Information processing device, information processing method, and information processing program

Abstract

To provide information from multiple sources in a more appropriate manner. [Solution] The information processing device according to the present invention comprises a determination unit, a calculation unit, a selection unit, and a provision unit. The determination unit determines a target category distribution, which is the target category distribution of the content provided to the user, based on information about the user. The calculation unit calculates a score for each candidate content based on the user's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution of the content selected to be provided to the user and the target category distribution determined by the determination unit. The selection unit selects content to be provided to the user from among multiple candidate content based on the score for each candidate content calculated by the calculation unit. The provision unit provides the user with information on the multiple content selected by the selection unit.

Description

Technical Field

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

Background Art

[0002] Conventionally, techniques related to information distribution via the Internet are known. As an example of such a technique, there is known a technique of arranging and displaying information of articles including article headings and thumbnails in a columnar form for each tab in a timeline format (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the prior art, there is room for further improvement in providing information of a plurality of articles more appropriately. This also applies to, for example, contents of types other than articles (such as videos, music, posted information, etc.).

[0005] The present application has been made in view of the above, and an object thereof is to provide an information processing apparatus, an information processing method, and an information processing program capable of more appropriately providing information of a plurality of contents.

Means for Solving the Problems

[0006] The information processing device according to this application comprises a determination unit, a calculation unit, a selection unit, and a provision unit. The determination unit determines a target category distribution, which is the target category distribution of the content provided to the user, based on information about the user. The calculation unit calculates a score for each candidate content based on the user's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution of the content selected to be provided to the user and the target category distribution determined by the determination unit. The selection unit selects content to be provided to the user from among multiple candidate content based on the score for each candidate content calculated by the calculation unit. The provision unit provides the user with information on the multiple content selected by the selection unit. [Effects of the Invention]

[0007] According to one embodiment, the effect is achieved that information from multiple content types can be provided more appropriately. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 shows an example of information processing according to the embodiment. [Figure 2] Figure 2 shows an example of the configuration of an information processing system according to the embodiment. [Figure 3] Figure 3 shows an example of the configuration of an information processing device according to the embodiment. [Figure 4] Figure 4 shows an example of a user information table stored in the user information storage unit of the information processing device according to the embodiment. [Figure 5] Figure 5 shows an example of an article-related information table stored in the article-related information storage unit of the information processing apparatus according to the embodiment. [Figure 6] Figure 6 is a diagram illustrating an example of a distribution penalty calculation, which is an example of a distribution score calculated by the distribution score calculation unit of the information processing device according to the embodiment. [Figure 7]Figure 7 shows an example of a plurality of articles selected by the selection unit of the information processing device according to the embodiment. [Figure 8] Figure 8 shows an example of how multiple articles are selected by the selection unit of the information processing device according to the embodiment. [Figure 9] Figure 9 is a diagram illustrating the selection of the first article by the selection unit of the information processing device according to this embodiment. [Figure 10] Figure 10 is a diagram illustrating the selection of the second article by the selection unit of the information processing device according to the embodiment. [Figure 11] Figure 11 is a diagram illustrating the selection of the third article by the selection unit of the information processing device according to the embodiment. [Figure 12] Figure 12 is a diagram illustrating the categories of multiple articles selected by the selection unit of the information processing device according to the embodiment. [Figure 13] Figure 13 shows an example of an article screen that includes a list of article information provided by the information processing unit according to the embodiment and displayed on a terminal device. [Figure 14] Figure 14 is a flowchart showing an example of information processing by the processing unit of the information processing device according to the embodiment. [Figure 15] Figure 15 is a flowchart showing an example of the process by which the processing unit of the information processing device according to the embodiment provides a list of article information. [Figure 16] Figure 16 is a hardware configuration diagram showing an example of a computer that implements the respective functions of the information processing device and terminal device according to the embodiment. [Modes for carrying out the invention]

[0009] Hereinafter, embodiments for implementing an information processing apparatus, an information processing method, and an information processing program according to the present application (hereinafter referred to as "embodiments") will be described in detail with reference to the drawings. Note that the information processing apparatus, the information processing method, and the information processing program according to the present application are not limited by this embodiment. Also, each embodiment can be appropriately combined within a range that does not conflict with the processing content. In addition, in the following embodiments, the same parts are denoted by the same reference numerals, and duplicate explanations are omitted.

[0010] 〔1. An example of information processing〕 First, an example of information processing according to the embodiment will be described using FIG. 1. FIG. 1 is a diagram showing an example of information processing according to the embodiment.

[0011] The information processing apparatus 1 provides a content service for distributing content. The content is, for example, an article such as a news article, a video, music, or posting information, etc., but is not limited to such examples. The posting information is, for example, posting information for an article (e.g., a posting comment, etc.), a posting text or a posting image on an SNS (Social Networking Service), etc., but is not limited to such examples.

[0012] Hereinafter, it will be described assuming that the content provided by the information processing apparatus 1 is an article such as a news article. However, the content provided by the information processing apparatus 1 may be content of a type other than an article (e.g., a video, music, posting information, etc.), or may be a plurality of types of content.

[0013] As shown in FIG. 1, the information processing apparatus 1 receives submissions of a plurality of article-related information from a plurality of submitter terminals 3 (step S1). The plurality of submitter terminals 3 are terminals of different submitters O. The submitter O is, for example, an employee of a news organization, a journalist, etc., but is not limited to such examples.

[0014] Article-related information is, for example, information regarding an article (an example of content) submitted by submitter O. Article-related information includes, for example, an article (for example, an article body including text and images), a title of the article, a thumbnail image of the article, a category of the article, a text snippet, a submission date and time (or creation date and time), keywords, a submitter name, a link to related articles, and the like, but is not limited to such examples. The text snippet is a snippet of the text of a news article.

[0015] Further, the information processing apparatus 1 determines a target category distribution that is a target of the category distribution of articles provided to user U based on information regarding user U of the article information distribution service (step S2). Information regarding user U includes, for example, at least one of attribute information and behavior history of user U. The target category distribution is information indicating a target of the provision ratio of articles for each category to user U, and includes a combination of a category and a target of the provision ratio for each category.

[0016] For example, assume that the categories of articles are five categories: sports, entertainment, business, IT, and politics. In this case, the target category distribution is indicated by a combination of a category and a target of the provision ratio, for example, sports 40%, entertainment 30%, business 15%, IT 10%, politics 5%, and the like.

[0017] The information processing apparatus 1 determines the target category distribution based on at least one of attribute information and behavior history of user U. For example, the information processing apparatus 1 can determine, as the target category distribution, a category distribution corresponding to the attributes of user U based on the attribute information of user U. The information processing apparatus 1 determines the target category distribution corresponding to the attributes of user U based on information in which the target category distribution is set for each combination of the contents of the attribute items (for example, a 30-year-old male in Tokyo, a 20-year-old female in Osaka Prefecture, etc.).

[0018] Furthermore, the information processing device 1 can estimate user U's level of interest for each category based on user U's behavioral history, and determine a target category distribution where the estimated level of interest for each category is used as the provision rate for each category. For example, based on user U's behavioral history, the information processing device 1 can estimate the proportion of each category of content (e.g., articles, videos, music, posted information, keywords, etc.) that user U has viewed, selected, or searched for, as user U's level of interest for each category.

[0019] Furthermore, the information processing device 1 can determine the target category distribution using a model that takes information including at least one of the user U's attribute information and behavioral history as input and outputs information indicating the provision rate for each category. Such a model is, but is not limited to, a regression model, a support vector machine, a gradient boosting decision tree, or a convolutional neural network. The model may also be a generative AI (Artificial Intelligence) such as a Large Language Model (LLM).

[0020] Next, the information processing device 1 calculates a score (hereinafter sometimes referred to as a ranking score) for each candidate article (an example of candidate content) based on the interest score and the distribution score (step S3). For example, the information processing device 1 calculates an interest score for each candidate article based on information about user U and information about the candidate article. Information about user U includes, for example, at least one of the following: user U's attribute information and behavioral history.

