Information processor, method for processing information, and program

Abstract

To provide an information processor which can generate a proper content on the basis of the result of analysis of a browsing log.SOLUTION: An information processor includes acquisition means for: acquiring a log indicating that a document output by a model generating a document including content instructed by an instruction sentence has been viewed; specifying means for specifying a characteristic of the content having a predetermined degree of attention on the basis of the log; and creation means for creating the instruction sentence so as to cause the document to be generated in accordance with the specified characteristic.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to techniques for content feedback analysis. [Background technology]

[0002] Conventionally, there is known a technology for acquiring a viewing log of content by a viewing user and providing content that matches the viewing user's preferences. For example, Patent Document 1 describes a technology for changing the placement, arrangement, etc. of content based on statistics obtained by statistically processing the viewing log and viewing correlation. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-362539 Summary of the Invention [Problem to be solved by the invention]

[0004] However, even with the technique of Patent Document 1, the results of analyzing the viewing logs cannot always be appropriately reflected in content generation.

[0005] An object of the present disclosure is to provide an information processing device capable of generating appropriate content based on the analysis results of a viewing log. [Means for solving the problem]

[0006] According to one aspect of the present disclosure, there is provided an information processing device, an acquisition means for acquiring a log indicating a record of viewing of a document output by a model that generates a document including the content specified by the instruction sentence; a specifying means for specifying characteristics of the content having a predetermined level of attention based on the log; a generating means for generating the instruction sentence so as to generate the document in accordance with the specified characteristics; Equipped with.

[0007] In another aspect of the present disclosure, an information processing method includes: 1. A computer-implemented information processing method, comprising: A log showing the record of the document being viewed is acquired for the document output by the model that generates a document containing the content specified in the instruction sentence, Identifying characteristics of the content that have a predetermined level of interest based on the log; The instructions are created to generate the document according to the identified characteristics.

[0008] In yet another aspect of the disclosure, a program includes: A log showing the record of the document being viewed is acquired for the document output by the model that generates a document containing the content specified in the instruction sentence, Identifying characteristics of the content that have a predetermined level of interest based on the log; The computer is caused to perform a process of creating the instructions to generate the document according to the identified characteristics.

[0009] According to yet another aspect of the present disclosure, there is provided an information processing device, a relationship specifying means for specifying a relationship between the attention level of a document and a feature quantity representing a feature of content included in the document; a feature quantity specifying means for specifying a feature quantity having a desired degree of attention based on the relationship; a content creation means for creating content having characteristics represented by the identified feature amount; Equipped with. [Effects of the Invention]

[0010] According to the present disclosure, it is possible to generate appropriate content based on the analysis results of the viewing log. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a diagram illustrating an overall configuration of a content analysis system according to the present disclosure. [Figure 2]FIG. 1 is a block diagram illustrating a hardware configuration of an information processing device according to the present disclosure. [Figure 3] FIG. 1 is a block diagram illustrating a functional configuration of an information processing device according to the present disclosure. [Figure 4] An example of a label is shown below. [Figure 5] An example of a label assignment method will be shown below. [Figure 6] An example of a title analysis is shown below. [Figure 7] This is an example of a scatter plot showing the results of a title analysis. [Figure 8] An example of analysis of the text is shown below. [Figure 9] This is an example of the analysis results of the main text presented in a scatter plot. [Figure 10] 10 is a flowchart of an article generation process. [Figure 11] 10 is a flowchart of a viewing log analysis process. [Figure 12] FIG. 10 is a block diagram showing a functional configuration of another information processing device according to the present disclosure. [Figure 13] Here are other examples of analysis in the text. [Figure 14] This is an example of the analysis results of the main text presented in a scatter plot. [Figure 15] FIG. 10 is a block diagram showing a functional configuration of another information processing device according to the present disclosure. [Figure 16] 10 is a flowchart of a process performed by another information processing device according to the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, preferred embodiments of the present disclosure will be described with reference to the drawings. First Embodiment [Overall configuration] 1 shows the overall configuration of a content analysis system to which an information processing device according to the present disclosure is applied. The content analysis system 1 includes a content publishing system 2, an information processing device 10, a business operator's terminal device 5, and a viewer's terminal device 20. It is assumed that there are multiple terminal devices 5 and multiple terminal devices 20.

[0013] The content publishing system 2 is a system that publishes content prepared by a business operator to a large number of viewers. The content publishing system 2 of this embodiment provides a website that can be accessed by a terminal device 20. Hereinafter, the website provided by the content publishing system 2 will also be simply referred to as a "website." The content publishing system 2 is configured, for example, by a server device.

