may be associated with each product to be analyzed. In this case, the phrase acquisition unit 102A may cause the language model 111A to generate appeal phrases for each product to which one of these attributes, for example, the "food" attribute, is associated. Then, for each product to which an attribute other than "food" is associated, the phrase acquisition unit 102A may cause the language model 111A to select appeal phrases from the appeal phrases generated for each product to which the "food" attribute is associated. In this way, the appeal points of the analysis target having other attributes can be expressed by the appeal phrases generated for each product to which the "food" attribute is associated.

[0054] (Example of obtaining characteristic phrases and appealing phrases) FIG. 4 is a diagram showing an example of acquiring feature words and appeal words. FIG. 4 shows an example of acquiring feature words and appeal words from a document 401 that explains "Beer A" and a document 402 that explains "Shampoo C." Document 401 may be, for example, a web page introducing Beer A. Document 401 may also be a document containing reviews from people who have drunk Beer A. The same applies to document 402.

[0055] 4, the data acquiring unit 101A acquires the phrase "pleasant bitterness" and the phrase "profound taste" as characteristic phrases contained in the document 401. The phrase acquiring unit 102A then acquires the appeal phrase "refreshing feeling" using the characteristic phrase "pleasant bitterness," and acquires the appeal phrase "satisfaction" using the characteristic phrase "profound taste."

[0056] These appealing phrases all indicate the points that Beer A appeals to its consumers, in other words, the values ​​that consumers feel about Beer A, and are useful information for, for example, marketing, but are not included in document 401. A user of information processing device 1A can obtain such useful information simply by inputting document 401 into information processing device 1A.

[0057] 4, the data acquiring unit 101A acquires the phrase "carbonated bubbles" and the phrase "gentle fragrance" as characteristic phrases contained in the document 402. The phrase acquiring unit 102A then acquires the appeal phrase "refreshing feeling" using the characteristic phrase "carbonated bubbles," and acquires the appeal phrase "relaxing effect" using the characteristic phrase "gentle fragrance."

[0058] In the example of Fig. 4, the appeal phrase "refreshing feeling" is a appeal phrase acquired for both beer A and shampoo C. In this way, the phrase acquisition unit 102A can also acquire a common appeal phrase for products with different attributes, namely beer A and shampoo C. This shows that "refreshing feeling" is a point that appeals to consumers across the boundaries of product attributes.

[0059] (Graph example) The display control unit 104A may display, on the display device, a graph in which nodes representing the analysis target and nodes representing appeal phrases generated or selected for the analysis target are connected by edges as the analysis result of the analysis target. This provides the effect of presenting the analysis result to the user in an intuitively easy-to-understand form in addition to the effect provided by the information processing device 1.

[0060] Fig. 5 is a diagram showing an example of a graph showing the analysis results. Specifically, graph 501 shown in Fig. 5 shows the analysis results for "Beer A" shown in Fig. 4. Graph 501 includes node N1, which indicates the beer A being analyzed, and nodes n1 and n2, which indicate the appealing phrases "satisfaction" and "refreshing feeling" generated for beer A, respectively.

[0061] In addition, in graph 501, node N1 is connected to nodes n1 and n2 by edges, which makes it possible to recognize at a glance that the appeal of beer A to consumers is its satisfaction and refreshing feeling.

[0062] Furthermore, in graph 501, the phrase "profound taste" is displayed in association with the edge connecting node N1 and node n1. This phrase is a characteristic phrase of beer A obtained from document 401 in FIG. 4, and is the basis for the appeal phrase "satisfaction." Similarly, in graph 501, the phrase "pleasant bitterness" is displayed in association with the edge connecting node N1 and node n2. This phrase is a characteristic phrase of beer A obtained from document 401 in FIG. 4, and is the basis for the appeal phrase "refreshing."

[0063] In this way, the display control unit 104A may display the characteristic phrases that are the source of the appeal phrases in association with the appeal phrases. This allows the user to recognize which appeal phrases were generated or selected from which characteristic phrases. The user can then perform analysis taking into account the characteristic phrases, and can also confirm whether appropriate appeal phrases have been generated or selected.

[0064] Furthermore, when the information processing device 1A is used for marketing, it is preferable to analyze multiple products and services and acquire appealing phrases for each analysis target, which makes it possible to associate and display products and services that share common appealing phrases.

