Article recommendation method and system

Abstract

The invention discloses an article recommendation method and system and belongs to the field of information processing. The method comprises the steps of obtaining an article tag corresponding to a first user, wherein the article tag corresponding to the first user is used for representing article content interested by the first user; an article knowledge graph is obtained, the article knowledge graph comprises article entities and connection relations between the article entities, the types of the article entities in the article knowledge graph comprise articles and article tags, and the article tags in the article knowledge graph are used for representing article content; obtaining at least one target article according to an article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article is matched with the article tag corresponding to the first user; and recommending the at least one target article to the first user. According to the method, the article interested by the user can be recommended to the user in combination with the article knowledge graph, so that the article recommendation pertinence is relatively high.

Description

Technical Field

[0001] The present application relates to the field of information processing, and in particular to an article recommendation method and system. Background Art

[0002] An article recommendation system is a system that can recommend articles to users, which can help users quickly obtain the required articles from a large number of articles.

[0003] At present, when users browse articles, the article recommendation system usually records the number of page views of the article, and recommends articles with higher page views to users based on the number of page views. However, the current article recommendation system recommends the same articles to different users, resulting in poor targeting of recommended articles. Summary of the invention

[0004] The present application provides an article recommendation method and system, which combines the article knowledge graph to recommend articles that the user is interested in, and the recommended articles are highly targeted. The technical solution of the present application is as follows.

[0005] In a first aspect, a method for recommending articles is provided, the method comprising:

[0006] Acquire an article tag corresponding to a first user, where the article tag corresponding to the first user is used to represent the content of the article that the first user is interested in;

[0007] Obtaining an article knowledge graph, wherein the article knowledge graph includes article entities and connection relationships between article entities, the types of article entities in the article knowledge graph include articles and article tags, and the article tags in the article knowledge graph are used to represent article content;

[0008] Acquire at least one target article according to the article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article matches the article tag corresponding to the first user;

[0009] The at least one target article is recommended to the first user.

[0010] Optionally, the type of article entity in the article knowledge graph further includes article popularity, and obtaining at least one target article according to the article tag corresponding to the first user and the article knowledge graph includes:

[0011] Acquire multiple articles according to the article tags corresponding to the first user and the article knowledge graph, wherein the article tags of the multiple articles match the article tags corresponding to the first user;

[0012] Obtaining the article popularity of each of the multiple articles according to the article knowledge graph;

[0013] According to the article popularity of the multiple articles, the at least one target article is obtained from the multiple articles, and the article popularity of the at least one target article is greater than a preset popularity.

[0014] Optionally, obtaining the article knowledge graph includes:

[0015] Get multiple articles;

[0016] According to the article content of each of the plurality of articles, obtaining at least one article tag for characterizing the article content of each of the plurality of articles;

[0017] Acquire multiple article entities and relationships between the multiple article entities according to the multiple articles and article tags of the multiple articles;

[0018] The article knowledge graph is generated according to the multiple article entities and the relationships between the multiple article entities.

[0019] Optionally, the acquiring of the article knowledge graph further includes: acquiring the article popularity of each of the plurality of articles according to the interactive behavior data corresponding to each of the plurality of articles, wherein the interactive behavior data corresponding to each of the plurality of articles is data generated by the user's interactive behavior with each of the plurality of articles, and the interactive behavior data includes at least one of browsing data, like data, collection data, sharing data, and scoring data;

[0020] The obtaining of multiple article entities and the relationships between the multiple article entities based on the multiple articles and the article tags of the multiple articles includes: obtaining the relationships between the multiple article entities and the multiple article entities based on the multiple articles, the article tags of the multiple articles and the article popularity of the multiple articles.

[0021] Optionally, each of the plurality of articles corresponds to a plurality of pieces of interaction behavior data, and the interaction type corresponding to each of the plurality of interaction behavior data includes any one of browsing, liking, collecting, analyzing, and scoring; and obtaining the article popularity of each of the plurality of articles according to the interaction behavior data corresponding to each of the plurality of articles includes:

[0022] For each interactive behavior data corresponding to each article in the plurality of articles: obtaining a time decay score of each interactive behavior data according to the interaction time corresponding to each interactive behavior data; and obtaining a score weight of each interactive behavior data according to the interaction type corresponding to each interactive behavior data, wherein the score weight of each interactive behavior data is used to characterize the importance of the interactive behavior that generates each interactive behavior data;

[0023] The article popularity of each of the multiple articles is obtained according to the time decay scores of the multiple pieces of interaction behavior data corresponding to each of the multiple articles and the score weights of the multiple pieces of interaction behavior data.

[0024] Optionally, obtaining the article tag corresponding to the first user includes: obtaining a user portrait of the first user, the user portrait including the article tag corresponding to the first user.

[0025] Optionally, obtaining the user portrait of the first user includes:

[0026] Acquire the interactive behavior data of the first user, where the interactive behavior data of the first user is data generated by the first user's interactive behavior with the article, and the interactive behavior data of the first user includes at least one of browsing data, like data, collection data, sharing data, and scoring data;

[0027] Acquire article content that the first user is interested in based on the interaction behavior data of the first user;

[0028] Acquire an article tag corresponding to the first user according to the article content that the first user is interested in;

[0029] Generate a user profile of the first user according to the article tag corresponding to the first user.

[0030] Optionally, recommending the at least one target article to the first user includes:

[0031] For each target article in the at least one target article, obtaining an estimated score of each target article for the first user according to the interaction behavior data of at least one second user strongly associated with the first user and each target article and the association degree between the first user and the at least one second user, wherein the estimated score of each target article for the first user is used to characterize the degree of matching between the article content of each target article and the article content that the first user is interested in, and the association degree between the first user and any of the second users is determined according to the interaction behavior data of the first user and at least one article and the interaction behavior data of the second user and the at least one article;

[0032] sorting the at least one target article according to an estimated score of the at least one target article for the first user;

[0033] The at least one target article is recommended to the first user in the sorted order.

[0034] In a second aspect, an article recommendation system is provided, the system comprising:

[0035] A first acquisition module is used to acquire an article tag corresponding to a first user, where the article tag corresponding to the first user is used to represent the article content that the first user is interested in;

[0036] A second acquisition module is used to acquire an article knowledge graph, wherein the article knowledge graph includes article entities and connection relationships between article entities, and the types of article entities in the article knowledge graph include articles and article tags. The article tags in the article knowledge graph are used to represent the content of the article;

[0037] A third acquisition module is used to acquire at least one target article according to the article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article matches the article tag corresponding to the first user;

[0038] A recommendation module is used to recommend the at least one target article to the first user.

[0039] Optionally, the type of article entity in the article knowledge graph further includes article popularity, and the third acquisition module is used to:

[0040] Acquire multiple articles according to the article tags corresponding to the first user and the article knowledge graph, wherein the article tags of the multiple articles match the article tags corresponding to the first user;

[0041] Obtaining the article popularity of each of the multiple articles according to the article knowledge graph;

[0042] According to the article popularity of the multiple articles, the at least one target article is obtained from the multiple articles, and the article popularity of the at least one target article is greater than a preset popularity.

[0043] Optionally, the second acquisition module is used to:

[0044] Get multiple articles;

[0045] According to the article content of each of the plurality of articles, obtaining at least one article tag for characterizing the article content of each of the plurality of articles;

[0046] Acquire multiple article entities and relationships between the multiple article entities according to the multiple articles and article tags of the multiple articles;

[0047] The article knowledge graph is generated according to the multiple article entities and the relationships between the multiple article entities.

[0048] Optionally, the second acquisition module is used to:

[0049] Obtaining the article popularity of each article according to the interactive behavior data corresponding to each of the plurality of articles, wherein the interactive behavior data corresponding to each article is data generated by the user's interactive behavior with each of the articles, and the interactive behavior data includes at least one of browsing data, like data, collection data, sharing data, and scoring data;

[0050] According to the multiple articles, the article tags of the multiple articles and the article popularity of the multiple articles, the multiple article entities and the relationships between the multiple article entities are obtained.

[0051] Optionally, each of the plurality of articles corresponds to a plurality of pieces of interaction behavior data, and the interaction type corresponding to each piece of interaction behavior data in the plurality of interaction behavior data includes any one of browsing, liking, collecting, analyzing and scoring; the second acquisition module is used to:

[0052] For each interactive behavior data corresponding to each article in the plurality of articles: obtaining a time decay score of each interactive behavior data according to the interaction time corresponding to each interactive behavior data; and obtaining a score weight of each interactive behavior data according to the interaction type corresponding to each interactive behavior data, wherein the score weight of each interactive behavior data is used to characterize the importance of the interactive behavior that generates each interactive behavior data;

[0053] The article popularity of each of the multiple articles is obtained according to the time decay scores of the multiple pieces of interaction behavior data corresponding to each of the multiple articles and the score weights of the multiple pieces of interaction behavior data.

[0054] Optionally, the first acquisition module is used to acquire a user portrait of the first user, where the user portrait includes an article tag corresponding to the first user.