[0014] The content in this embodiment is, for example, a piece of writing such as a company's brand story, a column article, or a news article, and hereinafter may be simply referred to as an "article."

[0015] The information processing device 10 generates an article of a business operator based on a request from the terminal device 5 and transmits it to the content publishing system 2. The information processing device 10 also acquires an article viewing log from the content publishing system 2 and analyzes trends in articles that have interested viewers. The information processing device 10 communicates with the content publishing system 2 and the terminal device 5 via a network such as the Internet.

[0016] The terminal device 5 is operated by a person in charge (hereinafter also referred to as a "user") of a business (company, etc.) and is used to make a request to create an article to be published on a website. The terminal device 5 is configured, for example, by a personal computer or a tablet terminal, and communicates with the information processing device 10 via a network such as the Internet.

[0017] The terminal device 20 is operated by a viewer and used to view articles published by businesses on their websites. The terminal device 20 is configured, for example, by a personal computer or a tablet terminal, and communicates with the content publishing system 2 via a network such as the Internet.

[0018] [Hardware configuration] 2 is a block diagram showing the hardware configuration of an information processing device 10 according to the first embodiment. As shown in the figure, the information processing device 10 includes an interface (I / F) 11, a processor 12, a memory 13, a recording medium 14, and a database (DB) 15.

[0019] The I / F 11 communicates with the content publishing system 2 and the terminal device 5 via a network such as the Internet.

[0020] The processor 12 is a computer such as a CPU (Central Processing Unit), and controls the entire information processing device 10 by executing a pre-prepared program. The processor 12 may be a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating Point number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination thereof. The processor 12 executes article generation processing and viewing log analysis processing, which will be described later.

[0021] The memory 13 is configured by a ROM (Read Only Memory), a RAM (Random Access Memory), etc. The memory 13 is also used as a working memory while the processor 12 is executing various processes.

[0022] The recording medium 14 is a non-volatile, non-transitory recording medium such as a disk-shaped recording medium or a semiconductor memory, and is configured to be detachable from the information processing device 10. The recording medium 14 records various programs to be executed by the processor 12. When the information processing device 10 executes various processes, the programs recorded on the recording medium 14 are loaded into the memory 13 and executed by the processor 12.

[0023] The DB 15 stores data used to generate articles, such as questionnaires and syntax tables, which will be described later, etc. The DB 15 also stores a viewing log, which will be described later.

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

[0025] [Function Configuration] 3 is a block diagram showing the functional configuration of the information processing device 10 of the first embodiment. Functionally, the information processing device 10 includes a generation unit 100 and an analysis unit 200 in addition to the DB 15 described above.

[0026] (Generation part) First, the generation unit 100 will be described. The generation unit 100 generates articles for businesses and transmits them to the content publishing system 2. The generation unit 100 includes an information acquisition unit 101, a syntax determination unit 102, a directive generation unit 103, an article generation unit 104, and an article editing unit 105.

[0027] The generation unit 100 transmits the business operator's profile and the questionnaire input screen to the terminal device 5 via the I / F 11. The user operates the terminal device 5 to input the business operator's profile (hereinafter also referred to as "business operator profile") and questionnaire responses (hereinafter also referred to as "questionnaire responses"), and transmits them to the information processing device 10.

[0028] The generation unit 100 receives the business profile and the questionnaire responses from the terminal device 5. The business profile and the questionnaire responses are input to the information acquisition unit 101. The information acquisition unit 101 outputs the business profile and the questionnaire responses to the syntax determination unit 102 and the instruction sentence generation unit 103.

[0029] The syntax determination unit 102 selects a syntax to be used for generating a directive statement from among multiple syntaxes stored in the DB 15. A syntax is a template of a directive statement for instructing the generation of an article. For example, the DB 15 stores a syntax table that associates answers to certain items in a questionnaire with syntaxes. The syntax determination unit 102 refers to the syntax table and selects a syntax that corresponds to the answer to a certain item among the questionnaire answers input from the information acquisition unit 101 as the syntax to be used for generating a directive statement. The syntax determination unit 102 outputs the selected syntax to the directive statement generation unit 103.

[0030] The instruction sentence generation unit 103 generates an instruction sentence by inputting the company profile and the questionnaire responses into corresponding input fields of the syntax. The instruction sentence generation unit 103 outputs the generated instruction sentence to the article generation unit 104.

[0031] The article generation unit 104 generates an article including a title and a main text based on the instruction statement. Specifically, the article generation unit 104 inputs the instruction statement to the generation AI and obtains an article from the generation AI as a response to the instruction statement. The article generation unit 104 uses, for example, a large-scale language model (LLM) as the generation AI. The LLM is a natural language processing model that uses a large amount of text data to learn the relationships between words in a sentence. The LLM generates related strings related to an input target string from the input target string. An example of an LLM is ChatGPT by OpenAI. The article generation unit 104 outputs the generated article to the article editing unit 105.

