A content recommendation method and related apparatus

Abstract

Embodiments of the present application disclose a content recommendation method and related device, at least related to machine learning in artificial intelligence, by obtaining content viewing history data of a user in a content platform, time consumption proportion data of the user is obtained, which can objectively and quantitatively reflect the user's preference degree for different contents in the content platform, and become a user feature reflecting the user's preference. A plurality of user categories are obtained by classifying the user based on the time consumption proportion data, and the users in the same user category have similar preferences for the content. When pushing target content to a target user, the target user category to which the target user belongs is determined based on the foregoing user classification. Since the content preferences of the users belonging to the target user category in the content platform are relatively similar, the target content recommended to the target user can be determined according to the users belonging to the target user category, avoiding various problems of user tags and content tags in recommendation, and ensuring the accuracy of content recommendation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, and in particular, to a content recommendation method and related device. BACKGROUND

[0002] In order to improve the user experience of the content platform, the content platform will recommend the content in the content platform, such as videos, live rooms, etc., to the user, in order to improve the user stickiness of the content platform.

[0003] In the related art, the content recommendation is mainly performed by using the label matching method. First, the label of the user and the label of the content in the platform are determined. The label of the user is determined according to the preference selected by the user when registering, and the label of the content is determined according to the content type involved in the content itself. Then, based on the label similarity between the user and the content, it is determined which content is recommended to the user.

[0004] However, the accuracy of the label itself is actually not high. For example, the user randomly selects the preference when registering, or the user accidentally selects the wrong label, etc., which will result in inaccurate user labels. Moreover, the identification dimension of the label is artificially specified, and as the content type is constantly updated, it is difficult for the label to comprehensively cover all possible content types.

[0005] Based on the above problems of accuracy and comprehensiveness of the label, the effect of recommending the content based on the label similarity is not ideal. SUMMARY

[0006] In order to solve the above technical problems, the present application provides a content recommendation method and related device, which are used to solve the problem of ideal effect of recommending the content to the user based on the label similarity.

[0007] The embodiments of the present application disclose the following technical solutions:

[0008] On the one hand, the present application provides a content recommendation method, which comprises:

[0009] obtaining content viewing history data of a user in a content platform;

[0010] determining time-consuming proportion data of the user according to the viewing time length of the identified content in the content viewing history data, the time-consuming proportion data being used to identify the time-consuming distribution of the user in viewing different contents in the content platform through the content platform;

[0011] classifying the user in the content platform based on the time-consuming proportion data, to obtain a plurality of user categories;

[0012] determining a target user in the content platform, the target user belonging to a target user category in the plurality of user categories;

[0013] determine target content recommended to the target user according to users belonging to the target user category in the content platform.

[0014] In another aspect, the present application provides a content recommendation device, the device comprising: an acquisition unit, a first determination unit, a classification unit, a second determination unit and a third determination unit;

[0015] The acquisition unit is configured to acquire content viewing history data of users in a content platform.

[0016] The first determination unit is configured to determine time consumption proportion data of the users according to viewing time lengths of identified content in the content viewing history data, the time consumption proportion data being used to identify time consumption distribution of different content viewed by the users in the content platform.

[0017] The classification unit is configured to classify the users in the content platform based on the time consumption proportion data to obtain a plurality of user categories.

[0018] The second determination unit is configured to determine a target user in the content platform, the target user belonging to a target user category in the plurality of user categories.

[0019] The third determination unit is configured to determine target content recommended to the target user according to users belonging to the target user category in the content platform.

[0020] In another aspect, the present application provides a computer device, the device comprising a processor and a memory:

[0021] The memory is configured to store program code and transmit the program code to the processor.

[0022] The processor is configured to execute the method according to the instructions in the program code.

[0023] In another aspect, an embodiment of the present application provides a computer readable storage medium for storing a computer program, the computer program being used to execute the method according to the above aspect.

[0024] In another aspect, an embodiment of the present application provides a computer program product or a computer program, the computer program product or the computer program comprising computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute the method according to the above aspect.