[0055] Optionally, the first acquisition module is used to:

[0056] Acquire the interactive behavior data of the first user, where the interactive behavior data of the first user is data generated by the first user's interactive behavior with the article, and the interactive behavior data of the first user includes at least one of browsing data, like data, collection data, sharing data, and scoring data;

[0057] Acquire article content that the first user is interested in based on the interaction behavior data of the first user;

[0058] Acquire an article tag corresponding to the first user according to the article content that the first user is interested in;

[0059] Generate a user profile of the first user according to the article tag corresponding to the first user.

[0060] Optionally, the recommendation module is used to:

[0061] For each target article in the at least one target article, obtaining an estimated score of each target article for the first user according to the interaction behavior data of at least one second user strongly associated with the first user and each target article and the association degree between the first user and the at least one second user, wherein the estimated score of each target article for the first user is used to characterize the degree of matching between the article content of each target article and the article content that the first user is interested in, and the association degree between the first user and any of the second users is determined according to the interaction behavior data of the first user and at least one article and the interaction behavior data of the second user and the at least one article;

[0062] sorting the at least one target article according to an estimated score of the at least one target article for the first user;

[0063] The at least one target article is recommended to the first user in the sorted order.

[0064] In a third aspect, an article recommendation system is provided, including a memory and a processor;

[0065] The memory is used to store computer programs;

[0066] The processor is used to execute the computer program stored in the memory so that the article recommendation system performs the article recommendation method provided by the first aspect or any optional implementation of the first aspect.

[0067] In a fourth aspect, a terminal device is provided, which may be a smart phone, a tablet computer, a laptop computer or a desktop computer, and the terminal device includes an article recommendation system as provided in the second aspect or any optional implementation of the second aspect, or includes an article recommendation system as provided in the third aspect.

[0068] In a fifth aspect, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed, the article recommendation method provided in the first aspect or any optional method of the first aspect is implemented.

[0069] In a sixth aspect, a computer program product is provided, the computer program product comprising a program or code, and when the program or code is executed, the article recommendation method provided in the first aspect or any optional manner of the first aspect is implemented.

[0070] The beneficial effects of the technical solution provided by this application include:

[0071] The article recommendation method and system provided by the present application, the article recommendation system obtains the article tag and article knowledge graph corresponding to the first user, obtains at least one target article according to the article tag corresponding to the first user and the knowledge graph, and recommends the at least one target article to the first user, wherein the article tag corresponding to the first user is used to characterize the content of the article that the first user is interested in, and the article tag of the at least one target article matches the article tag corresponding to the first user. Therefore, the present application recommends articles of interest to the user according to the content of the article that the user is interested in in combination with the article knowledge graph, and the recommended articles are highly targeted. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0073] Figure 1 is a flow chart of an article recommendation method provided in an embodiment of the present application;

[0074] Figure 2 It is a schematic diagram of an article knowledge graph provided in an embodiment of the present application;

[0075] Figure 3 is a schematic diagram of another article knowledge graph provided in an embodiment of the present application;

[0076] Figure 4 This is a schematic diagram of digitizing article tags provided by an embodiment of the present application;

[0077] Figure 5 is a schematic diagram of an article recommendation system provided in an embodiment of the present application;

[0078] Figure 6 It is a schematic diagram of another article recommendation system provided in an embodiment of the present application.

[0079] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0081] Users often need to search for articles to obtain the knowledge they want to know. For example, in the current technological environment, R&D personnel in the field of oil refining and chemical industry often need to find and expand related articles in order to better understand new technologies, new materials and new methods, thereby improving R&D efficiency. In traditional technology, users need to manually search for articles to obtain relevant knowledge, such as going to the library to read relevant articles or searching for relevant articles on the Internet. However, it is difficult for users to manually search for articles, and the search process is complicated, resulting in a high time cost for R&D projects. In order to solve the problems of traditional technology, article recommendation systems came into being. Article recommendation systems are recommendation systems that can recommend articles to users, which can help users quickly obtain the required articles from a large number of articles. The current article recommendation system recommends articles to users based on the number of views of the articles, and the articles recommended to different users are the same, resulting in poor pertinence of the recommended articles.

[0082] The embodiment of the present application provides an article recommendation method and system, and the article recommendation method is performed by an article recommendation system. The article recommendation system recommends articles whose article tags match the article tags corresponding to the user to the user according to the article tags corresponding to the user and the article knowledge graph. The article tags corresponding to the user are used to characterize the article content that the user is interested in. Therefore, the embodiment of the present application recommends articles to the user according to the article content that the user is interested in in combination with the article knowledge graph. The articles recommended to different users are usually different, so the recommended articles are more targeted.

[0083] The technical solution of the present application is introduced below, and firstly, an embodiment of the article recommendation method of the present application is introduced.

[0084] Please refer to Figure 1 , which shows a flow chart of an article recommendation method provided in an embodiment of the present application. The article recommendation method is executed by an article recommendation system. The article recommendation method includes the following steps S101 to S104.

[0085] S101. Obtain an article tag corresponding to a first user, where the article tag corresponding to the first user is used to represent the content of the article that the first user is interested in.

[0086] Wherein, the first user is any user registered in the article recommendation system. When the first user logs in to the article recommendation system, the article recommendation system obtains the article tag corresponding to the first user. In one embodiment, the article recommendation system stores the article tag corresponding to the first user, and the article recommendation system obtains the article tag corresponding to the first user stored in the article recommendation system. Wherein, the article tag corresponding to the first user stored in the article recommendation system may be the article tag set by the first user when registering in the article recommendation system, or may be the article tag generated by the article recommendation system according to the interactive behavior data of the first user. The interactive behavior data of the first user is the data generated by the first user interacting with the article managed by the article recommendation system. The interactive behavior data of the first user includes at least one of browsing data, like data, collection data, sharing data and scoring data. The browsing data of the first user is the data generated by the first user browsing the article managed by the article recommendation system, the like data of the first user is the data generated by the first user liking the article managed by the article recommendation system, the collection data of the first user is the data generated by the first user collecting the article managed by the article recommendation system, and the scoring data of the first user is the data generated by the first user scoring the article managed by the article recommendation system. For example, the score of the scoring data is 1 to 5 points.

[0087] In an optional embodiment, the article recommendation system obtains the interactive behavior data of the first user, the article recommendation system determines the article content that the first user is interested in based on the interactive behavior data of the first user, and the article recommendation system obtains the article tag corresponding to the first user based on the article content that the first user is interested in. In a specific embodiment, the article recommendation system obtains multiple interactive behavior data of the first user, each interactive behavior data of the multiple interactive behavior data corresponds to an article, the article recommendation system determines the article content that the first user is interested in based on the article content corresponding to the multiple interactive behavior data, and the article recommendation system generates the article tag corresponding to the first user based on the article content that the first user is interested in.

[0088] In an optional embodiment, the article recommendation system obtains a user portrait of the first user, and the user portrait of the first user includes the article tag corresponding to the first user. In one embodiment, the article recommendation system stores a user portrait of the first user, and when the first user logs into the article recommendation system, the article recommendation system obtains the user portrait of the first user stored in the article recommendation system. The user portrait of the first user stored in the article recommendation system is generated by the article recommendation system based on the interactive behavior data of the first user. The following describes the process of the article recommendation system generating the user portrait of the first user based on the interactive behavior data of the first user.

[0089] In an optional embodiment, the article recommendation system obtains the interactive behavior data of the first user, the article recommendation system determines the content of the article that the first user is interested in based on the interactive behavior data of the first user, the article recommendation system obtains the article tag corresponding to the first user based on the content of the article that the first user is interested in, and the article recommendation system generates a user portrait of the first user based on the article tag corresponding to the first user. In a specific embodiment, the article recommendation system obtains multiple interactive behavior data of the first user, each interactive behavior data of the multiple interactive behavior data corresponds to an article, and each interactive behavior data is data generated when the first user interacts with the corresponding article, the article recommendation system determines the article content corresponding to the multiple interactive behavior data, the article recommendation system determines the article content that the first user is interested in based on the article content that the first user is interested in, the article recommendation system generates at least one article tag corresponding to the first user based on the article content that the first user is interested in, and the article recommendation system generates a user portrait of the first user based on the at least one article tag corresponding to the first user. In a specific embodiment, the at least one article tag is a plurality of article tags, and the article recommendation system generates a user portrait of the first user based on the plurality of article tags. For example, the article recommendation system generates a user portrait of the first user based on article tags whose repetition rate is greater than a preset repetition rate among the plurality of article tags. The repetition rate of any article tag among the plurality of article tags refers to the probability of the any article tag appearing in the plurality of article tags.

[0090] In an optional embodiment, the article recommendation system determines the article content that the first user is interested in based on the article content corresponding to the multiple interactive behavior data of the first user, including: the article recommendation system obtains the score weight of each interactive behavior data in the multiple interactive behavior data and the time decay score of each interactive behavior, the article recommendation system determines the weighted time decay score of each interactive behavior data according to the score weight of each interactive behavior data and the time decay score of each interactive behavior, and the article recommendation system determines the article content that the first user is interested in based on the weighted time decay score of the multiple interactive behavior data. In one embodiment, for each interactive behavior data in the multiple interactive behavior data, the article recommendation system determines the product of the score weight of each interactive behavior data and the time decay score of each interactive behavior data as the weighted time decay score of each interactive behavior data. The article recommendation system determines the article content corresponding to the interactive behavior data in the multiple interactive behavior data whose weighted time decay score is greater than the weighted time decay score threshold as the article content that the first user is interested in.