[0032] The article editing unit 105 generates an editing screen for editing an article and transmits it to the terminal device 5. Then, the article editing unit 105 accepts editing operations for the article from the terminal device 5. For example, a user can perform editing operations such as correcting text, inserting images, and changing fonts on an article displayed on the display of the terminal device 5. The article editing unit 105 performs editing processing for the article in accordance with the editing operations of the user.

[0033] Furthermore, the article editing unit 105 accepts a publishing operation for an article from the terminal device 5. The publishing operation is an operation for indicating the intention to publish the article to others. When a user decides to publish an article on a website, the user performs a publishing operation for the article. When the article editing unit 105 accepts a publishing operation from the terminal device 5, the article editing unit 105 transmits the article to the content publishing system 2 and outputs the article to the analysis unit 200.

[0034] The content publishing system 2 publishes articles on a website. A viewer operates a terminal device 20 to access the website provided by the content publishing system 2 and view the articles of the business. The website provided by the content publishing system 2 has multiple hierarchies, for example, a top page and other web pages. The top page displays the titles of the articles of the business in a list or the like. A viewer selects the title of the article they wish to view from the top page and moves to the page of the relevant article.

[0035] The content publishing system 2 also records an access history from the terminal device 20. The access history includes, for example, information that uniquely identifies the viewer, the access date and time, the accessed page, and the duration of stay on the accessed page.

[0036] (Analysis Department) Next, the analysis unit 200 will be described. The analysis unit 200 analyzes the browsing trends of viewers based on the viewing logs of articles published on the website. Specifically, the analysis unit 200 analyzes the browsing trends of viewers for three elements of an article: the "title," "body," and "web design." For example, the analysis unit 200 analyzes the correlation between the characteristics of the "title" and "body" of an article and the browsing behavior. The analysis unit 200 also performs AB testing for the "web design" of an article and compares the elements of the web page of each article to analyze effective web design. The analysis of the "title" and "body" of an article will be described below.

[0037] The analysis unit 200 includes a label assignment unit 201 , a log recording unit 202 , an article analysis unit 203 , and a correction unit 204 .

[0038] The labeling unit 201 assigns labels to articles input from the article editing unit 105. A label is information for representing the characteristics of an article. Specifically, the labeling unit 201 assigns labels such as those shown in FIG. 4 to the "title" and "body" of an article.

[0039] 4, the label assignment unit 201 assigns a title label including "number of characters_medium", "text tone_joy", "presence of numbers_no", "article subject genre_izakaya", and "image genre_products" to article A in order to represent the characteristics of the title of article A. Also, in FIG. 4, the label assignment unit 201 assigns a body label including "number of characters_medium", "text tone_joy", "target age group_20s", "target gender_female", "article subject genre_cafe", "image genre_products", and "image size_medium" to article A in order to represent the characteristics of the body of article A.

[0040] FIG. 5 shows an example of a label assignment method. FIG. 5(A) is an example of a title label assignment method, and FIG. 5(B) is an example of a body label assignment method. As shown in FIGS. 5(A) and 5(B), a "title label" and a "body label" include a label item and a label value. The label items are determined in advance, and the label assignment unit 201 generates a title label and a body label by determining the label value corresponding to the label item.

[0041] The "determination data" in Figures 5(A) and 5(B) is data used to determine the value of a label, and includes articles and syntax. Articles are articles published on a website. Syntax is syntax used to generate the article. Note that the label assignment unit 201 may use, in addition to syntax, business profiles, questionnaire responses, instructional statements, and the like used to generate the article as determination data. For label items where the determination data is an article, the label assignment unit 201 determines the label value based on the title and text of the article published on the website. For label items where the determination data is syntax, the label assignment unit 201 determines the label value based on data such as syntax used to generate the article.

[0042] The "determination method" in Figures 5(A) and 5(B) refers to a method for determining the label value, and includes a system-based determination method and a generation AI-based determination method. The system determines the label value from the determination data according to predetermined criteria. For example, the label assignment unit 201 may determine the number of characters or image size as large, medium, or small based on a predetermined threshold. The label assignment unit 201 may also perform a character search of the determination data based on predetermined words, etc., to determine the presence or absence of numbers, each genre, and each target.