[0065] For example, when a claim phrase acquired for a certain analysis target is selected, the display control unit 104A may display other analysis targets for which the same claim phrase has been acquired, thereby allowing the user to recognize what other analysis targets have the claim point that the user has focused on.

[0066] Furthermore, if other appealing phrases are acquired for the other analysis targets, the display control unit 104A may also display the other appealing phrases, thereby providing the user with further information that is useful for marketing, etc.

[0067] For example, suppose an operation to select node n2 is performed on graph 501. In this case, the display control unit 104A may display graph 502 shown in Fig. 5. In addition to nodes N1, n1, and n2 that were included in graph 501, graph 502 also includes nodes N2, N3, n3, and n4.

[0068] Node N2 is a node indicating "Beer B," one of the analysis targets. Node n2, which represents the appealing phrase "refreshing feeling," and node n3, which represents the appealing phrase "easy to drink," are connected to node N2 by edges. The edge connecting nodes N2 and n2 is displayed in association with the characteristic phrase "citrus flavor," and the edge connecting nodes N2 and n3 is displayed in association with the characteristic phrase "light taste."

[0069] Node N3 is a node indicating "Shampoo C," one of the analysis targets. Node N23 is connected by edges to node n2, which represents the appeal phrase "refreshing feeling," and node n4, which represents the appeal phrase "relaxing effect." The edge connecting nodes N2 and n2 is associated with and displays the characteristic phrase "carbonated bubbles," and the edge connecting nodes N3 and n4 is associated with and displays the characteristic phrase "gentle fragrance."

[0070] From graph 502, the user can easily understand that among products that have the appeal point of refreshing feeling, there are also products that have additional appeal points such as satisfaction, ease of drinking, or a relaxing effect. Such information is useful, for example, when developing a new product or renewing an existing product, when considering what appeal points to make of the product. In particular, common appeal points between products with different attributes, such as beer and shampoo, are difficult to notice through normal analysis, but information processing device 1A can provide awareness of such appeal points.

[0071] For example, the connection of nodes n2 and n4 to node N3 suggests that the combination of refreshing and relaxing effects enhances the appeal to users. Therefore, a user considering renewing Beer A can consider a renewal that adds a new appeal point of relaxing effects to the existing appeal points of Beer A, such as satisfaction and refreshing feeling. In this way, appealing phrases can also be used to support the idea generation of product concepts.

[0072] (Classification of appealing phrases) As in the examples of Figures 4 and 5, it would be desirable to obtain common appealing phrases for analysis targets with different attributes. However, when the attributes of the analysis targets are different, the characteristic phrases are often phrases that correspond to the attributes, and therefore the appealing phrases are likely to be different depending on the attributes. Furthermore, even if the attributes of the analysis targets are the same, the documents that explain the analysis targets are diverse, so a variety of characteristic phrases will be obtained, and as a result, a variety of appealing phrases will be obtained, which may result in complicated analysis results.

[0073] In this regard, the information processing device 1A includes a classification unit 103A that classifies a plurality of appeal phrases acquired by the phrase acquisition unit 102A according to the content of the appeal phrases. Then, the display control unit 104A displays a representative phrase representing each appeal phrase of the same classification in each node corresponding to the appeal phrase of the same classification. This not only achieves the effect of the information processing device 1, but also has the effect of making it possible to present the analysis results in an easy-to-understand manner even when a variety of appeal phrases are acquired.

[0074] The method of classifying the appeal phrases is arbitrary and not particularly limited. For example, the classification unit 103A may vectorize each of the appeal phrases to be classified and calculate the similarity between each of the vectorized appeal phrases. In this case, the classification unit 103A may classify appeal phrases whose calculated similarity is equal to or greater than a predetermined threshold into the same group.

[0075] Furthermore, the classification unit 103A may cause the language model 111A to output the similarity between the claim phrases. In this case, the classification unit 103A may generate a prompt instructing the language model 111A to output the similarity between the claim phrases and input the generated prompt to the language model 111A. Such a prompt can be generated, for example, by inputting each claim phrase to be subjected to similarity judgment in the {claim phrase} portion of a template such as "Please express the similarity between {claim phrase} and {claim phrase} in a numerical range from 0 to 1." Note that the numerical range of the similarity is arbitrary.