[0091] In an optional embodiment, the article recommendation system obtains the score weight of each interactive behavior data, including: for each interactive behavior data of the multiple interactive behavior data of the first user, the article recommendation system obtains the interaction type corresponding to each interactive behavior data, and the article recommendation system obtains the score weight of each interactive behavior data according to the interaction type corresponding to each interactive behavior data. Among them, the interaction type corresponding to each interactive behavior data is the type of interactive behavior corresponding to each interactive behavior data, and the score weight of each interactive behavior data is the score weight corresponding to the interaction type corresponding to each interactive behavior data. In the embodiment of the present application, the interactive behavior data includes at least one of browsing data, like data, collection data, sharing data and scoring data, the interaction type of browsing data is browsing (or called browsing type), the interaction type of like data is like (or called like type), the interaction type of collection data is collection (or called collection type), the interaction type of sharing data is sharing (or called sharing type), and the interaction type of scoring data is scoring (or called scoring type). In one embodiment, the article recommendation system stores a score weight corresponding to each interaction type among multiple interaction types (such as browsing, liking, collecting, sharing and scoring). For each interaction behavior data of the multiple interaction behavior data of the first user, the article recommendation system obtains the score weight corresponding to the interaction type corresponding to each interaction behavior data, and the article recommendation system determines the score weight corresponding to the interaction type corresponding to each interaction behavior data as the score weight of each interaction behavior data. For example, the article recommendation system stores a score weight A1 corresponding to the browsing type, a score weight A2 corresponding to the like type, a score weight A3 corresponding to the collection type, a score weight A4 corresponding to the sharing type, and a score weight A5 corresponding to the scoring type; for each interactive behavior data of the first user: if the interaction type corresponding to each interactive behavior data is the browsing type, the article recommendation system determines that the score weight of each interactive behavior data is A1; if the interaction type corresponding to each interactive behavior data is the like type, the article recommendation system determines that the score weight of each interactive behavior data is A2; if the interaction type corresponding to each interactive behavior data is the collection type, the article recommendation system determines that the score weight of each interactive behavior data is A3; if the interaction type corresponding to each interactive behavior data is the sharing type, the article recommendation system determines that the score weight of each interactive behavior data is A4; if the interaction type corresponding to each interactive behavior data is the scoring type, the article recommendation system determines that the score weight of each interactive behavior data is A5.

[0092] In an optional embodiment, the article recommendation system obtains the time decay score of each interactive behavior, including: for each interactive behavior data of the first user, the article recommendation system obtains the interaction time corresponding to each interactive behavior data, and the article recommendation system determines the time decay score of each interactive behavior data according to the interaction time corresponding to each interactive behavior data. Among them, the interaction time corresponding to each interactive behavior data of the first user is the time when each interactive behavior data is generated, and is also the time when the first user interacts with the article corresponding to each interactive behavior data, such as the moment when the first user likes the article corresponding to each interactive behavior data, the moment when the first user collects the article corresponding to each interactive behavior data, and any moment in the time period when the first user browses the article corresponding to each interactive behavior data (such as the start time or the end time). In a specific embodiment, for each of the plurality of interaction behavior data of the first user, the article recommendation system determines the time difference between the interaction time corresponding to each interaction behavior data and the current moment (for example, the moment when the article recommendation system obtains the time decay score of each interaction behavior data), and the article recommendation system determines the time decay score of each interaction behavior data according to the time difference between the interaction time corresponding to each interaction behavior data and the current moment and the time decay score formula. The time decay score formula is: S = e -a×t , S represents the time decay score of the interaction behavior data, t represents the time difference between the interaction time corresponding to the interaction behavior data and the current moment, and a represents the time decay coefficient, which is usually a constant, for example, a is 0.099. According to the time decay score formula, the smaller the time difference between the interaction time corresponding to the interaction behavior data and the current moment, the greater the time decay score of the interaction behavior data; the greater the time difference between the interaction time corresponding to the interaction behavior data and the current moment, the smaller the time decay score of the interaction behavior data.

[0093] In an optional embodiment, the article recommendation system generates a user portrait of the first user based on multiple article tags corresponding to the first user, including: the article recommendation system determines the repetition rate of each article tag in the multiple article tags, the article recommendation system determines at least one article tag in the multiple article tags whose repetition rate is greater than a preset repetition rate, and the article recommendation system generates a user portrait of the first user based on the at least one article tag whose repetition rate is greater than the preset repetition rate. For example, the article recommendation system depicts the user portrait of the first user based on the at least one article tag, and specifically, the article recommendation system associates and stores the at least one article tag with the user identifier (e.g., user account) of the first user.

[0094] In an optional embodiment, the article recommendation system determines the article content that the first user is interested in based on the article content corresponding to the multiple interactive behavior data of the first user, generates multiple article tags corresponding to the first user based on the article content that the first user is interested in, and generates a user portrait of the first user based on the multiple article tags corresponding to the first user, that is, the article recommendation system generates a user portrait of the first user based on the multiple interactive behavior data of the first user. In this embodiment, the score weight of each interactive behavior data of the first user is used to characterize the influence of each interactive behavior data on the user portrait of the first user. The interactive behavior data with a larger score weight has a greater influence on the user portrait of the first user, and the interactive behavior data with a smaller score weight has a smaller influence on the user portrait of the first user; the time decay score of each interactive behavior data of the first user is used to characterize the influence of each interactive behavior data on the user portrait of the first user. The interactive behavior data with a larger time decay score has a greater influence on the user portrait of the first user, and the interactive behavior data with a smaller time decay score has a smaller influence on the user portrait of the first user; the repetition rate of each article tag corresponding to the first user is used to characterize the influence of each article tag on the user portrait of the first user. The article tag with a larger repetition rate has a greater influence on the user portrait of the first user; the article tag with a smaller repetition rate has a smaller influence on the user portrait of the first user. In some embodiments, the repetition rate of the article tag is also referred to as the tag score of the article tag.

[0095] In an optional embodiment, the article recommendation system includes an interactive behavior database, which includes multiple interactive behavior data of at least one user, each of the multiple interactive behavior data is data generated by an interactive behavior between a user and an article managed by the article recommendation system, the at least one user includes a first user, and the article recommendation system obtains multiple interactive behavior data of the first user in the interactive behavior database. Among them, the interactive behavior database is generated by the article recommendation system according to the operation log generated by the at least one user operating the article managed by the article recommendation system, the article recommendation system obtains the operation log of the at least one user in the article recommendation system, and the article recommendation system generates the interactive behavior database through the operation log of the at least one user in the article recommendation system. In a specific embodiment, the article recommendation system obtains the multiple interactive behavior data of the at least one user from the operation log of the at least one user in the article recommendation system, the article recommendation system performs structured processing on the multiple interactive behavior data respectively, and the article recommendation system generates the interactive behavior database according to the multiple interactive behavior data after structured processing. The article recommendation system performs structured processing on the multiple interactive behavior data respectively, which means converting the unstructured data in the multiple interactive behavior data into structured data. The unstructured data refers to data that does not use a fixed structure, and the structured data refers to data that uses a fixed structure. Optionally, each interactive behavior data includes interaction time, interaction type, interactive user (referring to the user who has interactive behavior), and interactive article (referring to the article that has interactive behavior with the user, such as the article that the user likes). In the operation log, the interaction time is the international standard time. In the process of the article recommendation system performing structured processing on each interactive behavior data, the article recommendation system converts the international standard type of interaction time included in each interactive behavior data into a time of a timestamp type. The time of the international standard type is, for example, YYYY-MM-DDTHH:mm:ss.sssZ, where YYYY represents a four-digit year, MM represents a month, DD represents a day, T is a separator between date and time, HH represents hours, mm represents minutes, ss.sss represents seconds and milliseconds, and Z represents the time zone. The time of the timestamp type is the number of seconds or milliseconds from 00:00:00 on January 1, 1970 (converted to 8:00:00 on January 1, 1970, Beijing time). Optionally, the article recommendation system deletes invalid data from the multiple interactive behavior data obtained from the operation log before performing structured processing on the multiple interactive behavior data. The invalid data includes but is not limited to interactive behavior data with missing interaction time, interactive behavior data with missing interaction type, and interactive behavior data with missing interactive users.Optionally, the interactive behavior database also includes sample types of interactive behavior data. The sample types can be positive samples or negative samples. For example, browsing data, like data, collection data, sharing data, and scoring data with a score greater than 3 points are all positive samples, and scoring data with a score less than 3 points are negative samples.

[0096] S102. Obtain an article knowledge graph, which includes article entities and connection relationships between article entities. The types of article entities in the article knowledge graph include articles and article tags. The article tags in the article knowledge graph are used to represent article content.