[0043] The generative AI inputs judgment data into the generative AI to determine the label value. In Figures 5(A) and 5(B), the generative AI includes an emotion estimation model and an image classification model. The emotion estimation model is a natural language processing model that takes a sentence as input and classifies the input sentence into one of the tones of joy, anger, sadness, or happiness, and uses BERT (Bidirectional Encoder Representations from Transformers) or similar. The image classification model is an AI model that takes an image as input and determines the genre (e.g., person, product, store, etc.) to which the input image belongs, and is composed of a CNN (Convolutional Neural Network) or similar.

[0044] Note that the title label and the body label shown in Fig. 4 are examples and are not limited to these. Also, the labeling method shown in Fig. 5 is an example and is not limited to these.

[0045] Returning to FIG. 3, the labeling unit 201 outputs the results of labeling each article to the article analysis unit 203.

[0046] The log recording unit 202 receives the access history from the content publishing system 2. Based on the access history, the log recording unit 202 calculates the total number of accesses and the total duration of stay for each article. Then, the log recording unit 202 associates information that uniquely identifies the article with the total number of accesses (hereinafter also referred to as "number of views") and the total duration of stay (hereinafter also referred to as "viewing time"), and records these in DB 15. The information that associates information that uniquely identifies the article with the number of views and viewing time will be referred to as a "viewing log" hereinafter.

[0047] The article analysis unit 203 analyzes the title and text of each article based on the labeling results of each article input from the labeling unit 201 and the viewing log acquired from the DB 15 .

[0048] (1) Title Analysis The article analysis unit 203 analyzes the characteristics of titles that have viewing trends based on the title label of each article and the number of views of each article. Figures 6 and 7 show examples of correlation analysis by the article analysis unit 203. As shown in Figure 6, the article analysis unit 203 analyzes the relationship between each label included in the title label and the number of views. Then, the article analysis unit 203 expresses the analysis results in a scatter diagram as shown in Figure 7.

[0049] The analysis table 60 in Fig. 6 includes analysis data 60a and analysis results 60b. The analysis data 60a includes the label assignment results for each article and the number of views for each article. In Fig. 6, there are articles A to P, and the labels assigned to each article are flagged with "1."

[0050] The analysis result 60b includes "number of labels," "label × number of views," and "average number of views." "Number of labels" indicates the number of articles that include the label in that column. "Label × number of views" indicates the total number of views for that article. "Average number of views" is the value obtained by dividing "label × number of views" by "number of labels," and represents the average number of views for the label in that column. For example, for "small character count" in Figure 6, "6" is entered in "number of labels," indicating that there are six articles (articles B, D, I, J, K, and O) with the title "small character count." Furthermore, "687," which is the total number of views for the six articles, is entered in "label × number of views." Furthermore, "average number of views" is entered as "114.5," which represents the average number of views for "small character count."

[0051] Then, the article analysis unit 203 analyzes the labels whose average number of views is equal to or greater than a predetermined threshold TH1 as labels that have a high correlation with views (that is, characteristics of titles that tend to be viewed).

[0052] Next, the article analysis unit 203 expresses the analysis results of FIG. 6 in a scatter diagram as shown in FIG. 7. FIG. 7 is a scatter diagram with the average number of views on the vertical axis and the number of labels on the horizontal axis. In FIG. 7, the predetermined threshold TH1 is 100, and the article analysis unit 203 analyzes "writing tone_easy" and "number of characters_small" as labels that have a high correlation with views. The article analysis unit 203 outputs the analysis results to the correction unit 204.

[0053] (2) Analysis of the text The article analysis unit 203 analyzes the characteristics of body text that show viewing trends based on the body text label of each article, the number of views of each article, and the viewing time of each article. Figures 8 and 9 show examples of correlation analysis by the article analysis unit 203. As shown in Figure 8, the article analysis unit 203 analyzes the relationship between each label included in the body text label and the average viewing time. Then, the article analysis unit 203 expresses the analysis results in a scatter diagram as shown in Figure 9.

[0054] The analysis table 80 in Fig. 8 includes analysis data 80a and analysis results 80b. The analysis data 80a includes the label assignment results for each article, the number of views for each article, the viewing time for each article, and the average viewing time for each article (viewing time / number of views). In Fig. 8, there are articles A to P, and the label assigned to each article is flagged as "1."

[0055] The analysis result 80b includes the "number of labels," "label × (viewing time / number of views)," and "average viewing time." The "number of labels" indicates the number of articles that include the label in that column. The "label × (viewing time / number of views)" indicates the sum of the average viewing times (viewing time / number of views) of the articles. The "average viewing time" is the value obtained by dividing "label × (viewing time / number of views)" by the "number of labels," and represents the average viewing time of the labels in that column. For example, for "small character count" in FIG. 8, "5" is entered in the "number of labels," indicating that there are five articles (articles B, D, I, K, and O) whose main text is "small character count." Furthermore, "14.17," which is the sum of the average viewing times (viewing time / number of views) of the five articles, is entered in the "label × (viewing time / number of views)." Furthermore, "2.83," which represents the average viewing time of "small character count," is entered in the "average viewing time."