[0076] Alternatively, the classification unit 103A may cause the language model 111A to determine whether the appeal phrases are similar. In this case, the classification unit 103A may generate a prompt instructing the language model 111A to determine whether the input appeal phrases are similar, and input the generated prompt to the language model 111A. Such a prompt can be generated, for example, by inputting each appeal phrase to be judged for similarity in the {appeal phrase} portion of a template such as "Please answer 'YES' if {appeal phrase} and {appeal phrase} are similar, and 'NO' if they are not similar." Note that the format in which the similarity judgment result is output is arbitrary.

[0077] Alternatively, the classification unit 103A may cause the language model 111A to classify the appeal phrases. In this case, the classification unit 103A may generate a prompt to instruct the language model 111A to classify the appeal phrases, and input the prompt together with a list of appeal phrases to be classified to the language model 111A. Such a prompt may be, for example, a standard phrase such as "Please classify the phrases shown in the list according to their contents."

[0078] The representative phrase of each category may be a phrase that represents each appeal phrase of that category, and the method for determining the representative phrase may be arbitrary. For example, the display control unit 104A may select one appeal phrase from among the appeal phrases of the same category and set the representative phrase of that category. Also, for example, the display control unit 104A may have the user input the representative phrase. Furthermore, for example, the display control unit 104A may input a list of appeal phrases of the same category into the language model 111A, output a generic name for those phrases, and set the output generic name as the representative phrase.

[0079] (Processing flow) The flow of processing executed by the information processing device 1A will be described with reference to Fig. 6. Fig. 6 is a flow diagram showing the flow of processing executed by the information processing device 1A. The flow of Fig. 6 includes each step of the analysis method according to this exemplary embodiment.

[0080] In S11, the data acquisition unit 101A accepts input of a document describing each analysis target. The document may be input in any manner. For example, the data acquisition unit 101A may accept input of the document via the input unit 13A, or may accept input of the document via the communication unit 12A. Furthermore, the data acquisition unit 101A may accept input of an image showing the analysis target instead of or together with the document describing the analysis target.

[0081] In S11, the data acquisition unit 101A may accept the specification of attributes to be analyzed instead of accepting the input of a document. In this case, the data acquisition unit 101A may acquire the attributes by searching for documents describing products, services, etc. having the input attributes or images related to the products, services, etc. This allows the user to very easily grasp the appealing points to consumers of products, services, etc. having attributes in which the user is interested.

[0082] For example, a user who is developing a new healthcare product can specify that the attributes to be analyzed are healthcare-related products and services. This allows the data acquisition unit 101A to automatically collect documents, etc. related to various healthcare-related products and services and acquire feature data from the collected documents, etc. This allows the user to quickly understand the appealing points of existing healthcare-related products and services to consumers, which can help with the creation of new products, or in other words, support the decision-making of users considering the specifications, etc. of new products.

[0083] In S12 (data acquisition process), the data acquisition unit 101A acquires, from one of the documents input in S11, one or more characteristic words or phrases that indicate characteristics of the analysis target and are contained in the document, as characteristic data. Note that, if an image showing the analysis target is input in S11, the data acquisition unit 101A generates characteristic words or phrases that indicate characteristics of the analysis target from the input image, and acquires the characteristic words or phrases as characteristic data.

[0084] In S13 (phrase acquisition process), the phrase acquisition unit 102A uses the feature data acquired in S12 to cause the language model 111A to generate appeal phrases that represent the points that the analysis target appeals to a predetermined target person. Note that one appeal phrase may be generated for one piece of feature data (for example, one feature phrase), or multiple appeal phrases may be generated for one piece of feature data.

[0085] In S14, the data acquisition unit 101A determines whether acquisition of feature words and generation of appeal words have been completed for all documents input in S11. If the determination in S14 is NO, the process returns to S12, and the data acquisition unit 101A acquires feature words from documents for which feature words have not been acquired. On the other hand, if the determination in S14 is YES, the process proceeds to S15.

[0086] In S15, the classification unit 103A classifies the multiple claim phrases generated by repeating the processes of S12 to S14 according to their contents. Furthermore, the display control unit 104A determines a representative phrase for each classification. Note that if the number of generated claim phrases is small (for example, if the number of generated claim phrases is equal to or smaller than a predetermined lower limit), the process of S15 may be omitted.