[0025] It can be seen from the technical solution that, by obtaining the content viewing history data of the user in the content platform, the time length of the user viewing different contents on the content platform can be determined, and the time consumption proportion data of the user is obtained according to the content viewing history data, which is used to identify the time consumption distribution of the user viewing different contents, that is, the proportion of the time of the user on the content platform consumed by different contents. Since the consumption of time can objectively and quantitatively reflect the preference degree of the user for different contents in the content platform, the time consumption proportion data can be used to represent the content preference of the user from the time dimension, which can replace the user label in the related art as a user feature reflecting the user preference. The user is classified by the time consumption proportion data to obtain a plurality of user categories, and the users in the same user category obtained in this way have similar preferences for contents. When recommending target contents for a target user, the target user category to which the target user belongs can be determined based on the foregoing user classification. Since the users in the content platform belonging to the target user category have similar content preferences, the users in the target user category can be used as the basis for content recommendation, and this kind of recommendation based on objective data such as time consumption proportion data does not need to consider various user labels and content labels in the related art, thereby breaking the shackles of the original labels, avoiding various problems of labels in recommendation, and ensuring the accuracy of content recommendation. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0027] Figure 1 An application scenario schematic diagram of a content recommendation method provided by an embodiment of the present application;

[0028] Figure 2 A flowchart of a content recommendation method provided by an embodiment of the present application;

[0029] Figure 3 A schematic diagram of an application scenario embodiment of a content recommendation method provided by an embodiment of the present application;

[0030] Figure 4 A schematic diagram of a content recommendation system provided by an embodiment of the present application;

[0031] Figure 5 A schematic diagram of a content recommendation device provided by an embodiment of the present application;

[0032] Figure 6 A structural diagram of a server provided for an embodiment of the present application is shown in FIG. 1.

[0033] Figure 7 A structural diagram of a terminal device provided for an embodiment of the present application is shown in FIG. 2. DETAILED DESCRIPTION

[0034] Embodiments of the present application are described below with reference to the accompanying drawings.

[0035] In view of the problem that the content recommendation effect is not ideal due to the content recommendation manner based on the label similarity of users and content in the related art, the embodiments of the present application provide a content recommendation method and related device, which avoid various problems of the content recommendation manner based on labels and improve the accuracy of content recommendation.

[0036] The content recommendation method provided by the embodiments of the present application is implemented based on artificial intelligence. Artificial intelligence (AI) is a theory, method, technology and application system for using a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive environment, acquire knowledge and use knowledge to obtain optimal results. In other words, artificial intelligence is a comprehensive technology of computer science, which attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making.

[0037] Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, including both hardware-level technologies and software-level technologies. Artificial intelligence basic technologies generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0038] In the embodiments of the present application, the artificial intelligence technologies mainly involved include the above-mentioned machine learning / deep learning, etc.

[0039] The content recommendation method provided in the application can be applied to a content recommendation device with data processing capability, such as a terminal device and a server. The terminal device can be a smart phone, a desktop computer, a notebook computer, a tablet computer, a smart speaker, a smart watch, a smart television, a set-top box, and the like, but is not limited thereto. The server can be a stand-alone physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the application.

[0040] The content recommendation device can have machine learning capability. Machine learning is a multi-disciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, and other disciplines. It is a specialized study of how computers simulate or implement human learning behavior to acquire new knowledge or skills, reorganize existing knowledge structure to continuously improve performance. Machine learning is the core of artificial intelligence and the fundamental approach to making computers intelligent, and its applications are widespread in various fields of artificial intelligence. Machine learning and deep learning usually include artificial neural networks and other technologies.

[0041] In the content recommendation method provided in the embodiments of the application, the artificial intelligence model used mainly involves the application of machine learning, which classifies users in the content platform through clustering methods and the like in machine learning.

[0042] In addition, the content recommendation device provided in the embodiments of the application also has cloud computing capability, which processes the content viewing history data of users in the content platform through cloud computing capability to recommend target content to target users. Big data refers to a collection of data that cannot be captured, managed and processed within a certain time range using conventional software tools, and is a massive, high-growth and diversified information asset that requires new processing modes to have stronger decision-making, insight discovery and process optimization capabilities. With the advent of the cloud era, big data has attracted more and more attention, and big data requires special technology to effectively process large amounts of data over time. The technologies suitable for big data include large-scale parallel processing databases, data mining, distributed file systems, distributed databases, cloud computing platforms, the Internet and scalable storage systems.