[0097] When the article recommendation system recommends an article to the first user, the article recommendation system obtains the article knowledge graph. For example, the article recommendation system stores the article knowledge graph, the article recommendation system obtains the article knowledge graph stored by the article recommendation system, or the article recommendation system obtains the article knowledge graph from a server. Among them, the article knowledge graph includes article entities and connection relationships between article entities, and the types of article entities include articles and article tags. Article tags are used to characterize article content. For example, article tags include keywords, authors (that is, the author of the article), article types (that is, the type of article content, for example, article types include computer science, chemistry, etc.), etc. In one embodiment, please refer to Figure 2 , which shows a schematic diagram of an article knowledge graph provided by an embodiment of the present application. The article knowledge graph includes articles 1 to 4, article tags 1 to 8, and one-way arrow lines connecting articles 1 to 4 and article tags 1 to 8. The one-way arrow line points from the article tag to the article. The one-way arrow line is used to represent the relationship between articles 1 to 4 and article tags 1 to 8. The article entity connected to the starting end of the one-way arrow line is the attribute of the article entity connected to the end of the one-way arrow line. Specifically, Figure 2As shown, article 1 is connected to article tags 1 to 3 respectively, and the one-way arrow line between article 1 and article tags 1 to 3 indicates that article tags 1 to 3 are article tags of article 1; article 2 is connected to article tag 1, article tag 3, article tag 4, and article tag 5 respectively, and the one-way arrow line between article 2 and article tag 1, article tag 3, article tag 4, and article tag 5 indicates that article tag 1, article tag 3, article tag 4, and article tag 5 are all article tags of article 2; article 3 is connected to article tag 3, article tag 7, and article tag 8 respectively, and the one-way arrow line between article 3 and article tag 3, article tag 7, and article tag 8 indicates that article tag 3, article tag 7, and article tag 8 are all article tags of article 3; article 4 is connected to article tag 6 and article tag 8 respectively, and the one-way arrow line between article 4 and article tag 6 and article tag 8 indicates that article tag 6 and article tag 8 are all article tags of article 4. In a specific example, articles 1 to 4 are all articles in the field of chemistry, for example, article 1 is "Simple Chemical Operations" (i.e., the name of article 1 is "Simple Chemical Operations"), article 2 is "Introduction to Toluene" (i.e., the name of article 2 is "Introduction to Toluene"), article 3 is "Use of Cyclohexane" (i.e., the name of article 3 is "Use of Cyclohexane"), and article 4 is "Introduction to Cyclohexane" (i.e., the name of article 4 is "Introduction to Cyclohexane"), article label 1 is "Author 1", article label 2 is "Filtration", article label 3 is "Catalyst Separation", article label 4 is "Tube Reactor", article label 5 is "Toluene", article label 6 is "Author 2", article label 7 is "Test Tube", and article label 8 is "Cyclohexane". The article "Simple Chemical Operations" is connected to the article labels "Author 1", "Filtration", and "Catalyst Separation" respectively, and the one-way arrow line between the article "Simple Chemical Operations" and the article labels "Author 1", "Filtration", and "Catalyst Separation" indicates that the author of the article "Simple Chemical Operations" is "Author 1" and the content includes "Filtration" and "Catalyst Separation". The article “Introduction to Toluene” is connected to the article labels “Author 1”, “Catalyst Separation”, “Tubular Reactor” and “Toluene” respectively. The one-way arrow line between the article “Introduction to Toluene” and the article labels “Author 1”, “Catalyst Separation”, “Tubular Reactor” and “Toluene” indicates that the author of the article “Introduction to Toluene” is “Author 1” and the content includes “Catalyst Separation”, “Tubular Reactor” and “Toluene”. Figure 2 Only an example of an article knowledge graph is shown. In actual applications, the article knowledge graph may include more article entities and more complex connection relationships, which is not limited in the embodiments of the present application.

[0098] In an optional embodiment, the type of article entity in the article knowledge graph also includes article popularity, which is used to characterize the popularity of article access. The article popularity of any article can be determined based on the user's interactive behavior data on the article. Figure 3 , which shows a schematic diagram of another article knowledge graph provided by an embodiment of the present application. The entities of the knowledge graph include articles 1 to 4, article tags 1 to 8, article popularity 1 to 4, one-way arrow lines connecting articles 1 to 4 with article tags 1 to 8, and one-way arrow lines connecting articles 1 to 4 with article popularity 1 to 4. The one-way arrow lines connecting articles 1 to 4 with article tags 1 to 8 point from article tags to articles. These one-way arrow lines are used to indicate the relationship between articles 1 to 4 and article tags 1 to 8. The one-way arrow lines connecting articles 1 to 4 with article popularity 1 to 4 point from article popularity to articles. These one-way arrow lines are used to indicate the relationship between articles 1 to 4 and article popularity 1 to 4. The article entity connected to the starting end of each one-way arrow line is the attribute of the article entity connected to the end of the one-way arrow line. Specifically, as Figure 3 As shown, article 1 is connected to article tags 1 to 3 and article heat 1 respectively, and the one-way arrow line between article 1 and article tags 1 to 3 indicates that article tags 1 to 3 are article tags of article 1, and the one-way arrow line between article 1 and article heat 1 indicates that article heat 1 is the article heat of article 1; article 2 is connected to article tag 1, article tag 3, article tag 4, article tag 5, and article heat 2 respectively, and the one-way arrow line between article 2 and article tag 1, article tag 3, article tag 4, and article tag 5 indicates that article tag 1, article tag 3, article tag 4, and article tag 5 are all article tags of article 2, and the one-way arrow line between article 2 and article heat 2 indicates that article heat 2 is the article heat of article 2. Article popularity; Article 3 is connected to Article Tag 3, Article Tag 7, Article Tag 8, and Article Popularity 3 respectively, and the one-way arrow line between Article 3 and Article Tag 3, Article Tag 7, and Article Tag 8 indicates that Article Tag 3, Article Tag 7, and Article Tag 8 are all article tags of Article 3, and the one-way arrow line between Article 3 and Article Popularity 3 indicates that Article Popularity 3 is the article popularity of Article 3; Article 4 is connected to Article Tag 6, Article Tag 8, and Article Popularity 4 respectively, and the one-way arrow line between Article 4 and Article Tag 6, and Article Tag 8 indicates that Article Tag 6 and Article Tag 8 are all article tags of Article 4, and the one-way arrow line between Article 4 and Article Popularity 4 indicates that Article Popularity 4 is the article popularity of Article 4.

[0099] Among them, the article knowledge graph can be pre-generated by the article recommendation system according to the articles managed by the article recommendation system. In a specific embodiment, the article recommendation system obtains multiple articles managed by the article recommendation system, and the article recommendation system obtains at least one article tag for characterizing the article content of each article according to the article content of each of the multiple articles, and the article recommendation system obtains the relationship between multiple article entities and the multiple article entities according to the multiple articles and the article tags of the multiple articles, and the article recommendation system generates the article knowledge graph according to the relationship between the multiple article entities and the multiple article entities. In the case where the article knowledge graph also includes article heat, the article recommendation system also obtains the article heat of each article according to the interactive behavior data corresponding to each of the multiple articles, and the article recommendation system obtains the relationship between multiple article entities and the multiple article entities according to the multiple articles, the article tags of the multiple articles and the article heat of the multiple articles, and then generates the article knowledge graph according to the relationship between the multiple article entities and the multiple article entities. For example, the article recommendation system connects the multiple article entities according to the relationship between the multiple article entities to obtain the article knowledge graph. The interactive behavior data corresponding to each article is the data generated when the user interacts with each article.

[0100] In an optional embodiment, each of the multiple articles corresponds to multiple pieces of interactive behavior data, and the multiple pieces of interactive behavior data are data generated by at least one user interacting with each article. For each piece of interactive behavior data corresponding to each of the multiple articles: the article recommendation system obtains the time decay score of each piece of interactive behavior data according to the interaction time corresponding to each piece of interactive behavior data, and the article recommendation system obtains the score weight of each piece of interactive behavior data according to the interaction type corresponding to each piece of interactive behavior data. The article recommendation system obtains the article popularity of each article according to the time decay score of the multiple pieces of interactive behavior data corresponding to each of the multiple articles and the score weight of the multiple pieces of interactive behavior data. The score weight of each piece of interactive behavior data is used to characterize the importance of the interactive behavior that generates each piece of interactive behavior data. For example, the article recommendation system determines the time decay score of each interactive behavior data according to the time difference between the interaction time corresponding to each interactive behavior data and the current time and the time decay score formula, obtains the score weight of each interactive behavior data according to the interaction type corresponding to each interactive behavior data, and determines the article popularity of each article according to the time decay score of the interactive behavior data corresponding to each article in the multiple articles, the score weight of each interactive behavior data corresponding to each article, and the popularity formula. The popularity formula is: H represents the popularity of the article, m represents the total number of interactive behavior data corresponding to the article, Si represents the time decay score of the i-th interactive behavior data of the article, Wi represents the score weight of the i-th interactive behavior data, i and m are integers greater than or equal to 1, i≤m.

[0101] In an optional embodiment, the article recommendation system obtains at least one article tag of each of the multiple articles through a tag acquisition model, and the tag acquisition model is used to obtain article tags according to the article content. For example, for each of the multiple articles, the article recommendation system inputs each article into the tag acquisition model, and the tag acquisition model obtains (e.g., calculates) the article tag of each article according to the article content of each article, and outputs the article tag of each article, and the article recommendation system obtains the article tag output by the tag acquisition model.