[0056] Then, the article analysis unit 203 analyzes the label whose average viewing time is equal to or greater than a predetermined threshold TH2 as a label that has a high correlation with viewing (that is, a characteristic of the text that tends to be viewed).

[0057] Next, the article analysis unit 203 expresses the analysis results of FIG. 8 in the form of a scatter diagram as shown in FIG. 9. FIG. 9 is a scatter diagram in which the vertical axis represents the average viewing time (s) and the horizontal axis represents the number of labels. In FIG. 9, the predetermined threshold TH2 is 7, and the article analysis unit 203 analyzes "image genre_products," "article subject genre_restaurants," "target gender_female," "target age group_50s," "number of characters_medium," and "text tone_relaxed" as labels highly correlated with viewing. The article analysis unit 203 outputs the analysis results to the correction unit 204.

[0058] The correction unit 204 changes various data used to generate articles based on the analysis results input from the article analysis unit 203. The various data include, for example, syntax, directives, questionnaires, and editing screens.

[0059] For example, the correction unit 204 may change the syntax based on the characteristics of titles that tend to be viewed. If the title label that has a high correlation with viewing is "Text tone_easy" and "Number of characters_small", the correction unit 204 adds instructions such as "Please create a title with a fun tone" or "Please keep the number of characters in the title to within XX characters" to the syntax.

[0060] Furthermore, the correction unit 204 may change the syntax based on the characteristics of the text that is likely to be viewed. If the text label that has a high correlation with viewing is "Text tone_easy" and "Number of characters_medium", the correction unit 204 adds instructions such as "Please write text with a fun tone" or "Please keep the number of characters in the text between XX and XX characters" to the syntax.

[0061] Furthermore, the correction unit 204 may modify the questionnaire based on the characteristics of the text that tend to be viewed. For example, the correction unit 204 may add items to the questionnaire for specifying the target age group of the article, the target gender of the article, the tone of the text, etc. In this case, the correction unit 204 adds a message to the questionnaire based on the characteristics of the text that tend to be viewed. For example, if the text labels that have a high correlation with viewing are "Text tone_easy," "Target gender_female," and "Target age_50s," the correction unit 204 adds messages such as "Text with a fun tone tends to be preferred" and "Articles aimed at women in their 50s tend to be viewed" to the questionnaire.

[0062] Furthermore, the correction unit 204 may add a message to the editing screen based on the characteristics of the title and the body text that tend to be viewed. For example, if the title label and the body text label that have a high correlation with viewing include "image genre_product", the correction unit 204 may add a message such as "If you include a product image in the title, it will be more likely to be viewed" or "If you include a product image in the body text, it will be more likely to be viewed" to the editing screen.

[0063] In the above configuration, the syntax determination unit 102, the instruction sentence generation unit 103, the article generation unit 104, the article editing unit 105, and the log recording unit 202 are examples of an acquisition means, the label assignment unit 201 and the article analysis unit 203 are examples of an identification means, and the correction unit 204 is an example of a creation means.

[0064] [Processing flow] Next, the article generation process by the generation unit 100 and the viewing log analysis process by the analysis unit 200 will be described.

[0065] (Article generation process) 10 is a flowchart of the article generation process performed by the generation unit 100. This process is realized by the processor 12 shown in FIG. 2 executing a program prepared in advance and operating as each element shown in FIG.

[0066] First, the generation unit 100 receives an operator profile and questionnaire responses from the terminal device 5. The operator profile and questionnaire responses are input to the information acquisition unit 101 (step S11). The information acquisition unit 101 outputs the operator profile and questionnaire responses to the syntax determination unit 102 and the instruction sentence generation unit 103.

[0067] Next, the syntax determination unit 102 selects a syntax to be used for generating a directive sentence from among the multiple syntaxes stored in the DB 15 (step S12). For example, the syntax determination unit 102 refers to a syntax table and selects a syntax corresponding to an answer to a certain item among the questionnaire answers input from the information acquisition unit 101 as a syntax to be used for generating a directive sentence. The syntax determination unit 102 outputs the selected syntax to the directive sentence generation unit 103.

[0068] Next, the instruction sentence generation unit 103 generates an instruction sentence by inputting the business profile and the questionnaire responses into corresponding input fields of the syntax (step S13). The instruction sentence generation unit 103 outputs the generated instruction sentence to the article generation unit 104.