[0021] For example, the information processing device 1 can calculate the interest score of candidate articles using a model that takes as input information including at least one piece of information from user U's attribute information and behavioral history, and information about candidate articles, and outputs information indicating the interest score of the candidate articles. Such a model may be, but is not limited to, a regression model, a support vector machine, a gradient boosting decision tree, or a convolutional neural network. The model may also be a generative AI such as a large-scale language model.

[0022] Furthermore, the information processing device 1 calculates a distribution score. The distribution score is a score corresponding to the distribution difference, which is the difference between the category distribution of the articles selected in step S4 as to be provided to user U and the target category distribution determined in step S2. The distribution score is calculated using a formula that approaches zero as the distribution difference is small, but it is not limited to this example, and may also be calculated using a formula that approaches zero as the distribution difference is large.

[0023] For example, if the distribution score is calculated using a formula that approaches zero as the distribution difference decreases, the information processing device 1 calculates the ranking score of the candidate articles by subtracting the distribution score, weighted accordingly, from the value obtained by dividing the interest score by the number of selections up to the next article in step S4. The number of selections up to the next article in step S4 can also be said to be the order in which the candidate articles are selected, and is an example of the display order described later.

[0024] Furthermore, if the information processing device 1 calculates the distribution score using a formula that approaches zero as the distribution difference increases, for example, it can calculate the ranking score of candidate articles by adding the distribution scores, each weighted, to the value obtained by dividing the interest score by the number of selections up to the next article in step S4.

[0025] Next, the information processing device 1 selects an article to be provided to user U from among multiple candidate articles based on the ranking score for each candidate article calculated in step S3 (step S4). For example, the information processing device 1 selects the candidate article with the highest ranking score from among the multiple candidate articles for which ranking scores were calculated in step S3 as the article to be provided.

[0026] The information processing device 1 selects a predetermined number of articles to be provided by repeating the processes of step S3 and step S4 a predetermined number of times (for example, 10 or 20) which is the number of articles to be provided. Specifically, the information processing device 1 calculates the ranking scores of multiple candidate articles excluding the candidate article selected as an article to be provided in step S4, and selects a predetermined number of articles to be provided by repeating the process of selecting the candidate article with the highest ranking score among the calculated ranking scores a predetermined number of times.

[0027] Furthermore, the method for calculating the ranking score is not limited to the example described above. For example, the information processing device 1 can also calculate the ranking score of candidate articles by subtracting or adding weighted distribution scores to the interest score, without using the number of selections up to the next article in step S4.

[0028] Next, the information processing device 1 provides user U with a group of article content containing information on multiple target articles selected in step S4 (step S5). The information on target articles provided to user U includes, for example, thumbnail images and headlines of the target articles, and the information on multiple target articles is displayed in a list on user U's terminal device 2.

[0029] Here, let's assume that the interest scores for the five sports candidate articles are 100, 80, 70, 50, and 40, and the interest scores for the five entertainment candidate articles are 30, 20, 15, 10, and 5. In this case, if we select 5 articles to offer from 10 candidate articles using only user U's interest scores for the candidate articles, the five sports candidate articles will be selected as the 5 articles to offer, as shown in article group A in Figure 1.

[0030] On the other hand, the information processing device 1 uses a distribution score that corresponds to the target category distribution (50% sports and 50% entertainment), in addition to the user U's interest score for candidate articles. Therefore, when the information processing device 1 selects 5 articles to provide from 10 candidate articles, as shown in article group B in Figure 1, 3 sports candidate articles (interest scores of 100, 80, and 70) and 2 entertainment candidate articles (interest scores of 30 and 20) will be selected as the 5 articles to provide.

[0031] In the example shown in Figure 1, the articles are selected in the following order: sports articles with an interest score of 100, entertainment articles with an interest score of 30, sports articles with an interest score of 80, entertainment articles with an interest score of 20, and sports articles with an interest score of 70. However, the example is not limited to this. For example, depending on the weight values ​​used to calculate the ranking score, a sports article with an interest score of 70 may be selected before an entertainment article with an interest score of 20.

[0032] In this way, the information processing device 1 can suppress bias in the categories of articles provided to user U and can provide information on multiple articles more appropriately. Below, the configuration of the information processing system, including the information processing device 1, multiple terminal devices 2, and multiple contributor terminals 3, which perform such processing, will be described in detail.

[0033] [2. Configuration of the Information Processing System] Figure 2 shows an example of the configuration of an information processing system according to the embodiment. As shown in Figure 2, the information processing system 100 according to the embodiment includes an information processing device 1, a plurality of terminal devices 2, and a plurality of submitter terminals 3.

[0034] Multiple terminal devices 2 are used by different users U. Multiple submitter terminals 3 are used by different submitters O. Each terminal device 2 and each submitter terminal 3 is, for example, a notebook PC (Personal Computer), a desktop PC, a smartphone, a tablet PC, or a wearable device. Wearable devices are, for example, smart glasses or smartwatches, but are not limited to these examples.

[0035] Each of the information processing device 1, terminal device 2, and submitter terminal 3 is connected to each other via network N, either by wire or wireless, enabling communication. Note that the information processing system 100 shown in Figure 2 may include multiple information processing devices 1, etc.

[0036] Network N includes, for example, WANs (Wide Area Networks) such as the Internet, and mobile communication networks such as LTE (Long Term Evolution), 4G (4th Generation), and 5G (5th Generation: 5th Generation Mobile Communication System).

[0037] Each terminal device 2 and each submitter terminal 3 can connect to the network N via short-range wireless communication such as a mobile communication network, Bluetooth®, or Wi-Fi (Local Area Network), and communicate with the information processing device 1.

[0038] [3. Configuration of Information Processing Device 1] Figure 3 shows an example of the configuration of an information processing device 1 according to an embodiment. As shown in Figure 3, the information processing device 1 includes a communication unit 10, a storage unit 11, and a processing unit 12.

[0039] [3.1. Communications Section 10] The communication unit 10 is implemented, for example, by a communication module or a NIC (Network Interface Card). The communication unit 10 is connected to the network N by wire or wireless connection and transmits and receives information with various other devices. For example, the communication unit 10 transmits and receives information with terminal device 2 and submitter terminal 3 via the network N.

[0040] [3.2. Storage section 11] The memory unit 11 is implemented by, for example, semiconductor memory elements such as RAM (Random Access Memory) and flash memory, or by storage devices such as hard disks and optical discs. The memory unit 11 includes a user information storage unit 20 and an article-related information storage unit 21.

[0041] [3.2.1. User information storage unit 20] The user information storage unit 20 stores various information about user U. Figure 4 is a diagram showing an example of a user information table stored in the user information storage unit 20 of the information processing device 1 according to this embodiment.

[0042] In the example shown in Figure 4, the user information table stored in the user information storage unit 20 includes information on items such as "User ID (Identifier)," "Attribute Information," and "Behavioral History." The "User ID" is an identifier that identifies user U, and is assigned to each user U.

[0043] "Attribute information" refers to attribute information that indicates the attributes of user U associated with the "User ID". User U's attributes include, for example, demographic attributes and psychographic attributes. Demographic attributes are demographic attributes and include multiple attribute items such as age, gender, occupation, place of residence, annual income, and family structure.

[0044] Psychographic attributes are psychological attributes and include multiple attribute items related to, for example, lifestyle, values, and interests. For example, each of the multiple attribute items in a psychographic attribute is an object of interest to user U, such as sports, entertainment, business, IT, politics, economics, international affairs, life, incidents / accidents, travel, fashion, science, books, art, technology, music, movies, etc. Attribute items that user U is interested in are assigned a value of 1, and those that are not are assigned a value of 0, but this is not an exhaustive example. For example, each of the multiple attribute items in a psychographic attribute may be indicated by user U's level of interest.