[0102] In an optional embodiment, the article tags in the above-mentioned article knowledge graph are numerical tags. Before the article recommendation system generates the article knowledge graph based on the above-mentioned multiple article entities, the article recommendation system numericalizes the article tags in the multiple article entities respectively, that is, expresses the article tags in the article knowledge graph with numerical values. For example, the article recommendation system numericalizes each article tag in the multiple article entities through a numerical model. In one embodiment, for each article tag in the multiple article entities, the article recommendation system inputs each article tag into the numerical model, causes the numerical model to numericalize each article tag, and outputs the numericalized article tag (for example, the numericalized article tag is called a numerical tag), and the article recommendation system obtains the numerical tag output by the numerical model.

[0103] The numerical model may be a trained model, for example, the numerical model may be a bidirectional encoder representation from transformers (BERT) model. Figure 4 is a schematic diagram of digitizing article tags provided in an embodiment of the present application, such as Figure 4As shown, the article tags include catalyst separation, filtration, tubular reactor, toluene, author 1, author 2, and chemistry. The tag types of the article tags "catalyst separation", "filtration", "tubular reactor" and "toluene" are all keywords, the tag types of the article tags "author 1" and "author 2" are all authors, and the tag type of the article tag "chemistry" is article type. The article recommendation system inputs the multiple article tags into the numerical model according to the tag types of the multiple article tags; the numerical model digitizes the article tags "catalyst separation", "filtration", "tubular reactor" and "toluene" to obtain corresponding numerical tags "0.074", "-0.162", "-0.417" and "-0.215" respectively; the numerical model digitizes the article tags "author 1" and "author 2" to obtain corresponding numerical tags "0.019" and "-0.039" respectively; and the numerical model digitizes the article tag "chemistry" to obtain the corresponding numerical tag "0.093".

[0104] S103. Obtain at least one target article based on the article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article matches the article tag corresponding to the first user.

[0105] In one embodiment, the types of article entities in the article knowledge graph include articles and article tags. The article recommendation system searches the article knowledge graph for at least one article tag that matches the article tag corresponding to the first user. For example, the article recommendation system searches the article knowledge graph for at least one article tag that is the same as the article tag corresponding to the first user. The article recommendation system determines at least one article in the article knowledge graph that has a connection relationship with the at least one article tag. The at least one article is the target article. The article recommendation system obtains the at least one target article. As an example, the article tags corresponding to the first user include article tag 1 and article tag 8. The article recommendation system Figure 2 The article knowledge graph shown in the figure searches for article tags that match the article tags corresponding to the first user. The article recommendation system can find article tag 1 and article tag 8. The article recommendation system determines in the article knowledge graph that articles 1 and 2 are connected to article tag 1, and determines that articles 3 and 4 are connected to article tag 8. The article recommendation system determines articles 1 to 4 as target articles, and obtains articles 1 to 4.

[0106] In an optional embodiment, the article entity in the article knowledge graph also includes article popularity. The article recommendation system obtains multiple articles based on the article tag corresponding to the first user and the article knowledge graph, and the article tags of the multiple articles match the article tags corresponding to the first user. The article recommendation system obtains the article popularity of each of the multiple articles based on the article knowledge graph. The article recommendation system obtains at least one target article from the multiple articles based on the article popularity of the multiple articles, wherein the article popularity of at least one target article is greater than a preset popularity. In a specific embodiment, the article recommendation system searches for at least one article tag that matches the article tag corresponding to the first user in the article knowledge graph. For example, the article recommendation system searches for at least one article tag that is the same as the article tag corresponding to the first user in the article knowledge graph. The article recommendation system determines multiple articles that have a connection relationship with the at least one article tag in the article knowledge graph. The article recommendation system obtains the article popularity of each of the multiple articles in the article knowledge graph. For each of the multiple articles, the article recommendation system determines whether the article popularity of each article is greater than a preset popularity. If the article popularity of each article is greater than the preset popularity, the article recommendation system determines that each article is a target article. If the article popularity of each article is not greater than the preset popularity, the article recommendation system determines that each article is not a target article. In this way, the article recommendation system can determine at least one target article from the multiple articles, and the article recommendation system obtains the at least one target article. As an example, the article tags corresponding to the first user include article tag 1 and article tag 8. The article recommendation system is as follows. Figure 3 The article knowledge graph shown in the figure searches for article tags that match the article tags corresponding to the first user. The article recommendation system can find article tags 1 and article tags 8. The article recommendation system determines articles 1 and 2 that are connected to article tag 1 in the article knowledge graph, and determines articles 3 and 4 that are connected to article tag 8. The article recommendation system obtains the article heat of each article in articles 1 to 4 in the article knowledge graph. The article recommendation system can obtain article heat 1 of article 1, article heat 2 of article 2, article heat 3 of article 3, and article heat 4 of article 4. Assuming that article heat 1 and article heat 2 are both greater than the preset heat, and article heat 3 and article heat 4 are both less than the preset heat, the article recommendation system determines article 1 and article 2 as target articles, and the article recommendation system obtains article 1 and article 2.

[0107] S104. Recommend the at least one target article to the first user.

[0108] In an optional embodiment, the article recommendation system sorts the at least one target article, and the article recommendation system recommends the at least one target article to the first user according to the sorting order. In one embodiment, for each target article in the at least one target article, the article recommendation system obtains the estimated score of each target article for the first user based on the interactive behavior data of at least one second user strongly associated with the first user and the each target article and the association degree between the first user and the at least one second user, and the article recommendation system sorts the at least one target article according to the estimated score of the at least one target article for the first user. For example, the article recommendation system sorts the at least one target article according to the estimated score of the at least one target article for the first user using a bubble sort algorithm, a selection sort algorithm, or an insertion sort algorithm. Among them, the estimated score of each target article in the at least one target article for the first user is used to characterize the degree of matching between the article content of each target article and the article content of interest to the first user, and the association degree between the first user and any second user is determined based on the interactive behavior data of the first user and the at least one article and the interactive behavior data of the second user and the at least one article. It can be seen that the at least one article is an article with which both the first user and the second user have interacted.

[0109] In one embodiment, the article recommendation system determines at least one article that has interacted with both the first user and the at least one second user, and the article recommendation system determines the association between the first user and each second user based on the interaction behavior data generated by the interaction between the first user and the at least one article (i.e., the interaction behavior data of the first user corresponding to the at least one article) and the interaction behavior data generated by each second user interacting with the at least one article (i.e., the interaction behavior data of the second user corresponding to the at least one article). For each target article in the at least one target article, the article recommendation system obtains the interaction behavior data between each second user in the at least one second user and each target article, and the article recommendation system obtains the interest score of each second user for each target article based on the interaction behavior data between each second user and each target article. The article recommendation system uses an estimated scoring formula to determine the estimated score of each target article for the first user based on the interest score of the at least one second user for each target article and the association between the first user and the at least one second user. The estimated scoring formula is: sim(u,vj)]. Z represents the estimated score of a target article for the first user, u represents the first user, vj represents the jth second user among at least one second user strongly associated with the first user, sim(u,vj) represents the association between the first user and the jth second user, Rj represents the interest score of the jth second user in the target article, k represents the number of the at least one second user, k and j are both integers greater than or equal to 1, j≤k. In an embodiment of the present application, the second user strongly associated with the first user satisfies: the first user has interacted with the same article and the difference between the first user's interest score in the same article and the second user's interest score in the same article is less than a preset difference.

[0110] In an optional embodiment, the article recommendation system obtains the interest score of each second user for each target article based on the interaction behavior data between each second user and each target article, including: the article recommendation system determines the weighted score of all interaction behavior data between the second user and the target article, and the article recommendation system determines the weighted score of all interaction behavior data between the second user and the target article as the interest score of the second user for the target article. In one embodiment, the multiple interaction behavior data between the second user and the target article include browsing data, like data, collection data, sharing data and scoring data, the article recommendation system obtains the score weight of each interaction behavior data in the multiple interaction behavior data, and the article recommendation system obtains the score corresponding to each interaction behavior data in the multiple interaction behavior data, and the article recommendation system performs weighted summation on the multiple interaction behavior data according to the score weight of the multiple interaction behavior data to obtain the weighted score of all interaction behavior data between the second user and the target article. Among them, the implementation process of the article recommendation system obtaining the score weight of each interaction behavior data can refer to the relevant description in S101. The article recommendation system obtains the score corresponding to each piece of the multiple pieces of interaction behavior data, including: for each piece of interaction behavior data in the browsing data, like data, collection data and sharing data, the article recommendation system determines that the score corresponding to each piece of interaction behavior data is a default score (for example, 1), and the article recommendation system determines that the score corresponding to the scoring data is the score of the scoring data.

[0111] The embodiment of the present application takes the example of an article recommendation system determining a weighted score of all interaction behavior data between each second user and each target article as the second user's interest score for the target article. In an optional embodiment, when the interaction behavior data between the second user and the target article includes scoring data, the article recommendation system determines the score of the scoring data as the second user's interest score for the target article. Other methods may also be used to determine the second user's interest score for the target article.