[0087] In S16, the display control unit 104A generates a graph in which nodes representing each analysis target and nodes representing appeal phrases generated for the analysis target (generated by repeating the processes of S12 to S14) are connected by edges. At this time, the display control unit 104A sets the nodes corresponding to appeal phrases classified in the same category in S15 as nodes of representative phrases representing each appeal phrase of the category.

[0088] For example, suppose that a claim phrase "low price" in one analysis target and a claim phrase "worth more than the price" in another analysis target are classified into the same category, and the representative phrase of this category is determined to be "good cost performance." In this case, the display control unit 104A sets the claim phrase nodes associated with these analysis target nodes to all be "good cost performance."

[0089] In S17, the display control unit 104A displays the graph generated in S16 on the display device. This completes the process of FIG. 6. The acquired appeal phrases do not necessarily need to be presented to the user. For example, the phrase acquisition unit 102A may store the acquired appeal phrases in a storage device such as the storage unit 11A as attribute information corresponding to the appeal phrases. Such attribute information can be used for classifying the analysis target, etc.

[0090] In addition, in S13, the word / phrase acquisition unit 102A may cause the language model 111A to select, from among the candidate appeal phrases, those that match the feature data acquired in S12. In this case, the acquired appeal phrases will not be more diverse than necessary, so the process of S15 may be omitted.

[0091] As described above, the word / phrase acquiring unit 102A may input, as candidates, appeal phrases generated or selected for each analysis target included in a group of analysis targets with common attributes, together with the feature data, to the language model 111A, and select appeal phrases that match the feature data from the input candidates. This not only provides the effect of the information processing device 1, but also provides the effect of preventing the total number of appeal phrases from diverging, without requiring a person to set candidates in advance.

[0092] When the above configuration is adopted, first, the processes from S11 to S14 are performed for the group of analysis targets for which candidates are to be generated, and the appeal phrases generated by these processes are set as candidates. Alternatively, after the processes from S11 to S14 are performed, the process from S15 may be performed again, and each representative phrase determined in S15 may be set as a candidate. After determining candidates in this way, appeal phrases may be selected from the candidates for the remaining analysis targets in S13. Note that, as mentioned above, new appeal phrases may be generated for analysis targets for which no suitable candidates exist.

[0093] [Modification] The execution entity of each process described in the above exemplary embodiment is arbitrary and is not limited to the above example. For example, a system having the same functions as the information processing devices 1 and 1A can be constructed using multiple devices that can communicate with each other. Furthermore, the execution entity of each process shown in the flowchart of FIG. 6 may be a single device (which can also be called a processor) or multiple devices (which can also be called processors).

[0094] [Software implementation example] Some or all of the functions of the information processing device 1, 1A may be realized by hardware such as an integrated circuit (IC chip), or may be realized by software.

[0095] In the latter case, the information processing device 1 or 1A is realized by, for example, a computer that executes instructions of a program, which is software that realizes each function. An example of such a computer (hereinafter referred to as computer C) is shown in Fig. 7. Fig. 7 is a block diagram showing the hardware configuration of computer C that functions as information processing device 1 or 1A.

[0096] The computer C includes at least one processor C1 and at least one memory C2. The memory C2 stores a program (analysis program) P for causing the computer C to operate as the information processing device 1 or 1A. In the computer C, the processor C1 reads and executes the program P from the memory C2, thereby realizing each function of the information processing device 1 or 1A.

[0097] The processor C1 may be, for example, a central processing unit (CPU), a graphic processing unit (GPU), a digital signal processor (DSP), a micro processing unit (MPU), a floating point number processing unit (FPU), a physics processing unit (PPU), a tensor processing unit (TPU), a quantum processor, a microcontroller, or a combination thereof. The memory C2 may be, for example, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), or a combination thereof.

[0098] The computer C may further include a RAM (Random Access Memory) for expanding the program P during execution and for temporarily storing various data. The computer C may also include a communication interface for transmitting and receiving data to and from other devices. The computer C may also include an input / output interface for connecting input / output devices such as a keyboard, mouse, display, and printer.

[0099] Furthermore, the program P can be recorded on a non-transitory tangible recording medium M that can be read by the computer C. Such a recording medium M can be, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit. The computer C can acquire the program P via such a recording medium M. The program P can also be transmitted via a transmission medium. Such a transmission medium can be, for example, a communication network or broadcast waves. The computer C can also acquire the program P via such a transmission medium.