[0045] "Activity history" includes the activity history of user U associated with the "User ID". User U's activity history includes, for example, search history, browsing history, and transaction history. User U's search history includes, for example, information on user U's search history on web search services. User U's browsing history includes, for example, user U's content browsing history on online services. Transaction history includes, for example, information on user U's transaction history of goods on online services.

[0046] The content viewing history of user U in the online service includes the viewing history of articles provided to user U by the article information distribution service described above. The article viewing history includes the history of selecting (e.g., clicking or tapping) information on articles displayed in a list on user U's terminal device 2.

[0047] [3.2.2. Article-related information storage section 21] The article-related information storage unit 21 stores various types of information related to articles. Figure 5 shows an example of an article-related information table stored in the article-related information storage unit 21 of the information processing device 1 according to this embodiment.

[0048] In the example shown in Figure 5, the article-related information table stored in the article-related information storage unit 21 includes information on items such as "article ID," "category," "article," "headline," "thumbnail image," and "submission date and time." The "article ID" is an identifier that identifies the article and is assigned to each article.

[0049] The "Category" is information indicating the category of the article corresponding to the "Article ID". Examples of article categories include, but are not limited to, broad categories such as sports, entertainment, business, IT, politics, economics, international, lifestyle, and incidents / accidents.

[0050] In the example shown in Figure 5, only the major categories are shown, but subcategories and minor categories are also included. Subcategories, for example, if the higher-level major category is "sports," would include baseball, soccer, rugby, basketball, volleyball, etc. Minor categories, for example, if the higher-level subcategory is "soccer," would include national leagues such as the J.League, Premier League, Serie A, and Bundesliga, but could also include domestic soccer categories such as J1, J2, J3, and high school soccer.

[0051] An "article" is the article corresponding to the "article ID," and includes information such as the article's body, images, and layout. The article's body consists of strings, and the article's images consist of the article's image information. Note that an "article" may also be the address of the article corresponding to the "article ID" (for example, a URL (Uniform Resource Locator)).

[0052] "Headline" is the text information of the headline corresponding to, for example, "Article ID". "Thumbnail image" is the image information of the thumbnail image of the article corresponding to, for example, "Article ID". "Submission date and time" is information indicating the date and time when the article corresponding to, for example, "Article ID" was submitted.

[0053] Although not shown in Figure 5, the article-related information table includes information such as the article's text snippet, keywords, author's name, and links to related articles, corresponding to each "article ID".

[0054] [3.3. Processing Unit 12] The processing unit 12 is a controller, and is realized, for example, by a CPU (Central Processing Unit) or MPU (Micro Processing Unit) executing various programs stored in the memory device inside the information processing device 1 using RAM as the working area.

[0055] The processing unit 12 may be partially or entirely implemented by an integrated circuit, such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0056] As shown in Figure 3, the processing unit 12 includes an acquisition unit 30, a determination unit 31, a reception unit 32, a calculation unit 33, an adjustment unit 34, a selection unit 35, and a provision unit 36, and realizes or executes the information processing functions and operations described below. Note that the internal configuration of the processing unit 12 is not limited to the configuration shown in Figure 3, and other configurations are also acceptable as long as they perform the information processing described later.

[0057] [3.3.1. Acquisition part 30] The acquisition unit 30 acquires various information from an external information processing device, terminal device 2, or submitter terminal 3 via the communication unit 10, and stores the acquired information in the storage unit 11.

[0058] For example, the acquisition unit 30 can acquire user information transmitted from the terminal device 2 and store the acquired user information in the user information storage unit 20. The acquisition unit 30 can also acquire article-related information transmitted from the submitter terminal 3 and store the acquired article-related information in the article-related information storage unit 21.

[0059] Furthermore, the acquisition unit 30 acquires various types of information from the storage unit 11. For example, the acquisition unit 30 acquires information included in user information from the user information storage unit 20, etc. The acquisition unit 30 also acquires information included in article-related information from the article-related information storage unit 21, etc.

[0060] [3.3.2. Decision Section 31] The decision unit 31 makes various decisions. For example, based on information about user U, the decision unit 31 determines the target category distribution, which is the target category distribution of articles provided to user U. Information about user U includes, for example, at least one of user U's attribute information and behavioral history. The target category distribution is information indicating the target percentage of articles provided to user U for each category, and includes a combination of category and target percentage for each category.

[0061] For example, suppose the article categories are sports, entertainment, business, IT, and politics. In this case, the target category distribution would be shown as a combination of categories and target distribution percentages, such as 40% sports, 30% entertainment, 15% business, 10% IT, and 5% politics.

[0062] The decision unit 31 determines the target category distribution based on, for example, at least one piece of information from user U's attribute information and behavioral history. For example, the decision unit 31 can determine a category distribution corresponding to user U's attributes as the target category distribution based on user U's attribute information.

[0063] The decision unit 31 determines the target category distribution according to the user U's attributes based on information in which the target category distribution is set for each combination of attribute item contents (for example, a male in his 30s from Tokyo, a female in her 20s from Osaka, etc.). Attribute items include, for example, demographic attribute items and psychographic attribute items.

[0064] Demographic attributes include items such as age, gender, occupation, place of residence, annual income, and family structure. Psychographic attributes include items that indicate the user's interests, such as sports, entertainment, business, IT, politics, economics, international affairs, lifestyle, incidents / accidents, travel, fashion, science, books, art, technology, music, and movies, but are not limited to these examples. For example, demographic attributes may include items that further classify each of the above items.

[0065] Furthermore, the decision unit 31 can estimate user U's level of interest for each category based on user U's behavioral history, and determine a target category distribution where the estimated level of interest for each category is used as the provision rate for each category.

[0066] The decision unit 31 can, for example, estimate the proportion of content that user U has viewed, selected, searched, etc., for each category, based on user U's behavioral history, as user U's level of interest in each category.

[0067] User U's behavioral history includes, for example, the history of articles provided to User U, and the decision unit 31 can estimate, for example, the proportion of categories of articles provided to User U (provision rate) as User U's level of interest for each category. The articles provided to User U are, for example, articles selected by User U from the article list, but are not limited to this example. The history of articles provided to User U may also be the selection history of article information included in the article information list (for example, tap history or click history).

[0068] Furthermore, if the content of the psychographic attribute items stored in the user information storage unit 20 represents the user U's level of interest for each category, the determination unit 31 can also use this information to determine the target category distribution.

[0069] Furthermore, the decision unit 31 can also determine the category distribution of articles provided to user U by including information that includes contextual information of user U as information about user U, in addition to or instead of at least one of user U's attribute information and behavioral history.

[0070] User U's contextual information is information about User U's context, and includes, but is not limited to, one or more of the following: User U's current location, current time, the physical environment in which User U is located, User U's physical state, and User U's emotions.

[0071] The physical environment in which user U is placed includes, but is not limited to, temperature, humidity, weather, illuminance, indoors, outdoors, or a combination of two or more of these. User U's physical state includes, but is not limited to, running, walking, sitting, etc. User U's emotions include, but are not limited to, laughing, being angry, sad, distressed, or a normal state.

[0072] The decision unit 31 can, for example, determine the target category distribution according to the user U's context based on information in which the target category distribution is set for each context.

[0073] Furthermore, the decision unit 31 can determine the target category distribution using a model that takes information including at least one of the user U's attribute information, behavioral history, and contextual information as input and outputs information indicating the provision rate for each category. Such a model is, but is not limited to, a regression model, a support vector machine, a gradient boosting decision tree, or a convolutional neural network.

[0074] Furthermore, the model used in the decision unit 31 may be a generative AI such as a large-scale language model. In this case, the decision unit 31 can determine the target category distribution by inputting to the model information that includes, for example, information containing at least one of the user U's attribute information, behavioral history, and context, and instruction information that instructs the model to determine the target category distribution from at least one of the user U's attribute information, behavioral history, and context.

[0075] The user U's behavioral history entered into the model may include, but is not limited to, the percentage of content (e.g., article information, articles, keywords, etc.) that user U viewed, selected, or searched for in each category.