[0069] Next, the article generation unit 104 generates an article including a title and a main text based on the instruction statement (step S14). Specifically, the article generation unit 104 inputs the instruction statement to the generation AI and obtains an article from the generation AI as a response to the instruction statement. The article generation unit 104 outputs the generated article to the article editing unit 105.

[0070] Next, the article editing unit 105 generates an editing screen for editing the article input from the article generating unit 104, and transmits it to the terminal device 5. Furthermore, the article editing unit 105 performs an editing process for the article in response to an editing operation by the user (step S15).

[0071] Next, when the article editing unit 105 receives a publishing operation from the terminal device 5, it transmits the article to the content publishing system 2 and outputs the article to the analysis unit 200 (step S16). Then, the process ends.

[0072] (Analysis of browsing logs) 11 is a flowchart of the viewing log analysis process performed by the analysis unit 200. This process is realized by the processor 12 shown in FIG. 2 executing a program prepared in advance and operating as each element shown in FIG.

[0073] First, the labeling unit 201 assigns labels to articles input from the article editing unit 105 (step S21). Specifically, the labeling unit 201 assigns labels to the "title" and "body" of the article as shown in Fig. 4. The labeling unit 201 outputs the labeling results for each article to the article analysis unit 203.

[0074] Next, the log recording unit 202 receives the access history from the content publishing system 2. Based on the access history, the log recording unit 202 obtains the number of views and the viewing time for each article. Then, the log recording unit 202 records a viewing log in which information that uniquely identifies the article is associated with the number of views and the viewing time in the DB 15 (step S22).

[0075] Next, the article analysis unit 203 performs an analysis based on the label assignment results for each article input from the label assignment unit 201 and the viewing log acquired from DB15 (step S23). Specifically, the article analysis unit 203 analyzes the characteristics of titles that tend to be viewed based on the title label of each article and the number of views of each article. The article analysis unit 203 also analyzes the characteristics of body text that tend to be viewed based on the body text label of each article and the number of views and viewing time of each article. The article analysis unit 203 outputs the analysis results to the correction unit 204.

[0076] Next, the correction unit 204 changes various data to be used in generating the article based on the analysis results input from the article analysis unit 203 (step S24), and the process then ends.

[0077] [Variations] Next, a modification of the first embodiment will be described.

[0078] The content publishing system 2 is not limited to publishing content on a website, and may also publish content on a predetermined social networking service (SNS), for example. In this case, the content publishing system 2 records the number of accesses to the content and reactions to the content (e.g., the number of "likes," the number of comments, the number of saves, the number of follows, etc.) instead of the access history. The analysis unit 200 then analyzes the characteristics of content that is likely to be viewed, using the number of accesses to the content and the reactions to the content.

[0079] Second Embodiment Next, a second embodiment will be described. The information processing device 10a of the second embodiment can create images and hashtags according to the content of an article. Furthermore, the information processing device 10a of the second embodiment can analyze hashtags in addition to analyzing the text of an article and images inserted in the article. This allows the user to understand images and hashtags that are likely to be viewed. Note that the overall configuration and hardware configuration are the same as those of the first embodiment, and therefore description thereof will be omitted.

[0080] [Function Configuration] 12 is a block diagram showing the functional configuration of an information processing device 10a according to the second embodiment. Functionally, the information processing device 10a includes a generation unit 100a and an analysis unit 200a.

[0081] (Generation part) First, the generation unit 100a will be described. The generation unit 100a is based on the generation unit 100 of the information processing device 10 according to the first embodiment, but further includes an additional element creation unit 106a. Note that the information acquisition unit 101a, syntax determination unit 102a, directive sentence generation unit 103a, and article generation unit 104a have the same configurations and operate in the same manner as the information acquisition unit 101, syntax determination unit 102, directive sentence generation unit 103, and article generation unit 104 of the generation unit 100, and therefore their description will be omitted.

[0082] The additional element creation unit 106a uses a generation AI to create images and hashtags related to the article.

[0083] Specifically, when a user wants to incorporate an image or a hashtag into an article, the user operates the terminal device 5 to send a request to create an image or a hashtag to the information processing device 10a. For example, the user sends a creation request to the information processing device 10a including the genre of the creation (i.e., whether to create an image or a hashtag) and the medium on which the article will be published (hereinafter also referred to as "publication target"). The creation request is input to the additional element creation unit 106a.

[0084] The additional element creation unit 106a selects a question and a syntax from the DB 15. The question is a question to the user, and includes questions about images and hashtags. The syntax selected by the additional element creation unit 106a is a syntax used to generate images and hashtags, and more specifically, is a template of an instruction statement for instructing the generation of images and hashtags.