[0100] Furthermore, each of the above functions of the information processing device 1 or 1A may be realized by a single processor provided in a single computer, by multiple processors provided in a single computer working together, or by multiple processors provided in each of multiple computers working together. Furthermore, a program for causing the information processing device 1 or 1A to realize each of the above functions may be stored in a single memory provided in a single computer, or may be distributed and stored in multiple memories provided in a single computer, or may be distributed and stored in multiple memories provided in each of multiple computers.

[0101] [Appendix 1] This disclosure includes the techniques described in the following appendices. However, the present invention is not limited to the techniques described in the following appendices, and various modifications are possible within the scope of the claims.

[0102] [Appendix A] (Appendix A1) An information processing device comprising: a data acquisition means for acquiring feature data representing the features of an object of analysis; and a phrase acquisition means for causing a machine-learned language model to generate appealing phrases representing the points that the object of analysis appeals to a predetermined target audience using the feature data, or for causing the language model to select from among the candidate appealing phrases those that match the feature data.

[0103] (Appendix A2) The information processing device described in Appendix A1, wherein the data acquisition means acquires characteristic words and phrases that indicate characteristics of the analysis target and are included in a document describing the analysis target as the characteristic data, and the phrase acquisition means generates the appeal phrase by inputting the characteristic words and phrases into the language model, or inputs the characteristic words and phrases and the candidates into the language model, thereby selecting from the candidates those that match the characteristic words and phrases as the appeal phrases.

[0104] (Appendix A3) The information processing device described in Appendix A1 or A2, wherein the data acquisition means generates characteristic words and phrases that indicate characteristics of the analysis target from an image showing the analysis target and acquires the characteristic words and phrases as the characteristic data, and the phrase acquisition means generates the appeal phrase by inputting the characteristic words and phrases into the language model, or inputs the characteristic words and phrases and the candidates into the language model and selects from the candidates those that match the characteristic words and phrases as the appeal phrases.

[0105] (Appendix A4) The information processing device according to any one of Appendices A1 to A3, wherein the phrase acquisition means inputs, together with the feature data, the appeal phrases generated or selected for each analysis target included in a part of a plurality of groups of analysis targets having common attributes into the language model as the candidates, and selects, from the input candidates, one that matches the feature data as the appeal phrase.

[0106] (Appendix A5) An information processing device described in any of Appendices A1 to A4, comprising a display control means for displaying on a display device a graph in which edges connect nodes indicating the object of analysis and nodes indicating the appeal phrases generated or selected for the object of analysis.

[0107] (Appendix A6) The information processing device according to Appendix A5 includes a classification means for classifying the generated or selected plurality of appeal phrases according to the content of the appeal phrases, and the display control means displays a representative phrase representing each appeal phrase of the category at each node corresponding to the appeal phrases of the same category.

[0108] [Appendix B] (Appendix B1) An analysis method including: a data acquisition process in which at least one processor acquires feature data representing the characteristics of an object of analysis; and a phrase acquisition process in which, using the feature data, a machine-learned language model generates appealing phrases representing the points that the object of analysis appeals to a predetermined target audience, or selects from among the candidate appealing phrases those that match the feature data, using the language model.

[0109] (Appendix B2) The analysis method described in Appendix B1, wherein in the data acquisition process, the at least one processor acquires characteristic words and phrases that indicate characteristics of the analysis target and are included in a document describing the analysis target as the characteristic data, and in the phrase acquisition process, the at least one processor inputs the characteristic words and phrases into the language model to generate the appeal phrase, or inputs the characteristic words and phrases and the candidates into the language model to select from the candidates those that match the characteristic words and phrases as the appeal phrase.

[0110] (Appendix B3) The analysis method described in Appendix B1 or B2, wherein in the data acquisition process, the at least one processor generates characteristic words and phrases that indicate the characteristics of the analysis target from an image showing the analysis target and acquires the characteristic words and phrases as the characteristic data, and in the phrase acquisition process, the at least one processor inputs the characteristic words and phrases into the language model to generate the appeal words and phrases, or inputs the characteristic words and phrases and the candidates into the language model to select from the candidates those that match the characteristic words and phrases as the appeal words and phrases.

[0111] (Appendix B4) The analysis method according to any one of Appendices B1 to B3, wherein in the phrase acquisition process, the at least one processor inputs the appeal phrases generated or selected for each analysis target included in a part of a plurality of groups of analysis targets having common attributes together with the feature data into the language model as the candidates, and selects the appeal phrase that matches the feature data from the input candidates as the appeal phrase.