[0076] Furthermore, the determination unit 31 can determine the category distribution of articles provided to user U for each predetermined number of articles (e.g., 10 or 20). For example, the determination unit 31 can determine the category distribution of articles provided to the same user U each time a predetermined number of articles are selected by the selection unit 35. The predetermined number is, for example, a number specified by user U, but may also be, for example, a number corresponding to user U's attributes. The predetermined number is, for example, the number of article information items included in the article information list provided to user U in response to an article list distribution request, but is not limited to this example.

[0077] Furthermore, the decision unit 31 determines a target category distribution that includes, for example, a target for the percentage of major categories provided, a target for the percentage of medium categories provided, or a target for the percentage of minor categories provided. It can also determine a target category distribution that includes two or more pieces of information from among the target for the percentage of major categories provided, the target for the percentage of medium categories provided, and the target for the percentage of minor categories provided.

[0078] [3.3.3. Reception Department 32] The reception unit 32 receives various requests and information from terminal devices 2 and submitter terminals 3 via the communication unit 10. For example, the reception unit 32 receives requests for article list distribution and article distribution from terminal devices 2. Article list distribution requests may include information specifying major categories and minor categories, for example. Article distribution requests may include information specifying articles.

[0079] Furthermore, the reception unit 32 receives submission requests transmitted from the submitter terminal 3. Submission requests include information about the article to be submitted. When the reception unit 32 receives a submission request, it stores the article-related information, which is information about the article included in the submission request, in the article-related information storage unit 21.

[0080] Article-related information included in the submission request may include, but is not limited to, the article itself (e.g., the article body including text and images), the article's headline, the article's thumbnail image, the article's category, the article's snippet, the submission date (or creation date), keywords, the submitter's name, and links to related articles.

[0081] Furthermore, the reception unit 32 receives, for example, the number of articles included in the article information list provided to user U in response to an article list distribution request (for example, the predetermined number mentioned above). For example, the reception unit 32 can receive a specification from user U regarding the number of articles included in the article information list. By operating the terminal device 2, user U can cause the terminal device 2 to execute a process that sends a specification request containing information specifying the number of articles included in the article information list to the information processing device 1. By receiving the specification request, the reception unit 32 can receive a specification from user U regarding the number of articles included in the article information list.

[0082] [3.3.4. Calculation Section 33] The calculation unit 33 calculates a ranking score for each candidate article based on the interest score of user U for the candidate article and a distribution score corresponding to the difference between the category distribution of the articles selected to be provided to user U and the target category distribution determined by the decision unit 31.

[0083] The calculation unit 33 calculates a ranking score for each candidate article corresponding to the target user, for example, when an article list distribution request is received by the reception unit 32, but is not limited to this example. If the article list distribution request received by the reception unit 32 does not include a category specification, the calculation unit 33 calculates a ranking score for each candidate article without narrowing down the category.

[0084] If the article list distribution request received by the reception unit 32 includes the specification of a major category, the calculation unit 33 calculates a ranking score for each candidate article belonging to the specified major category. For example, if the article list distribution request received by the reception unit 32 includes the specification of the major category "sports," the calculation unit 33 calculates a ranking score for each candidate article belonging to the "sports" category.

[0085] If the article list distribution request received by the reception unit 32 includes the specification of a subcategory, the calculation unit 33 calculates a ranking score for each candidate article belonging to the specified subcategory. For example, if the article list distribution request received by the reception unit 32 includes the specification of "baseball" as a subcategory, the calculation unit 33 calculates a ranking score for each candidate article belonging to the "baseball" category.

[0086] For example, the calculation unit 33 calculates a ranking score for each candidate article using the following formula (1). In the following formula (1), "display rank" is the order in which the article is presented to user U, and "distribution penalty" is an example of a distribution score, which is a distribution score calculated using a formula where the score approaches zero as the distribution difference described above becomes smaller. Also, in the following formula (1), "λ" is the weight for the value obtained by dividing the interest score by the display rank, and "1-λ" is the weight for the distribution penalty, which is an example of a distribution score. "λ" is a value defined, for example, 0≦λ≦1, but is not limited to such examples.

[0087]

number

[0088] The “distribution penalty” shown in equation (1) below is, for example, the KL (Kullback-Leibler) divergence, which can be expressed, for example, in equation (2) below. In equation (2) below, “P” is, for example, the target category distribution, and “Q” is, for example, the category distribution of articles selected to be provided to user U.

[0089]

number

[0090] The calculation unit 33 calculates the ranking score for each candidate article, excluding the candidate article selected by the selection unit 35 as an article to be provided to user U. For example, as described above, the calculation unit 33 can calculate the ranking score for each candidate article by dividing the user U's interest score for the candidate article by the article's ranking in the selection unit 35 (an example of the number of selections until the next article by the selection unit 35), and then subtracting a distribution penalty, which is an example of a distribution score, with a weight applied to each value.

[0091] If the article list distribution request received by the reception unit 32 includes the specification of a major category, the calculation unit 33 treats the distribution of multiple subcategories classified under the specified major category as the target category distribution used to calculate the distribution score. For example, if the article list distribution request received by the reception unit 32 includes the specification of the major category "sports," the calculation unit 33 treats the distribution of categories such as baseball, soccer, tennis, and basketball, which are classified under the sports category, as the target category distribution used to calculate the distribution score.

[0092] Furthermore, if the article list distribution request received by the reception unit 32 includes the specification of a subcategory, the calculation unit 33 treats the distribution of categories of multiple sub-items classified under the specified subcategory as the target category distribution used to calculate the distribution score. For example, if the article list distribution request received by the reception unit 32 includes the specification of baseball as a subcategory, the calculation unit 33 treats the distribution of categories such as Japanese professional baseball, MLB, independent baseball leagues, and high school baseball, which are classified under the baseball category, as the target category distribution used to calculate the distribution score.

[0093] The calculation unit 33 calculates the ranking score for each of the multiple candidate articles, excluding the candidate article selected by the selection unit 35, until the number of articles selected by the selection unit 35 up to the next article reaches a predetermined number. In other words, for each ranking position, the calculation unit 33 calculates the ranking score for each of the multiple candidate articles, excluding the candidate article selected by the selection unit 35.

[0094] The calculation unit 33 includes an interest score calculation unit 40 for calculating an interest score, a distribution score calculation unit 41 for calculating a distribution score, and a ranking score calculation unit 42 for calculating a ranking score.

[0095] [3.3.4.1. Interest Score Calculation Unit 40] The interest score calculation unit 40 calculates an interest score for each candidate article based, for example, on information about user U (for example, some or all of the user information described above) and information about the candidate articles. For example, the interest score calculation unit 40 can calculate the interest score for a candidate article using a model that takes as input information including at least one piece of information from user U's attribute information and behavioral history, and information about the candidate articles, and outputs information indicating the interest score of the candidate articles.

[0096] The model used in the interest score calculation unit 40 is, for example, a regression model, a support vector machine, a gradient boosting decision tree, or a convolutional neural network, but is not limited to these examples. Alternatively, the model used in the interest score calculation unit 40 may be a generative AI such as a large-scale language model. In this case, the interest score calculation unit 40 can calculate an interest score for a candidate article by inputting to the model input information that includes information about user U and information about candidate articles, and instruction information that instructs the model to determine user U's level of interest in the candidate articles based on the information about user U and the candidate articles.

[0097] [3.3.4.2. Distribution score calculation unit 41] The distribution score calculation unit 41 calculates a distribution score based, for example, on the target category distribution determined by the determination unit 31 and the category distribution (status) of the articles selected by the selection unit 35. The distribution score is a score corresponding to the distribution difference, which is the difference between the target category distribution determined by the determination unit 31 and the category distribution status of the articles selected by the selection unit 35, and is calculated, for example, using the above formula (2).

[0098] If Q is zero in any of the categories, then in equation (2) above, P / Q becomes infinite in the category where Q is zero, causing the KL divergence to diverge. Therefore, when calculating equation (2) above, the distribution score calculation unit 41 adds a positive constant (for example, "0.01") to Q in each category to avoid the KL divergence diverging.