[0112] In an optional embodiment, the article recommendation system determines the association degree between the first user and each second user, including: the article recommendation system determines at least one article with which both the first user and the second user have interacted, the article recommendation system determines a weighted score of all interaction behavior data between the first user and each article in the at least one article to obtain at least one first weighted score corresponding one-to-one to the at least one article, and the article system determines a weighted score of all interaction behavior data between the second user and each article in the at least one article to obtain at least one second weighted score corresponding one-to-one to the at least one article, the article recommendation system determines a first vector composed of the at least one first weighted score, and the article recommendation system determines a second vector composed of the at least one second weighted score, the article recommendation system determines the distance between the first vector and the second vector, for example, the article recommendation system uses the Euclidean distance formula to determine the distance between the first vector and the second vector, and the article recommendation system determines the distance between the first vector and the second vector as the association degree between the first user and the second user.

[0113] In the embodiment of the present application, an article recommendation system determines the estimated score of each target article for the first user based on the interest score of at least one second user for each target article and the association between the first user and the at least one second user as an example. In other embodiments, the article recommendation system determines the estimated score of each target article for the first user based on the article tag corresponding to the first user and the article tag of each target article. For example, the article recommendation system calculates the estimated score of each target article for the first user using an estimated score model based on the article tag corresponding to the first user and the article tag of each target article. In a specific embodiment, the article recommendation system concatenates the article label corresponding to the first user and the article label of each target article to form a concatenated label of the first user and each target article. For example, the article label corresponding to the first user includes article label 1 and article label 8, and the article label of the target article includes article label A and article label B. The article recommendation system concatenates the article label corresponding to the first user and the article label of the target article to form a concatenated label of "article label 1 article label 8 article label A article label B". The article recommendation system inputs the concatenated label of the first user and each target article into an estimated scoring model, so that the estimated scoring model calculates the estimated score of each target article for the first user based on the concatenated label of the first user and each target article. Optionally, the estimated rating model can be a lightweight gradient boosting machine learning (Light Gradient Boosting Machine, LightGBM) model. The LightGBM model is a fast, distributed, high-performance gradient boosting algorithm model based on the decision tree algorithm. The LightGBM model adopts the decision tree algorithm of the histogram algorithm and the histogram difference acceleration algorithm, which can reduce the computational cost of the article recommendation system when sorting the spliced ​​labels. The LightGBM model adopts a leaf growth algorithm with depth restrictions and a mutually exclusive feature bundling algorithm, which can improve the accuracy of the article recommendation system when sorting the spliced ​​labels. The LightGBM model adopts a unilateral gradient sampling algorithm, which can improve the speed of the article recommendation system when sorting the spliced ​​labels.

[0114] The embodiment of the present application takes the example of an article recommendation system recommending articles to users based on an article knowledge graph as an example. In other embodiments, the article recommendation system recommends articles to users based on an article index library. The article index library includes a label vector for characterizing article labels of each article. The article recommendation system obtains at least one target article whose article label matches the article label corresponding to the first user based on the article label corresponding to the first user and the article index library, and recommends the at least one target article to the first user. The article index library is generated by the article recommendation system based on multiple articles managed by the article recommendation system and the article labels of the multiple articles. In one embodiment, the label vector in the article index library is a numerical vector. The article recommendation system digitizes the article label of each of the multiple articles to obtain the numerical label of each article. The article recommendation system generates the numerical vector of the article label of each article based on the numerical label of each article. In a specific embodiment, continuing with Figure 4 For example, suppose Figure 4The article labels “catalyst separation”, “filtration”, “tubular reactor”, “toluene”, “author 1”, “author 2” and “chemistry” in the article are article labels of the same article (for example, article A). The article recommendation system obtains the numerical labels “0.074”, “-0.162”, “-0.417” and “-0.215” corresponding to the article labels “catalyst separation”, “filtration”, “tubular reactor” and “toluene” output by the numerical model, the numerical labels “0.019” and “-0.039” corresponding to the article labels “author 1” and “author 2”, and the numerical label “0.093” corresponding to the article label “chemistry”. The article recommendation system generates the numerical labels “0.074”, “-0.162”, “-0.417” and “-0.215” corresponding to the article labels “catalyst separation”, “filtration”, “tubular reactor” and “toluene” according to the numerical labels “0.074”, “-0.162”, “-0.417” and “-0.215” output by the numerical model. The article recommendation system generates a numerical vector [0.074, -0.162, -0.417, -0.215...] according to the numerical labels "0.019" and "-0.039" corresponding to the article labels "Author 1" and "Author 2", and the article recommendation system generates a numerical vector [0.019, -0.039...] according to the numerical label "0.093" corresponding to the article label "Chemistry" 3...], the article recommendation system concatenates the numerical vectors [0.074, -0.162, -0.417, -0.215...], [0.019, -0.039...] and [0.093...] to obtain the numerical vector [0.074, -0.162, -0.417, -0.215, ...0.019, -0.039...0.093...] of the article (ie, article A). The article recommendation system generates the article index library according to the numerical vectors of the multiple articles. For example, the article recommendation system constructs the article index library according to the numerical vectors of the multiple articles through a hierarchical navigable small world (HNSW) algorithm. Optionally, the article index library is a tree-type index library, which includes multiple index layers, each of the multiple index layers includes a numeralized vector of at least one article among the multiple articles, the bottom index layer includes a numeralized vector of each article among the multiple articles, and the numeralized vectors included in the multiple index layers decrease in sequence from the bottom layer to the top layer. For example, the first layer includes a numeralized vector of each article among the multiple articles, the number of numeralized vectors included in the second layer is half of the number of numeralized vectors included in the first layer, the number of numeralized vectors included in the third layer is half of the number of numeralized vectors included in the second layer, and so on.

[0115] In summary, the article recommendation method provided in the embodiment of the present application is executed by an article recommendation system. The article recommendation system obtains the article tag and article knowledge graph corresponding to the first user, the article tag is used to characterize the article content that the first user is interested in, the article recommendation system determines at least one target article based on the article tag corresponding to the first user and the knowledge graph, the article tag of the at least one target article matches the article tag corresponding to the first user, and the article recommendation system recommends the at least one target article to the first user. Therefore, the present application recommends articles to the first user based on the article content that the first user is interested in in combination with the article knowledge graph, and the recommended articles are highly targeted.

[0116] The following is a system embodiment of the present application, which can be used to execute the method embodiment of the present application. For details not disclosed in the system embodiment of the present application, please refer to the method embodiment of the present application.

[0117] Please refer to Figure 5 , which shows a schematic diagram of an article recommendation system 500 provided in an embodiment of the present application. The article recommendation system 500 is used to perform Figure 1 The method provided by the illustrated embodiment: The article recommendation system 500 includes a first acquisition module 501 , a second acquisition module 502 , a third acquisition module 503 and a recommendation module 504 .

[0118] A first acquisition module 501 is used to acquire an article tag corresponding to a first user, where the article tag corresponding to the first user is used to represent the content of the article that the first user is interested in;

[0119] The second acquisition module 502 is used to acquire an article knowledge graph, the article knowledge graph includes article entities and connection relationships between article entities, the types of article entities in the article knowledge graph include articles and article tags, and the article tags in the article knowledge graph are used to represent the content of the article;

[0120] A third acquisition module 503 is used to acquire at least one target article according to the article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article matches the article tag corresponding to the first user;

[0121] The recommendation module 504 is configured to recommend the at least one target article to the first user.

[0122] Optionally, the type of the article entity in the article knowledge graph further includes article popularity. The third acquisition module 503 is used to:

[0123] Acquire multiple articles according to the article tag corresponding to the first user and the knowledge graph of the article, where the article tags of the multiple articles match the article tag corresponding to the first user;

[0124] Obtain the article popularity of each of the multiple articles according to the article knowledge graph;

[0125] According to the article popularity of the multiple articles, the at least one target article is obtained from the multiple articles, and the article popularity of the at least one target article is greater than a preset popularity.

[0126] Optionally, the second acquisition module 502 is used to:

[0127] Get multiple articles;

[0128] According to the article content of each article in the plurality of articles, obtaining at least one article tag for characterizing the article content of each article;

[0129] Acquire multiple article entities and relationships between the multiple article entities according to the multiple articles and article tags of the multiple articles;

[0130] The article knowledge graph is generated according to the multiple article entities and the relationships between the multiple article entities.

[0131] Optionally, the second acquisition module 502 is used to:

[0132] Obtaining the article popularity of each article according to the interactive behavior data corresponding to each article in the plurality of articles, wherein the interactive behavior data corresponding to each article is data generated by the user's interactive behavior with each article, and the interactive behavior data includes at least one of browsing data, like data, collection data, sharing data, and scoring data;

[0133] According to the multiple articles, the article tags of the multiple articles and the article popularity of the multiple articles, the multiple article entities and the relationships between the multiple article entities are obtained.

[0134] Optionally, each of the plurality of articles corresponds to a plurality of pieces of interaction behavior data, and the interaction type corresponding to each piece of interaction behavior data in the plurality of pieces of interaction behavior data includes any one of browsing, liking, collecting, analyzing, and scoring; the second acquisition module 502 is used to:

[0135] For each interactive behavior data corresponding to each article in the multiple articles: obtain a time decay score for each interactive behavior data according to the interaction time corresponding to the each interactive behavior data; and obtain a score weight for each interactive behavior data according to the interaction type corresponding to the each interactive behavior data, and the score weight for each interactive behavior data is used to characterize the importance of the interactive behavior that generates the each interactive behavior data; obtain the article popularity of each article according to the time decay scores of the multiple interactive behavior data corresponding to each article in the multiple articles and the score weights of the multiple interactive behavior data.