[0112] (Appendix B5) An analysis method described in any of Appendices B1 to B4, including a display control process in which the at least one processor displays on a display device a graph in which edges connect nodes representing the analysis target and nodes representing the appeal phrases generated or selected for the analysis target.

[0113] (Appendix B6) The analysis method described in Appendix B5 includes a classification process in which the at least one processor classifies the generated or selected multiple claim phrases according to the content of the claim phrases, and in the display control process, the at least one processor displays a representative phrase that represents each claim phrase of the category at each node corresponding to the claim phrases of the same category.

[0114] [Appendix C] (Appendix C1) An analysis program that causes a computer to function as a data acquisition means that acquires feature data that represents the characteristics of an object of analysis, and as a phrase acquisition means that uses the feature data to generate appealing phrases that represent the points that the object of analysis appeals to a specified target audience in a machine-learned language model, or that causes the language model to select from among the candidate appealing phrases those that match the feature data.

[0115] (Appendix C2) The analysis program described in Appendix C1, wherein the data acquisition means acquires characteristic words and phrases that indicate characteristics of the analysis target and are included in a document that describes the analysis target as the characteristic data, and the phrase acquisition means generates the appeal phrase by inputting the characteristic words and phrases into the language model, or selects from the candidates those that match the characteristic words and phrases as the appeal phrase by inputting the characteristic words and phrases and the candidates into the language model.

[0116] (Appendix C3) The analysis program described in Appendix C1 or C2, wherein the data acquisition means generates characteristic words and phrases that indicate characteristics of the analysis target from an image showing the analysis target and acquires the characteristic words and phrases as the characteristic data, and the phrase acquisition means generates the appeal words and phrases by inputting the characteristic words and phrases into the language model, or inputs the characteristic words and phrases and the candidates into the language model and selects from the candidates those that match the characteristic words and phrases as the appeal words and phrases.

[0117] (Appendix C4) The analysis program according to any one of Appendices C1 to C3, wherein the phrase acquisition means inputs, together with the feature data, the appeal phrases generated or selected for each analysis target included in a part of a plurality of groups consisting of the analysis targets having common attributes, into the language model as the candidates, and selects, from the input candidates, one that matches the feature data as the appeal phrase.

[0118] (Appendix C5) An analysis program described in any of Appendices C1 to C4, which causes the computer to function as a display control means for displaying on a display device a graph in which edges connect nodes representing the analysis object and nodes representing the appeal phrases generated or selected for the analysis object.

[0119] (Appendix C6) The analysis program described in Appendix C5 causes the computer to function as a classification means for classifying the generated or selected plurality of appeal phrases according to the content of the appeal phrases, and the display control means displays a representative phrase representing each appeal phrase of the category at each node corresponding to the appeal phrases of the same category.

[0120] [Appendix D] (Appendix D1) An information processing device comprising at least one processor, the at least one processor executing a data acquisition process for acquiring feature data representing the features of an object of analysis, and a phrase acquisition process for causing a machine-learned language model to generate appealing phrases representing the points that the object of analysis appeals to a predetermined target audience using the feature data, or for causing the language model to select from among the candidate appealing phrases those that match the feature data.

[0121] The information processing device may further include a memory, and the memory may store a program for causing the at least one processor to execute each of the processes.

[0122] (Appendix D2) The information processing device described in Appendix D1, wherein in the data acquisition process, the at least one processor acquires characteristic words and phrases that indicate characteristics of the analysis target and are included in a document describing the analysis target as the characteristic data, and in the phrase acquisition process, the at least one processor generates the appeal phrase by inputting the characteristic words and phrases into the language model, or inputs the characteristic words and phrases and the candidates into the language model, thereby selecting from the candidates those that match the characteristic words and phrases as the appeal phrases.

[0123] (Appendix D3) The information processing device described in Appendix D1 or D2, wherein in the data acquisition process, the at least one processor generates characteristic words and phrases that indicate characteristics of the analysis target from an image showing the analysis target and acquires the characteristic words and phrases as the characteristic data, and in the phrase acquisition process, the at least one processor generates the appeal phrase by inputting the characteristic words and phrases into the language model, or inputs the characteristic words and phrases and the candidates into the language model and selects from the candidates those that match the characteristic words and phrases as the appeal phrases.