[0099] Figure 6 is a diagram illustrating an example of a distribution penalty calculation, which is an example of a distribution score calculated by the distribution score calculation unit 41 of the information processing device 1 according to the embodiment. In the example shown in Figure 6, an example is shown where the target category distribution is 50% sports and 50% entertainment, and "0.01" is used as the positive constant.

[0100] In Example A shown in Figure 6, the category distribution of articles selected by the selection unit 35 is 60% sports and 40% entertainment. In this case, for the sports category, P / Q = 0.51 / 0.61, and for the entertainment category, P / Q = 0.51 / 0.41. Therefore, the distribution score calculation unit 41 calculates a distribution score of 0.0196, as shown in Figure 6.

[0101] In Example B shown in Figure 6, the category distribution of articles selected by the selection unit 35 is 20% sports and 80% entertainment. In this case, for the sports category, P / Q = 0.51 / 0.21, and for the entertainment category, P / Q = 0.51 / 0.81. Therefore, the distribution score calculation unit 41 calculates a distribution score of 0.02123, as shown in Figure 6.

[0102] In Example C shown in Figure 6, the category distribution of the articles selected by the selection unit 35 is 50% sports and 50% entertainment. In this case, since P / Q = 0.51 / 0.51 for the sports category and P / Q = 0.51 / 0.51 for the entertainment category, the distribution score calculation unit 41 calculates 0 as the distribution score, as shown in Figure 6.

[0103] The distribution score calculated by the distribution score calculation unit 41 is not limited to the KL divergence, but may be, for example, the Wasserstein distance, the Helllinger distance, or other index values.

[0104] In the example described above, the distribution score is calculated using a formula that approaches zero as the distribution difference decreases. However, the method is not limited to this example, and the score may be calculated using a formula that approaches zero as the distribution difference decreases. For example, the distribution score calculated by the distribution score calculation unit 41 may be the angle between the target category distribution determined by the determination unit 31 and the category distribution of the articles selected by the selection unit 35, treating them as vectors.

[0105] [3.3.4.3. Ranking Score Calculation Section 42] The ranking score calculation unit 42 calculates a ranking score for each candidate article based on the interest score calculated by the interest score calculation unit 40 and the distribution score calculated by the distribution score calculation unit 41.

[0106] The ranking score calculation unit 42 can calculate the ranking score for each candidate article using the above formula (1) if, for example, a distribution penalty, which is an example of a distribution score, is calculated using a formula that approaches zero as the distribution difference becomes smaller. More specifically, the ranking score calculation unit 42 can calculate the ranking score for each candidate article by dividing the interest score calculated by the interest score calculation unit 40 by the article's ranking in the selection unit 35 (an example of the number of selections up to the next article by the selection unit 35), and then subtracting the distribution penalty, which is an example of a distribution score calculated by the distribution score calculation unit 41, with a weight assigned to each value.

[0107] Furthermore, if the ranking score calculation unit 42 calculates the distribution score using a formula that approaches zero as the distribution difference increases, for example, it can calculate the ranking score for each candidate article by adding the distribution score calculated by the distribution score calculation unit 41 to the value obtained by dividing the interest score calculated by the interest score calculation unit 40 by the ranking of the article in the selection unit 35 (an example of the number of selections up to the next article by the selection unit 35), with each value weighted accordingly.

[0108] The calculation of the ranking score by the ranking score calculation unit 42 is not limited to the example described above. The ranking score may be calculated, for example, based on the interest score calculated by the interest score calculation unit 40 and the distribution score calculated by the distribution score calculation unit 41, and the calculation method may not use the ranking of articles displayed in the selection unit 35 (an example of the number of selections until the next article).

[0109] For example, if the ranking score calculation unit 42 calculates a distribution penalty, which is an example of a distribution score, using a calculation formula that approaches zero as the distribution difference becomes smaller, it may be configured to calculate the ranking score of each candidate article by subtracting the distribution score calculated by the distribution score calculation unit 41 from the interest score calculated by the interest score calculation unit 40, each with a weighted subtraction.

[0110] Furthermore, if the ranking score calculation unit 42 is configured to calculate the ranking score for each candidate article by weighting and adding the interest score calculated by the interest score calculation unit 40 and the distribution score calculated by the distribution score calculation unit 41, for example, when the distribution score is calculated using a formula that approaches zero as the distribution difference increases.

[0111] Furthermore, if the content of the psychographic attribute items stored in the user information storage unit 20 represents the user U's level of interest for each category, the ranking score calculation unit 42 can also use such level of interest as the interest score to determine the ranking score of the candidate articles.

[0112] [3.3.5. Adjustment section 34] The adjustment unit 34 adjusts the weights used in the ranking score calculation unit 42 based on information about user U. For example, the adjustment unit 34 can adjust the "λ" that determines the weights "λ" and "1-λ" in the above equation (1) based on information about user U.

[0113] Information about user U may include, but is not limited to, one or more of the following: user U's attribute information, user U's behavioral history, or user U's contextual information.

[0114] For example, the adjustment unit 34 can set a value corresponding to user U's attributes as "λ" based on user U's attribute information. The adjustment unit 34 sets the value corresponding to user U's attributes in formula (1) above, based on information in which "λ" has been set for each combination of attribute item contents (for example, a male in his 30s from Tokyo, a female in her 20s from Osaka). Attribute items include, for example, demographic attribute items and psychographic attribute items.

[0115] Furthermore, the adjustment unit 34 can set "λ" in formula (1) above according to the user U's behavior history, based on the user U's behavior history. For example, the adjustment unit 34 can set "λ" in formula (1) above according to the user U's behavior pattern, based on information in which "λ" is set for each behavior pattern (for example, a pattern indicated by a range such as search rate, viewing rate, and selection rate for content for each category).

[0116] Furthermore, the adjustment unit 34 can set "λ" as information relating to user U, which includes contextual information of user U, in addition to or instead of at least one of user U's attribute information and behavioral history.

[0117] For example, the adjustment unit 34 can determine a value of "λ" appropriate to the user U's context based on information in which "λ" is set for each context.

[0118] Furthermore, the adjustment unit 34 can determine "λ" using a model that inputs information including at least one of the user U's attribute information, behavioral history, and contextual information, and outputs information indicating "λ". Such a model may be, but is not limited to, a regression model, a support vector machine, a gradient boosting decision tree, or a convolutional neural network.

[0119] Furthermore, the model used in the adjustment unit 34 may be, for example, a generative AI such as a large-scale language model. In this case, the adjustment unit 34 can determine "λ" by inputting input information into the model, for example, information including at least one of the user U's attribute information, behavioral history, and context, and instruction information that instructs the model to determine "λ" from at least one of the user U's attribute information, behavioral history, and context.

[0120] The user U's behavioral history entered into the model may include, but is not limited to, the percentage of content (e.g., article information, articles, keywords, etc.) that user U viewed, selected, or searched for in each category.

[0121] [3.3.6. Selection Section 35] The selection unit 35 selects from among multiple candidate articles the articles to be provided to user U (for example, the target user mentioned above) based on the ranking score for each candidate article calculated by the calculation unit 33.

[0122] The selection unit 35, for example, each time the calculation unit 33 calculates a score for each candidate article, selects the candidate article with the highest score as the article to be provided to user U. For example, the selection unit 35 selects the candidate article with the highest ranking score from among multiple candidate articles for which the calculation unit 33 has calculated a ranking score, as the article to be provided. Alternatively, the selection unit 35 can also select two or more candidate articles for which the calculation unit 33 has calculated a ranking score, in descending order of their ranking scores, as the articles to be provided.

[0123] Figure 7 shows an example of multiple articles selected by the selection unit 35 of the information processing device 1 according to the embodiment. In the example shown in Figure 7, an example is shown where the target category distribution determined by the decision unit 31 is 50% sports and 50% entertainment, and it also shows an example of multiple articles selected by the selection unit 35 when λ=0, λ=0.5, and λ=0.99.