[0085] For example, a table associating a disclosure target, a question sentence, and a syntax for each genre of creation is stored in DB 15. Based on a creation request from a user, additional element creation unit 106a refers to the table corresponding to the genre of creation and selects a question sentence and a syntax corresponding to the disclosure target.

[0086] The additional element creation unit 106a transmits the question to the terminal device 5. The question includes one or more question items, such as the content of the article to be published, the number of images and hashtags to be generated, and the style of the images and hashtags to be generated. The user inputs answers to each question item via the terminal device 5 and transmits them to the information processing device 10a.

[0087] The syntax includes one or more input fields. The additional element creation unit 106a generates an instruction sentence by inputting a response from the user into the corresponding input field of the syntax. The additional element creation unit 106a inputs the instruction sentence to the generation AI and obtains an image or a hashtag as a response from the generation AI. The additional element creation unit 106a transmits the image or hashtag to the terminal device 5. The user can insert the image or hashtag generated by the additional element creation unit 106a into an article by performing the editing operation described below.

[0088] The additional element creation unit 106a may obtain a description of the image from the generation AI instead of the image and transmit it to the terminal device 5. The user can prepare an image related to the article by taking a photo or the like in accordance with the description of the image.

[0089] The article editing unit 105a generates an editing screen for editing an article and transmits it to the terminal device 5. The article editing unit 105a then receives editing operations for the article from the terminal device 5. For example, a user can perform editing operations such as correcting text, inserting images or hashtags, and changing fonts on an article displayed on the display of the terminal device 5. The article editing unit 105a performs editing processing for the article in accordance with the user's editing operations. Then, when the article editing unit 105a receives a publishing operation from the terminal device 5, it transmits the article to the content publishing system 2 and outputs the article to the analysis unit 200a.

[0090] (Analysis Department) Next, the analysis unit 200a will be described. The analysis unit 200a has the same configuration as the analysis unit 200 of the information processing device 10 according to the first embodiment, but differs in the processing contents of the article analysis unit 203a and the correction unit 204a.

[0091] In the second embodiment, there are cases where a hash tag is inserted in the main text of an article. In this case, the labeling unit 201a does not assign a label related to the hash tag to the article, and the article analysis unit 203a analyzes the hash tag as it is.

[0092] The article analysis unit 203a analyzes the title and text of each article based on the label assignment results for each article input from the label assignment unit 201a, the hashtags included in each article, and the viewing log acquired from DB15.

[0093] (1) Title Analysis The analysis of the title is similar to the analysis of the title by the article analysis unit 203 of the information processing device 10 according to the first embodiment, and therefore a description thereof will be omitted.

[0094] (2) Analysis of the text The article analysis unit 203a analyzes the characteristics of the body text that show viewing trends based on the body text label of each article, the hashtag of each article, the number of views of each article, and the viewing time of each article. FIG. 13 shows an example of correlation analysis by the article analysis unit 203a. The analysis table 90 of FIG. 13 includes analysis data 90a and analysis results 90b. The analysis data 90a includes the label assignment results of each article, the hashtag of each article, the number of views of each article, the viewing time of each article, and the average viewing time of each article (viewing time / number of views). In FIG. 13, there are articles A to P, and the label assigned to each article is flagged with "1", and the hashtag assigned to each article is flagged with "1".

[0095] The analysis result 90b includes the "number of labels," "label x (view time / number of views)," and "average view time." The "number of labels" indicates the number of articles that include the label in that column or the number of articles that include the hashtag in that column. The "label x (view time / number of views)" indicates the total average view time (view time / number of views) of the articles. The "average view time" is the value obtained by dividing the "label x (view time / number of views)" by the "number of labels," and represents the average view time of the label or hashtag in that column.

[0096] For example, for the hashtag "#travel" in Figure 13, "6" is entered in "Number of Labels," indicating that there are six articles (articles A, D, F, I, K, and L) that have the hashtag "#travel." Furthermore, "43.65," which is the sum of the average viewing times (viewing time / number of views) of the six articles, is entered in "Labels x (viewing time / number of views)." Furthermore, "7.28," which represents the average viewing time for "#travel," is entered in "Average Viewing Time."

[0097] Then, the article analysis unit 203a analyzes the labels whose average viewing time is equal to or greater than a predetermined threshold TH3 as labels that have a high correlation with viewing (that is, characteristics of the text that tend to be viewed).

[0098] Next, the article analysis unit 203a expresses the analysis results of FIG. 13 in the form of a scatter diagram as shown in FIG. 14. FIG. 14 is a scatter diagram in which the vertical axis represents the average viewing time (s) and the horizontal axis represents the number of labels. In FIG. 14, the predetermined threshold TH3 is 7, and the article analysis unit 203a analyzes "image genre_products," "article subject genre_restaurants," "target gender_female," "target age group_50s," "number of characters_medium," "text tone_easy," and "#travel" as labels highly correlated with viewing. The article analysis unit 203a outputs the analysis results to the correction unit 204a.