[0124] (Appendix D4) The information processing device according to any one of appendices D1 to D3, wherein in the phrase acquisition process, the at least one processor inputs the appeal phrase generated or selected for each analysis target included in a part of a plurality of groups of analysis targets having common attributes together with the feature data into the language model as the candidate, and selects the appeal phrase that matches the feature data from the input candidates as the appeal phrase.

[0125] (Appendix D5) An information processing device described in any of Appendices D1 to D4, wherein the at least one processor executes a display control process to display on a display device a graph in which edges connect nodes indicating the analysis target and nodes indicating the appeal phrases generated or selected for the analysis target.

[0126] (Appendix D6) The information processing device described in Appendix D5, wherein the at least one processor executes a classification process to classify the generated or selected multiple appeal phrases according to the content of the appeal phrases, and in the display control process, the at least one processor displays a representative phrase representing each appeal phrase of the classification at each node corresponding to the appeal phrases of the same classification.

[0127] [Appendix E] A non-transient recording medium that records an analysis program that causes a computer to execute a data acquisition process that acquires feature data that represents the characteristics of an object of analysis, and a phrase acquisition process that uses the feature data to generate appealing phrases that represent the points that the object of analysis appeals to a specified target audience in a machine-learned language model, or to select from among the candidate appealing phrases those that match the feature data in the language model. [Explanation of symbols]

[0128] 1. Information processing equipment 101 Data acquisition unit (data acquisition means) 102 Word acquisition unit (word acquisition means) 1A Information processing equipment 101A data acquisition unit (data acquisition means) 102A Phrase acquisition unit (phrase acquisition means) 103A Classification section (classification means) 104A Display control unit (display control means) 111A Language Model

Claims

1. data acquisition means for acquiring feature data representing features of an analysis target; and a phrase acquisition means for causing a machine-learned language model to generate appealing phrases that express the points that the analysis subject appeals to a specified target audience using the feature data, or for causing the language model to select from among the candidate appealing phrases those that match the feature data.

2. the data acquisition means acquires, as the feature data, feature words and phrases that indicate features of the analysis target and are included in a document that describes the analysis target; 2. The information processing device according to claim 1, wherein the phrase acquisition means generates the appeal phrase by inputting the characteristic phrase into the language model, or inputs the characteristic phrase and the candidates into the language model, thereby selecting one of the candidates that matches the characteristic phrase as the appeal phrase.

3. the data acquisition means generates characteristic phrases that indicate characteristics of the analysis target from an image that shows the analysis target, and acquires the characteristic phrases as the characteristic data; 2. The information processing device according to claim 1, wherein the phrase acquisition means generates the appeal phrase by inputting the characteristic phrase into the language model, or inputs the characteristic phrase and the candidates into the language model, thereby selecting one of the candidates that matches the characteristic phrase as the appeal phrase.

4. 4. The information processing device according to claim 1, wherein the phrase acquisition means inputs, together with the feature data, the appeal phrases generated or selected for each analysis target included in a part of a plurality of groups consisting of the analysis targets having common attributes, as the candidates to the language model, and selects, from the input candidates, one that matches the feature data as the appeal phrase.

5. 4. The information processing device according to claim 1, further comprising a display control means for causing a display device to display a graph in which edges connect nodes representing the analysis target and nodes representing the appeal phrases generated or selected for the analysis target.

6. A classification means for classifying the generated or selected plurality of appeal phrases according to the content of the appeal phrases, The information processing device according to claim 5 , wherein the display control means displays a representative phrase representing each of the appeal phrases of the same category in each node corresponding to the appeal phrase of the same category.

7. At least one processor a data acquisition process for acquiring feature data representing the features of the analysis target; an analysis method including: a phrase acquisition process that uses the feature data to generate appealing phrases that express the points that the analysis subject appeals to a specified target audience using a machine-learned language model, or selects from among the candidate appealing phrases those that match the feature data using the language model.

8. Computer, data acquisition means for acquiring feature data representing the features of the analysis target; and An analysis program that functions as a phrase acquisition means that uses the feature data to generate appealing phrases that express the points that the analysis subject appeals to a specified target audience in a machine-learned language model, or that causes the language model to select from the candidate appealing phrases those that match the feature data.

Images