[0124] As shown in Figure 7, it can be seen that the category distribution of the multiple articles selected by the selection unit 35 approaches the target category distribution determined by the determination unit 31 as λ increases.

[0125] For example, each time the ranking score of multiple candidate articles, excluding the candidate article selected by the selection unit 35, is calculated by the calculation unit 33, the selection unit 35 selects an article from among the unselected candidate articles that is to be provided to user U.

[0126] Figure 8 shows an example of how multiple articles are selected by the selection unit 35 of the information processing device 1 according to the embodiment. In the example shown in Figure 8, the candidate articles selected by the selection unit 35 are five sports candidate articles with interest scores of 100, 80, 70, 50, and 40 for user U, and five entertainment candidate articles with interest scores of 30, 20, 15, 10, and 5 for user U.

[0127] In the example shown in Figure 8, the selection unit 35 selects a sports candidate article with an interest score of 100 as the first article, an entertainment candidate article with an interest score of 30 as the second article, a sports candidate article with an interest score of 80 as the third article, and an entertainment candidate article with an interest score of 20 as the fourth article.

[0128] The selection process by the selection unit 35 shown in Figure 8 will be explained in more detail. Figure 9 is a diagram illustrating the selection of the first article by the selection unit 35 of the information processing device 1 according to the embodiment. Figure 10 is a diagram illustrating the selection of the second article by the selection unit 35 of the information processing device 1 according to the embodiment. Figure 11 is a diagram illustrating the selection of the third article by the selection unit 35 of the information processing device 1 according to the embodiment. Figures 9 to 11 show an example where the target category distribution is 50% sports and 50% entertainment, with λ set to 0.99 and "0.01" used as a positive constant.

[0129] As shown in Figure 9, in the selection of the first article, since the ranking is "1", for a sports candidate article with an interest score of 100, interest score / ranking = 100 / 1, and for an entertainment candidate article with an interest score of 30, interest score / ranking = 30 / 1. Furthermore, the distribution score calculated by the distribution score calculation unit 41 of the calculation unit 33 is 1.6242 for both the sports candidate article with an interest score of 100 and the entertainment candidate article with an interest score of 30.

[0130] Therefore, the ranking score calculated by the ranking score calculation unit 42 of the calculation unit 33 is -0.608 for a sports candidate article with an interest score of 100, and -1.308 for an entertainment candidate article with an interest score of 30. In this case, the selection unit 35 selects the sports candidate article with an interest score of 100, which has the highest ranking score, as the first article. Note that the ranking score of a sports candidate article with an interest score of 80 or less is lower than the ranking score of a sports candidate article with an interest score of 100, and the ranking score of an entertainment candidate article with an interest score of 20 or less is lower than the ranking score of an entertainment candidate article with an interest score of 30.

[0131] As shown in Figure 10, in the selection of the second article, the sports candidate article with an interest score of 100 is excluded because it has already been selected by the selection unit 35. In the selection of the second article, since the display rank is "2", for the sports candidate article with an interest score of 80, interest score / display rank = 80 / 2, and for the entertainment candidate article with an interest score of 30, interest score / display rank = 30 / 2. In addition, the distribution score calculated by the distribution score calculation unit 41 of the calculation unit 33 is 1.9634 for the sports candidate article with an interest score of 80, and 0 for the entertainment candidate article with an interest score of 30.

[0132] Therefore, the ranking score calculated by the ranking score calculation unit 42 of the calculation unit 33 is -1.543 for a sports candidate article with an interest score of 80, and 0.15 for an entertainment candidate article with an interest score of 30. In this case, the selection unit 35 selects the entertainment candidate article with an interest score of 30, which has the highest ranking score, as the second article. Note that the ranking score of a sports candidate article with an interest score of 70 or less is lower than the ranking score of a sports candidate article with an interest score of 80, and the ranking score of an entertainment candidate article with an interest score of 20 or less is lower than the ranking score of an entertainment candidate article with an interest score of 30.

[0133] As shown in Figure 11, in the selection of the third article, the candidate sports article with an interest score of 100 and the candidate entertainment article with an interest score of 30 have already been selected by the selection unit 35 and are therefore excluded. In the selection of the third article, the display rank is "3", so for the candidate sports article with an interest score of 80, interest score / display rank = 80 / 3, and for the candidate entertainment article with an interest score of 20, interest score / display rank = 20 / 3. In addition, the distribution score calculated by the distribution score calculation unit 41 of the calculation unit 33 is 0.0580 for both the candidate sports article with an interest score of 80 and the candidate entertainment article with an interest score of 20.

[0134] Therefore, the ranking score calculated by the ranking score calculation unit 42 of the calculation unit 33 is -0.209 for a sports candidate article with an interest score of 80, and 0.009 for an entertainment candidate article with an interest score of 20. In this case, the selection unit 35 selects the sports candidate article with an interest score of 80, which has the highest ranking score, as the third article. Note that the ranking score of a sports candidate article with an interest score of 70 or less is lower than the ranking score of a sports candidate article with an interest score of 80, and the ranking score of an entertainment candidate article with an interest score of 15 or less is lower than the ranking score of an entertainment candidate article with an interest score of 20.

[0135] In this way, the information processing device 1 can mitigate the situation where the same category is selected consecutively. Furthermore, for users U for whom there is no information and an interest score cannot be calculated, the information processing device 1 selects articles from various categories, which can improve the daily click user (DCU), for example, especially among light users.

[0136] Here, we will explain the effect of selecting articles based on the ranking score calculated by the calculation unit 33. Figure 12 is a diagram illustrating the categories of multiple articles selected by the selection unit 35 of the information processing device 1 according to this embodiment. Figures 12(a) and (c) show the distribution of interest scores of users U who are interested in various categories, and Figures 12(b) and (d) show the distribution of interest scores of users U who are interested in a specific category.

[0137] Furthermore, Figures 12(a) and (b) show the state of articles selected using the interest score calculated by the calculation unit 33 but without using the distribution score, while Figures 12(c) and (d) show the state of articles selected using the ranking score calculated by the calculation unit 33.

[0138] In the examples shown in Figures 12(a) and (b), candidate articles are selected as articles to be provided in order of their interest score, resulting in only sports-related candidate articles being selected. On the other hand, in Figures 12(c) and (d), in addition to sports-related candidate articles, entertainment and economic candidate articles are also selected as articles to be provided. In this way, the information processing device 1 can control the diversity according to the user U.

[0139] Furthermore, the selection unit 35 can select articles to be provided to user U based on the ranking score calculated by the calculation unit 33 without using the display order. For example, the selection unit 35 can select a predetermined number of candidate articles from among several candidate articles for which a ranking score has been calculated, in descending order of ranking score, as articles to be provided to user U.

[0140] [3.3.7.Providing Department 36] The provision unit 36 ​​provides various information via the communication unit 10. For example, when the reception unit 32 receives a request for distribution of an article list, the provision unit 36 ​​provides the user U with an article information list containing information on multiple articles selected by the selection unit 35.

[0141] For example, when the receiving unit 32 receives a request for distribution of a list of articles, the providing unit 36 ​​can provide user U with information on multiple articles arranged in the order selected by the selection unit 35. In addition, when the receiving unit 32 receives a request for distribution of articles, the providing unit 36 ​​provides user U with the articles specified in the request for distribution of articles.

[0142] Figure 13 is a diagram showing an example of an article screen that includes a list of article information provided by the information processing device 1's provisioning unit 36 ​​and displayed on the terminal device 2 according to this embodiment. The article screen 60 shown in Figure 13 includes a tab column 61 and tab content 63. The tab column 61 includes a plurality of tabs 62a, 62b, 62c, 62d, and 62e that can be selected by the user U. Tab 62a is the tab "All", tab 62b is the tab "Sports", tab 62c is the tab "Entertainment", tab 62e is the tab "Economy", and tab 62e is the tab "Politics".