[0136] Optionally, the first acquisition module 501 is used to acquire a user portrait of the first user, where the user portrait includes an article tag corresponding to the first user.

[0137] Optionally, the first acquisition module 501 is used to:

[0138] Acquire interactive behavior data of the first user, where the interactive behavior data of the first user is data generated by the first user's interactive behavior with the article, and the interactive behavior data of the first user includes at least one of browsing data, like data, collection data, sharing data, and scoring data;

[0139] Acquire article content that the first user is interested in based on the interaction behavior data of the first user;

[0140] Obtaining article tags corresponding to the first user based on the content of the article that the first user is interested in;

[0141] Generate a user profile of the first user according to the article tag corresponding to the first user.

[0142] Optionally, a recommendation module 504 is provided for:

[0143] For each target article in the at least one target article, obtaining an estimated score of each target article for the first user according to the interaction behavior data of at least one second user strongly associated with the first user and each target article and the association degree between the first user and the at least one second user, wherein the estimated score of each target article for the first user is used to characterize the degree of matching between the article content of each target article and the article content that the first user is interested in, and the association degree between the first user and any of the second users is determined according to the interaction behavior data of the first user and at least one article and the interaction behavior data of the second user and the at least one article;

[0144] sorting the at least one target article according to the estimated score of the at least one target article for the first user;

[0145] The at least one target article is recommended to the first user according to the sorted order.

[0146] To summarize, the technical solution provided by the embodiment of the present application is that the article recommendation system obtains the article tag and article knowledge graph corresponding to the first user, the article tag is used to characterize the article content that the first user is interested in, and the article recommendation system determines at least one target article based on the article tag corresponding to the first user and the article knowledge graph, the article tag of the at least one target article matches the article tag corresponding to the first user, and the article recommendation system recommends the at least one target article to the first user. Therefore, the present application recommends articles to the first user based on the article content that the first user is interested in in combination with the article knowledge graph, and the recommended articles are more targeted.

[0147] An embodiment of the present application provides an article recommendation system, including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program stored in the memory so that the article recommendation system performs the article recommendation method provided in the above embodiment.

[0148] As an example, see Figure 6 , which shows a schematic diagram of another article recommendation system 600 provided in an embodiment of the present application. The article recommendation system 600 is a functional component deployed in a terminal device, which can be a smart phone, a tablet computer, a laptop computer or a desktop computer, etc. The article recommendation system 600 is used to perform Figure 1 The method provided by the illustrated embodiment.

[0149] Typically, the article recommendation system 600 includes: a processor 601 and a memory 602 .

[0150] The processor 601 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 601 may be implemented in at least one hardware form of digital signal processing (DSP), field programmable gate array (FPGA), and programmable logic array (PLA). The processor 601 may include but is not limited to a central processing unit (CPU). In some embodiments, the processor 601 may be integrated with a graphics processing unit (GPU), which is responsible for rendering and drawing the content to be displayed on the display screen. The processor 601 may also include an artificial intelligence (AI) processor to handle computing operations related to machine learning.

[0151] The memory 602 may include one or more computer-readable storage media, which may be non-transitory. The memory 602 may also include a high-speed random access memory and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 602 is used to store at least one instruction, which is used to be executed by the processor 601 to implement the article recommendation method provided in the embodiment of the present application.

[0152] In some embodiments, the article recommendation system 600 may also optionally include: a peripheral device interface 603 and at least one peripheral device. The processor 601, the memory 602 and the peripheral device interface 603 may be connected via a bus or a signal line. Each peripheral device may be connected to the peripheral device interface 603 via a bus, a signal line or a circuit board. The peripheral device may include: at least one of a radio frequency circuit 604, a display screen 605, a camera 606, an audio circuit 607, a positioning component 608 and a power supply 609.

[0153] The peripheral device interface 603 may be used to connect at least one peripheral device related to input / output (I / O) to the processor 601 and the memory 602. In some embodiments, the processor 601, the memory 602, and the peripheral device interface 603 are integrated on the same chip or circuit board; in some embodiments, any one or two of the processor 601, the memory 602, and the peripheral device interface 603 may be implemented on a separate chip or circuit board, which is not limited in this embodiment.

[0154] The radio frequency circuit 604 is used to receive and transmit radio frequency (RF) signals, also known as electromagnetic signals. The radio frequency circuit 604 communicates with communication networks and other communication devices through electromagnetic signals. The radio frequency circuit 604 converts electrical signals into electromagnetic signals for transmission, or converts received electromagnetic signals into electrical signals. Optionally, the radio frequency circuit 604 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, and the like. The radio frequency circuit 504 can communicate with other devices through at least one wireless communication protocol. The wireless communication protocol includes, but is not limited to: the World Wide Web, a metropolitan area network, an intranet, various generations of mobile communication networks (2G, 3G, 4G and 5G), and a wireless local area network, which is not limited in the embodiments of the present application.

[0155] The display screen 605 is used to display a user interface (UI). The UI may include graphics, text, icons, videos, and any combination thereof. When the display screen 605 is a touch display screen, the display screen 605 also has the ability to collect touch signals on the surface or above the surface of the display screen 605. The touch signal can be input to the processor 601 as a control signal for processing. At this time, the display screen 605 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen 605 can be a flexible display screen. Even, the display screen 605 can also be set to a non-rectangular irregular shape, that is, a special-shaped screen. The display screen 605 can be a liquid crystal display (LCD), an organic light-emitting diode (OLED) display screen, etc.

[0156] The camera assembly 606 is used to capture images or videos. Optionally, the camera assembly 606 includes a front camera and a rear camera. Typically, the front camera is arranged on the front panel of the terminal device, and the rear camera is arranged on the back of the terminal device. In some embodiments, there are at least two rear cameras, which are any one of a main camera, a depth of field camera, a wide-angle camera, and a telephoto camera, so as to realize the fusion of the main camera and the depth of field camera to realize the background blur function, the fusion of the main camera and the wide-angle camera to realize panoramic shooting and virtual reality (VR) shooting function or other fusion shooting functions. In some embodiments, the camera assembly 606 may also include a flash. The flash can be a monochrome temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cold light flash, which can be used for light compensation at different color temperatures.

[0157] The audio circuit 607 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, and convert the sound waves into electrical signals and input them into the processor 601 for processing, or input them into the radio frequency circuit 604 to achieve voice communication. For the purpose of stereo acquisition or noise reduction, there can be multiple microphones, which are respectively arranged in different parts of the terminal device. The microphone can also be an array microphone or an omnidirectional acquisition microphone. The speaker is used to convert the electrical signal from the processor 601 or the radio frequency circuit 604 into sound waves. The speaker can be a traditional film speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, it can not only convert the electrical signal into sound waves audible to humans, but also convert the electrical signal into sound waves inaudible to humans for purposes such as ranging.

[0158] The positioning component 608 is used to locate the geographical location of the terminal device to implement navigation or location based service (LBS). The positioning component 608 can be a positioning component based on the global positioning system (GPS), Beidou system or Galileo system.

[0159] The power supply 609 is used to supply power to various components in the terminal device. The power supply 609 can be an alternating current, a direct current, a disposable battery, or a rechargeable battery. When the power supply 609 includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged through a wired line, and a wireless rechargeable battery is a battery that is charged through a wireless coil.

[0160] In some embodiments, the article recommendation system 600 further includes one or more sensors 610 , including but not limited to: a fingerprint sensor 611 , an optical sensor 612 , a proximity sensor 613 , a pressure sensor 614 , an acceleration sensor 615 , and a gyroscope sensor 616 .

[0161] The fingerprint sensor 611 is used to collect the user's fingerprint, and the processor 601 identifies the user's identity based on the fingerprint collected by the fingerprint sensor 611, or the fingerprint sensor 611 identifies the user's identity based on the collected fingerprint. When the user's identity is identified as a trusted identity, the processor 601 authorizes the user to perform relevant sensitive operations, which include unlocking the screen, viewing encrypted information, downloading software, paying, and changing settings. The fingerprint sensor 611 can be set on the front, back, or side of the terminal device. When a physical button or manufacturer logo is set on the terminal device, the fingerprint sensor 611 can be integrated with the physical button or manufacturer logo.

[0162] The optical sensor 612 is used to collect the ambient light intensity. In one embodiment, the processor 601 can control the display brightness of the display screen 605 according to the ambient light intensity collected by the optical sensor 612. Specifically, when the ambient light intensity is high, the display brightness of the display screen 605 is increased; when the ambient light intensity is low, the display brightness of the display screen 605 is reduced. In another embodiment, the processor 601 can also dynamically adjust the shooting parameters of the camera assembly 606 according to the ambient light intensity collected by the optical sensor 612.

[0163] The proximity sensor 613, also known as a distance sensor, is usually disposed on the front panel of the display screen 605 of the terminal device. The proximity sensor 613 is used to collect the distance between the user and the display screen 605. In one embodiment, when the proximity sensor 613 detects that the distance between the user and the display screen 605 is gradually decreasing, the processor 601 controls the display screen 605 to switch from the screen-on state to the screen-off state; when the proximity sensor 613 detects that the distance between the user and the display screen 605 is gradually increasing, the processor 601 controls the display screen 605 to switch from the screen-off state to the screen-on state.