[0099] The correction unit 204a modifies various data used to generate articles based on the analysis results input from the article analysis unit 203a. The various data include, for example, syntax, directives, questionnaires, questions, and editing screens.

[0100] For example, the correction unit 204a may change the questionnaire, question text, or editing screen based on the characteristics of the text that tends to be viewed. If the hashtag highly correlated with views is "#travel," the correction unit 204a may add a message such as "Articles about travel tend to be viewed" to the questionnaire or question text. The correction unit 204a may also add a message such as "Articles tagged with #travel tend to be viewed" to the editing screen.

[0101] <Third embodiment> 15 is a block diagram showing the functional configuration of an information processing apparatus according to the third embodiment. The information processing apparatus 400 includes an acquisition unit 401, a specification unit 402, and a creation unit 403.

[0102] 16 is a flowchart of processing by the information processing device of the third embodiment. The acquisition unit 401 acquires a log showing a record of viewing of a document output by a model that generates a document including content specified by a directive (step S401). The identification unit 402 identifies characteristics of the content that have a predetermined level of attention based on the log (step S402). The creation unit 403 creates the directive so as to generate the document in accordance with the identified characteristics (step S403).

[0103] According to the information processing device 400 of the third embodiment, it is possible to generate appropriate content based on the analysis result of the viewing log.

[0104] Although the present disclosure has been described above with reference to the embodiments and examples, the present disclosure is not limited to the above-described embodiments and examples. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. [Explanation of symbols]

[0105] 1. Content Analysis System 2 Content Publishing System 5, 20 Terminal equipment 10. Information processing equipment 15 Database (DB) 100, 100a generation section 101, 101a Information acquisition section 102, 102a Syntax determination unit 103, 103a Directive sentence generation unit 104, 104a Article generation section 105, 105a Article Editorial Department 106a Additional Element Creation Department 200, 200a Analysis Department 201, 201a Label assignment section 202, 202a Log recording section 203, 203a Article Analysis Section 204, 204a correction section

Claims

1. an acquisition means for acquiring a log indicating a record of viewing of a document output by a model that generates a document including the content specified by the instruction sentence; a specifying means for specifying characteristics of the content having a predetermined level of attention based on the log; a generating means for generating the instruction sentence so as to generate the document in accordance with the specified characteristics; An information processing device comprising:

2. template storage means for storing templates of instruction sentences; The information processing apparatus according to claim 1 , wherein the creating means creates the template so as to generate the document in accordance with the specified characteristics.

3. a questionnaire sending / receiving means for sending a questionnaire to a user terminal and receiving questionnaire responses from the user terminal; the instruction text includes the questionnaire response; The information processing apparatus according to claim 1 , wherein the creation means creates the questionnaire so that the document is generated in accordance with the identified characteristics.

4. The log includes the number of views and the duration of views for each document; The information processing apparatus according to claim 1 , wherein the attention level is determined based on the number of times a document is viewed or the duration of time the document is viewed.

5. The log includes viewers' reactions to the document; The information processing device according to claim 1 , wherein the attention level is determined based on the reaction.

6. 2. The information processing device according to claim 1, wherein the identification means analyzes the relationship between the attention level of the document and the characteristics of the content contained in the document by correlation analysis, and identifies the characteristics of the content having a predetermined attention level based on the analysis results.

7. The information processing device of claim 1, wherein the content characteristics include one or more of the genre of the document, the number of characters in the document, the emotion of the document, the presence or absence of numbers, the layout of the document, the target demographic of the document, the images and colors contained in the document, and the hashtags attached to the document.

8. 1. A computer-implemented information processing method, comprising: A log showing the record of the document being viewed is acquired for the document output by the model that generates a document containing the content specified in the instruction sentence, Identifying characteristics of the content that have a predetermined level of interest based on the log; An information processing method for generating the instruction text to generate the document in accordance with the specified characteristics.

9. A log showing the record of the document being viewed is acquired for the document output by the model that generates a document containing the content specified in the instruction sentence, Identifying characteristics of the content that have a predetermined level of interest based on the log; a program that causes a computer to execute a process of creating the instruction sentence so as to generate the document in accordance with the specified characteristics;

10. a relationship specifying means for specifying a relationship between the attention level of a document and a feature quantity representing a feature of content included in the document; a feature quantity specifying means for specifying a feature quantity having a desired degree of attention based on the relationship; a content creation means for creating content having characteristics represented by the identified feature amount; An information processing device comprising:

Images