[0143] Although not shown in the diagram, tab column 61 also includes tabs other than tabs 62a, 62b, 62c, 62d, and 62e. User U can make tabs other than tabs 62a, 62b, 62c, 62d, and 62e selectable on terminal device 2 by swiping or scrolling. In the following, when multiple tabs including tabs 62a, 62b, 62c, 62d, and 62e are not individually distinguished, they may be referred to as tab 62.

[0144] In the tab content 63 shown in the figure, multiple article information items 64a, 64b, 64c, 64d, and 64e are arranged vertically. Each of the article information items 64a, 64b, 64c, 64d, and 64e is a UI that, when selected by user U, redirects to the linked article. For example, it includes information such as a thumbnail image of the article, a headline, and a link to the article. By selecting any of the multiple article information items, including article information items 64a, 64b, 64c, 64d, and 64e, user U can display the article corresponding to the selected article information on the terminal device 2.

[0145] Tab content 63 also includes article information other than article information 64a, 64b, 64c, 64d, and 64e. User U can select and display article information other than article information 64a, 64b, 64c, 64d, and 64e on terminal device 2 by swiping or scrolling.

[0146] In the example shown in Figure 13, article information is selected based on a ranking score using a distribution score in addition to an interest score. Articles 64a, 64c, and 64e are articles in the sports category, while article information 64b and 64d are articles in the entertainment category. In this way, the information processing device 1 can provide information on multiple articles more appropriately.

[0147] Furthermore, in the example shown in Figure 13, tab 62a is selected, and the target category distribution for the major category is determined by the determination unit 31. However, if tab 62b is selected by user U on the article screen 60, an article list distribution request including the specification of the category corresponding to tab 62b is sent from terminal device 2 to information processing device 1. In this case, the determination unit 31 determines the distribution of multiple subcategories, with the category corresponding to tab 62b as the major category, as the target category distribution. As a result, articles in multiple subcategories classified under the major category of sports are provided to user U, allowing the information processing device 1 to provide information on multiple articles more appropriately.

[0148] [4. Processing Procedure] Next, the procedure for information processing by the processing unit 12 of the information processing device 1 according to the embodiment will be described. Figure 14 is a flowchart showing an example of information processing by the processing unit 12 of the information processing device 1 according to the embodiment.

[0149] As shown in Figure 14, the processing unit 12 of the information processing device 1 determines whether or not the distribution determination timing has arrived (step S10). The distribution determination timing is, for example, a timing specified by the operator of the information processing device 1, a timing that occurs at a predetermined interval, or a timing when the number of times the list of article information is provided to the same user U reaches a predetermined number of times, but is not limited to these examples.

[0150] If the processing unit 12 determines that it is time to determine the distribution (step S10: Yes), it determines the category distribution of the articles provided to user U as the target category distribution (step S11).

[0151] If the processing in step S11 is completed, or if it is determined that the distribution determination timing has not yet arrived (step S10: No), the processing unit 12 determines whether or not the distribution timing has arrived (step S12). The distribution timing is, for example, the timing when the article list distribution request is received by the information processing device 1, the timing specified by the operator of the information processing device 1, or a timing that arrives at a predetermined interval, but is not limited to these examples.

[0152] If the processing unit 12 determines that it is time for delivery (step S12: Yes), it performs the process of providing the list of article information (step S13). The process of providing the list of article information in step S13 is the process shown in steps S20 to S25 in Figure 15, which will be described in detail later.

[0153] If the processing in step S13 is completed, or if it is determined that it is not time for distribution (step S12: No), the processing unit 12 determines whether it is time to terminate the operation (step S14). For example, the processing unit 12 determines that it is time to terminate the operation if the power to the information processing device 1 is turned off, or if it determines that a termination operation has been performed by operating on an unillustrated control unit of the information processing device 1.

[0154] If the processing unit 12 determines that it is not yet time to terminate the operation (step S14: No), it proceeds to step S10. If it determines that it is time to terminate the operation (step S14: Yes), it terminates the process shown in Figure 14.

[0155] Figure 15 is a flowchart showing an example of the article information list provision process by the processing unit 12 of the information processing device 1 according to the embodiment. As shown in Figure 15, the processing unit 12 calculates the user U's interest score for each candidate article (step S20). The processing unit 12 also performs weight adjustment (step S21).

[0156] Next, the processing unit 12 calculates a ranking score for each candidate article (step S22). Then, the processing unit 12 determines that the candidate article with the highest ranking score will be the article to be provided to user U (step S23).

[0157] If the processing in step S23 is completed, the processing unit 12 determines whether the number of articles to be provided determined in step S23 has reached a predetermined number (step S24). If the processing unit 12 determines that the number of articles to be provided determined in step S23 has not reached a predetermined number (step S24: No), the processing moves to step S22.

[0158] If the processing unit 12 determines that the number of articles to be provided, as determined in step S23, has reached a predetermined number (step S24: Yes), it provides the predetermined number of articles to be provided to user U (step S25) and terminates the process shown in Figure 15.

[0159] [5. Variations] The calculation unit 33 can exclude candidate articles with similar vectors from the calculation of the ranking score. In this case, the calculation unit 33 can calculate the vectors of the candidate articles using, for example, a text embedding model. The text embedding model is, for example, text-embedding-ada, but is not limited to such examples.

[0160] Furthermore, the calculation unit 33 can also calculate a ranking score using the posting order without excluding the candidate articles selected by the selection unit 35 as articles to be provided to user U. In this case, the selection unit 35 can perform the process of selecting the candidate article with the highest ranking score among the candidate articles other than the candidate article selected as an article to be provided to user U, according to the posting order, for each posting order.

[0161] Furthermore, while the above example used an article as an example of content, the content can also be other types of content (for example, videos, music, posted information, etc.).

[0162] For example, the calculation unit 33 can calculate a score for each candidate video based on the user U's interest score for the candidate video (an example of candidate content) and a distribution score corresponding to the difference between the category distribution of the videos selected to be provided to user U (an example of content) and the target category distribution determined by the decision unit 31. Then, the selection unit 35 selects a video to be provided to user U from among the multiple candidate videos based on the score for each candidate video calculated by the calculation unit 33. Subsequently, the provision unit 36 ​​provides user U with information on the multiple videos selected by the selection unit 35.

[0163] Furthermore, the calculation unit 33 can calculate a score for each candidate post information based on the interest score of user U for the candidate post information (an example of candidate content) and a distribution score corresponding to the difference between the category distribution of the post information (an example of content) selected to be provided to user U and the target category distribution determined by the decision unit 31.Then, the selection unit 35 selects the post information to be provided to user U from among the multiple candidate post information based on the score for each candidate post information calculated by the calculation unit 33.After that, the provision unit 36 ​​provides user U with information on the multiple post information selected by the selection unit 35.

[0164] Furthermore, the processing unit 12 of the information processing device 1 can select and provide multiple content items of two or more types. In this case, the processing unit 12 can treat the types of content as content categories.

[0165] For example, the calculation unit 33 can calculate a score for each candidate content based on the user U's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution (distribution of content types) of the content selected to be provided to user U and the target category distribution (distribution of target content types) determined by the decision unit 31. Then, the selection unit 35 selects the content to be provided to user U from among the multiple candidate content based on the score for each candidate content calculated by the calculation unit 33. After that, the provision unit 36 ​​provides user U with information on the multiple content selected by the selection unit 35.

[0166] [6. Hardware Configuration] Each of the information processing device 1 and terminal device 2 according to the above embodiment is implemented by a computer 80 having a configuration such as that shown in Figure 16. Figure 16 is a hardware configuration diagram showing an example of a computer 80 that implements the respective functions of the information processing device 1 and terminal device 2 according to the embodiment. The computer 80 has a CPU 81, RAM 82, ROM (Read Only Memory) 83, HDD (Hard Disk Drive) 84, communication interface (I / F) 85, input / output interface (I / F) 86, and media interface (I / F) 87.