[0164] The pressure sensor 614 can be set in the side frame of the terminal device and / or the lower layer of the touch display screen 605. When the pressure sensor 614 is set in the side frame of the terminal device, it can detect the user's grip signal of the terminal device, and the processor 601 performs left and right hand recognition or quick operation according to the grip signal collected by the pressure sensor 614. When the pressure sensor 614 is set in the lower layer of the touch display screen 605, the processor 601 controls the operability controls on the UI interface according to the user's pressure operation on the touch display screen 605. The operability controls include at least one of a button control, a scroll bar control, an icon control, and a menu control.

[0165] The acceleration sensor 615 can detect the magnitude of acceleration on the three coordinate axes of the coordinate system established by the terminal device. For example, the acceleration sensor 615 can be used to detect the components of gravity acceleration on the three coordinate axes. The processor 601 can control the touch display screen 605 to display the user interface in a horizontal view or a vertical view according to the gravity acceleration signal collected by the acceleration sensor 615. The acceleration sensor 615 can also be used for collecting game or user motion data.

[0166] The gyroscope sensor 616 can detect the body direction and rotation angle of the terminal device, and the gyroscope sensor 616 can cooperate with the acceleration sensor 615 to collect the user's 3D actions on the terminal device. The processor 601 can implement the following functions based on the data collected by the gyroscope sensor 612: motion sensing (such as changing the UI according to the user's tilt operation), image stabilization during shooting, game control, and inertial navigation.

[0167] Those skilled in the art will understand that Figure 6 The structure shown in the figure does not constitute a limitation on the article recommendation system 600. The article recommendation system 600 may include more or fewer components than those shown in the figure, or combine some components, or adopt a different component arrangement.

[0168] An embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed (for example, when executed by an article recommendation system, one or more processors, etc.), all or part of the steps of the article recommendation method provided in the above embodiment are implemented.

[0169] An embodiment of the present application provides a computer program product, which includes a program or code. When the program or code is executed (for example, when executed by an article recommendation system, one or more processors, etc.), all or part of the steps of the article recommendation method provided in the above embodiment are implemented.

[0170] It should be understood that the term "at least one" in this application refers to one or more, and "multiple" refers to two or more. The term "and / or" in this application is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, in order to facilitate clear description, in this application, the words "first", "second", "third" and the like are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art will understand that the words "first", "second", "third" and the like do not limit the quantity and execution order.

[0171] The method embodiments and system embodiments provided in the embodiments of the present application can refer to each other, and the embodiments of the present application are not limited thereto. The sequence of operations of the method embodiments provided in the embodiments of the present application can be appropriately adjusted, and the operations can also be increased or decreased in response to the situation. Any technician familiar with the technical field can easily think of the method of change within the technical scope disclosed in the present application, and it should be included in the protection scope of the present application, so it will not be repeated.

[0172] In the corresponding embodiments provided in the present application, it should be understood that the disclosed system and the like can be implemented by other configuration methods. For example, the system embodiments described above are only illustrative, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, for example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0173] The modules described as separate components may or may not be physically separate, and the components described as modules may or may not be physical modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0174] The above is only an exemplary embodiment of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. An article recommendation method, characterized in that: The method comprises: Acquire an article tag corresponding to a first user, where the article tag corresponding to the first user is used to represent the content of the article that the first user is interested in; Obtaining an article knowledge graph, wherein the article knowledge graph includes article entities and connection relationships between article entities, the types of article entities in the article knowledge graph include articles and article tags, and the article tags in the article knowledge graph are used to represent article content; Acquire at least one target article according to the article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article matches the article tag corresponding to the first user; The at least one target article is recommended to the first user.

2. The method according to claim 1, characterized in that The type of article entity in the article knowledge graph further includes article popularity, and obtaining at least one target article according to the article tag corresponding to the first user and the article knowledge graph includes: Acquire multiple articles according to the article tags corresponding to the first user and the article knowledge graph, wherein the article tags of the multiple articles match the article tags corresponding to the first user; Obtaining the article popularity of each of the multiple articles according to the article knowledge graph; According to the article popularity of the multiple articles, the at least one target article is obtained from the multiple articles, and the article popularity of the at least one target article is greater than a preset popularity.

3. The method according to claim 1, characterized in that The obtaining of the article knowledge graph includes: Get multiple articles; According to the article content of each of the plurality of articles, obtaining at least one article tag for characterizing the article content of each of the plurality of articles; Acquire multiple article entities and relationships between the multiple article entities according to the multiple articles and article tags of the multiple articles; The article knowledge graph is generated according to the multiple article entities and the relationships between the multiple article entities.

4. The method according to claim 3, characterized in that The obtaining of the article knowledge graph further includes: obtaining the article popularity of each of the plurality of articles according to the interactive behavior data corresponding to each of the plurality of articles, wherein the interactive behavior data corresponding to each of the plurality of articles is data generated by the user's interactive behavior with each of the plurality of articles, and the interactive behavior data includes at least one of browsing data, like data, collection data, sharing data, and scoring data; The acquiring, according to the plurality of articles and the article tags of the plurality of articles, a plurality of article entities and the relationship between the plurality of article entities comprises: According to the multiple articles, the article tags of the multiple articles and the article popularity of the multiple articles, the multiple article entities and the relationships between the multiple article entities are obtained.

5. The method according to claim 4, characterized in that Each of the plurality of articles corresponds to a plurality of pieces of interactive behavior data, and the interaction type corresponding to each piece of interactive behavior data in the plurality of pieces of interactive behavior data includes any one of browsing, liking, collecting, analyzing, and scoring; The obtaining the article popularity of each article according to the interactive behavior data corresponding to each article in the plurality of articles includes: For each interactive behavior data corresponding to each article in the plurality of articles: obtaining a time decay score of each interactive behavior data according to the interaction time corresponding to each interactive behavior data; and obtaining a score weight of each interactive behavior data according to the interaction type corresponding to each interactive behavior data, wherein the score weight of each interactive behavior data is used to characterize the importance of the interactive behavior that generates each interactive behavior data; The article popularity of each of the multiple articles is obtained according to the time decay scores of the multiple pieces of interaction behavior data corresponding to each of the multiple articles and the score weights of the multiple pieces of interaction behavior data.

6. An article recommendation system, characterized in that: The system comprises: A first acquisition module is used to acquire an article tag corresponding to a first user, where the article tag corresponding to the first user is used to represent the article content that the first user is interested in; A second acquisition module is used to acquire an article knowledge graph, wherein the article knowledge graph includes article entities and connection relationships between article entities, and the types of article entities in the article knowledge graph include articles and article tags. The article tags in the article knowledge graph are used to represent the content of the article; A third acquisition module is used to acquire at least one target article according to the article tag corresponding to the first user and the article knowledge graph, wherein the article tag of the at least one target article matches the article tag corresponding to the first user; A recommendation module is used to recommend the at least one target article to the first user.

7. The system according to claim 6, characterized in that The type of article entity in the article knowledge graph also includes article popularity, and the third acquisition module is used to: Acquire multiple articles according to the article tags corresponding to the first user and the article knowledge graph, wherein the article tags of the multiple articles match the article tags corresponding to the first user; Obtaining the article popularity of each of the multiple articles according to the article knowledge graph; According to the article popularity of the multiple articles, the at least one target article is obtained from the multiple articles, and the article popularity of the at least one target article is greater than a preset popularity.

8. The system according to claim 6, characterized in that The second acquisition module is used to: Get multiple articles; According to the article content of each of the plurality of articles, obtaining at least one article tag for characterizing the article content of each of the plurality of articles; Acquire multiple article entities and relationships between the multiple article entities according to the multiple articles and article tags of the multiple articles; The article knowledge graph is generated according to the multiple article entities and the relationships between the multiple article entities.

9. The system according to claim 8, characterized in that The second acquisition module is used to: Obtaining the article popularity of each article according to the interactive behavior data corresponding to each of the plurality of articles, wherein the interactive behavior data corresponding to each article is data generated by the user's interactive behavior with each of the articles, and the interactive behavior data includes at least one of browsing data, like data, collection data, sharing data, and scoring data; According to the multiple articles, the article tags of the multiple articles and the article popularity of the multiple articles, the multiple article entities and the relationships between the multiple article entities are obtained.

10. The system according to claim 9, characterized in that Each of the plurality of articles corresponds to a plurality of pieces of interactive behavior data, and the interaction type corresponding to each piece of interactive behavior data in the plurality of pieces of interactive behavior data includes any one of browsing, liking, collecting, analyzing and scoring; the second acquisition module is used to: For each piece of interaction behavior data corresponding to each of the plurality of articles: obtaining a time decay score of each piece of interaction behavior data according to an interaction time corresponding to each piece of interaction behavior data; and, obtaining a score weight of each piece of interaction behavior data according to the interaction type corresponding to each piece of interaction behavior data, wherein the score weight of each piece of interaction behavior data is used to characterize the importance of the interaction behavior that generates each piece of interaction behavior data; The article popularity of each of the multiple articles is obtained according to the time decay scores of the multiple pieces of interaction behavior data corresponding to each of the multiple articles and the score weights of the multiple pieces of interaction behavior data.

Images