[0167] The CPU 81 operates based on programs stored in the ROM 83 or HDD 84, and controls various parts of the system. The ROM 83 stores boot programs executed by the CPU 81 when the computer 80 starts up, as well as programs that depend on the computer 80's hardware.

[0168] HDD84 stores programs executed by CPU81 and data used by such programs. The communication interface85 receives data from other devices via network N (see Figure 2) and sends it to CPU81, and transmits the data generated by CPU81 to other devices via network N.

[0169] The CPU 81 controls output devices such as displays and printers, and input devices such as keyboards and mice, via the input / output interface 86. The CPU 81 acquires data from input devices via the input / output interface 86. The CPU 81 also outputs data it has generated to output devices via the input / output interface 86.

[0170] The media interface 87 reads a program or data stored in the recording medium 88 and provides it to the CPU 81 via the RAM 82. The CPU 81 loads the program from the recording medium 88 onto the RAM 82 via the media interface 87 and executes the loaded program. The recording medium 88 can be, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase Change Rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory.

[0171] For example, when the computer 80 functions as an information processing device 1 or terminal device 2 according to the embodiment, the CPU 81 of the computer 80 realizes the functions of the processing unit 12 by executing a program loaded on the RAM 82. The HDD 84 stores data from the storage unit 11. The CPU 81 of the computer 80 reads and executes these programs from the recording medium 88, but as another example, these programs may be obtained from other devices via a network N.

[0172] [7. Other] Furthermore, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically by known methods. In addition, the processing procedures, specific names, and various data and parameters shown in the above document and drawings can be changed at will unless otherwise specified. For example, the various information shown in each figure is not limited to the information shown.

[0173] Furthermore, the components of each illustrated device are functionally conceptual and do not necessarily need to be physically configured as shown. In other words, the specific forms of distribution and integration of each device are not limited to those shown, and all or part of them can be functionally or physically distributed and integrated in any unit according to various loads and usage conditions.

[0174] For example, the information processing device 1 described above may be implemented using multiple server computers, and the configuration can be flexibly changed, such as by calling external platforms via APIs or network computing depending on the function.

[0175] Furthermore, the embodiments and modifications described above can be combined as appropriate, provided that the processing content is not inconsistent.

[0176] [8. Effects] As described above, the information processing device 1 according to this embodiment comprises a determination unit 31, a calculation unit 33, a selection unit 35, and a provision unit 36. The determination unit 31 determines a target category distribution, which is the target category distribution of content provided to user U, based on information about user U. The content is, for example, articles, videos, music, and posted information, but is not limited to these examples. The calculation unit 33 calculates a score for each candidate content based on user U's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution of the content selected to be provided to user U and the target category distribution determined by the determination unit 31. The candidate content is, for example, candidate articles, candidate videos, candidate music, and candidate posted information, but is not limited to these examples. The selection unit 35 selects content to be provided to user U from among a plurality of candidate content based on the score for each candidate content calculated by the calculation unit 33. The provision unit 36 ​​provides user U with information on the plurality of content selected by the selection unit 35. This enables the information processing device 1 to provide information on multiple content more appropriately.

[0177] Furthermore, the calculation unit 33 calculates a score for each candidate content, excluding the candidate content selected by the selection unit 35 as content to be provided to user U. Each time the calculation unit 33 calculates a score for a candidate content, the selection unit 35 selects the candidate content with the highest score as content to be provided to user U. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0178] Furthermore, the calculation unit 33 calculates the score of candidate content by dividing the interest score by the number of selections up to the next content in the selection unit 35, and then subtracting the distribution score, weighted accordingly. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0179] Furthermore, it includes an adjustment unit 34 that adjusts the weights based on information about user U. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0180] Furthermore, the determination unit 31 determines the target category distribution using information including the user U's behavioral history with respect to content as information about user U. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0181] Furthermore, the determination unit 31 determines the target category distribution using information including the content provision history to user U as information about user U. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0182] Furthermore, the determination unit 31 determines the target category distribution using information including the context of user U as information about user U. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0183] Furthermore, the determination unit 31 determines the target category distribution for each predetermined number of content items. For example, the determination unit 31 determines the target category distribution each time a predetermined number of content items are selected by the selection unit 35. This allows the information processing device 1 to provide information on multiple content items more appropriately.

[0184] Although embodiments of the present application have been described in detail based on the drawings, these are illustrative examples, and the present invention can be implemented in various other forms, including those described in the disclosure section of the invention, based on the knowledge of those skilled in the art.

[0185] Furthermore, the terms "section, module, unit" mentioned above can be replaced with "means" or "circuit," etc. For example, the acquisition unit can be replaced with acquisition means or acquisition circuit. [Explanation of Symbols]

[0186] 1. Information Processing Device 2 Terminal devices 3. Submitter's Terminal 10 Communications Department 11 Storage section 12 Processing Units 20 User information storage unit 21 Article-related information storage section 30 Acquisition Department 31 Decision Section 32 Reception Department 33 Calculation Section 34 Adjustment part 35 Selection Section 36 Providing Department 40 Interest Score Calculation Unit 41 Distribution score calculation unit 42. Ranking Score Calculation Section 100 Information Processing Systems

Claims

1. A determination unit that determines a target category distribution, which is the target category distribution of the content provided to the user, based on information about the user, A calculation unit calculates a score for each candidate content based on the user's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution of the content selected to be provided to the user and the target category distribution determined by the decision unit. A selection unit selects content to be provided to the user from among a plurality of candidate content based on the score for each candidate content calculated by the calculation unit, The system includes a provisioning unit that provides information on a plurality of content selected by the selection unit to the user. An information processing device characterized by the following:

2. The calculation unit described above, The selection unit calculates a score for each candidate content, excluding the candidate content selected as content to be provided to the user. The aforementioned selection unit is Each time the calculation unit calculates a score for a candidate content, the candidate content with the highest score is selected as the content to be provided to the user. The information processing apparatus according to feature 1.

3. The calculation unit described above, The score of the candidate content is calculated by dividing the interest score by the number of selections up to the next content in the selection unit, and then subtracting the distribution score, weighted accordingly, from the result. The information processing apparatus according to feature 2.

4. The system includes an adjustment unit that adjusts the weight based on the information about the user. The information processing apparatus according to claim 3.

5. The aforementioned determination unit, The information including the user's behavioral history regarding the content is used as information about the user to determine the target category distribution. The information processing apparatus according to any one of claims 1 to 3.

6. The aforementioned determination unit, The target category distribution is determined using information including the content provision history to the aforementioned user as information about the aforementioned user. The information processing apparatus according to feature 5.

7. The aforementioned determination unit, The target category distribution is determined using information including the user's context as information about the user. The information processing apparatus according to feature 5.

8. The aforementioned determination unit, The target category distribution is determined for each predetermined number of content items. The information processing apparatus according to any one of claims 1 to 3.

9. A method of information processing performed by a computer, A decision step of determining a target category distribution, which is the target category distribution of the content provided to the user, based on information about the user, A calculation step for calculating a score for each candidate content based on the user's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution of the content selected to be provided to the user and the target category distribution determined by the decision step, A selection step in which, based on the score for each candidate content calculated in the calculation step, a selection step is made to select the content to be provided to the user from among the multiple candidate contents, The process includes a provisioning step of providing information on the multiple contents selected by the selection step to the user. An information processing method characterized by the following:

10. A decision procedure for determining the target category distribution, which is the goal of the category distribution of content provided to the user, based on information about the user, A calculation procedure for calculating a score for each candidate content based on the user's interest score for the candidate content and a distribution score corresponding to the difference between the category distribution of the content selected to be provided to the user and the target category distribution determined by the decision procedure, A selection procedure to select content to be provided to the user from among multiple candidate content based on the score for each candidate content calculated by the calculation procedure described above, The computer is instructed to perform a provision procedure that provides information on the multiple contents selected by the selection procedure to the user. An information processing program characterized by the following features.

Images