[0043] In order to facilitate understanding of the technical solutions of the application, the content recommendation method provided in the embodiments of the application will be introduced below in combination with actual application scenarios, taking a server as a content recommendation device.

[0044] Referring to Figure 1 , Figure 1 The application scenario of a content recommendation method provided in the embodiments of the application is shown in the figure.Figure 1 In the application scenario shown, the content platform includes a server 100, configured to recommend target content to a target user.

[0045] The server 100 obtains content viewing history data of a user in the content platform. The content platform obtains various types of content and distributes the content to the user for viewing different content. The content is information or experience presented to the user by a content creator. For example, the content can be a live streaming room, news, a video, an article, an audio book, etc. In the content platform, the content creator can be a user or a professional content creator. Figure 1 In the scenario shown, the content platform is a live streaming platform, which has 100 users. The server 100 obtains content viewing history data of the 100 users in the live streaming platform. The content viewing history data mainly includes behavior data generated by the user in the live streaming platform, such as viewing duration, sharing and collecting, and recharging data.

[0046] The server 100 determines time consumption ratio data of the user according to the viewing duration identified in the obtained content viewing history data. The time consumption ratio data is used to identify the time consumption distribution of the user in the content platform when viewing different content in the content platform. For example, the user views the content in the content platform through a client, and the server 100 obtains the time consumption ratio data of the user in the content platform according to the viewing duration of the user in the content platform. Figure 1 In the live streaming platform shown, the viewing duration of different live streaming rooms is different, and the time consumption ratio of the viewing duration is also different. Taking two users who only view two live streaming rooms as an example, the time consumption ratio data of user i in live streaming room A is 70%, and the time consumption ratio data of user i in live streaming room B is 30%. The time consumption ratio data of user j in live streaming room A is 30%, and the time consumption ratio data of user j in live streaming room B is 70%.

[0047] The total time consumed by different users in the live streaming platform is different. The time consumption ratio data can objectively reflect the preference of different users for different live streaming rooms under the premise that the total time consumed by different users in the live streaming platform is different, so as not to be affected by the absolute time length, and can quantify the preference of the user for different live streaming rooms, and then can represent the preference of the user based on the time consumption ratio data. For example, in the live streaming platform shown, the time consumption ratio data of user i in live streaming room A is 70%, and the time consumption ratio data of user i in live streaming room B is 30%. The time consumption ratio data of user j in live streaming room A is 30%, and the time consumption ratio data of user j in live streaming room B is 70%. Figure 1 In the scenario shown, based on the time consumption ratio data, it can be known that user i prefers live streaming room A, and user j prefers live streaming room B.

[0048] Since the time consumption ratio data can represent the preference of the user, the server 100 can classify the users based on the time consumption ratio data to obtain multiple user categories. Users belonging to the same user category have similar preferences for the same live streaming room. For example, in the live streaming platform shown, the 100 users are divided into two categories based on the time consumption ratio data, including 60 users including user i, which are divided into type I; and 40 users including user j, which are divided into type II. Figure 1 In the scenario shown, based on the time consumption ratio data, the 100 users are divided into two categories, including 60 users including user i, which are divided into type I; and 40 users including user j, which are divided into type II.

[0049] When recommending target content to a target user, taking user i as an example, based on the aforementioned user classification results, user i belongs to target user category type I. Since users belonging to the same user category have similar preferences for the same live stream, the target content to recommend to user i can be determined based on users belonging to type I. For example, in Figure 1 In the scenario shown, the live streaming platform also has live streaming rooms C and D. Among the users viewing live streaming room C, 90% belong to type I, and among the users viewing live streaming room D, 10% belong to type I. Users of type I may prefer live streaming room C, so live streaming room C can be recommended to user i as the target content.

[0050] As can be seen from the above, recommendation methods based on objective data such as time consumption percentage do not need to consider various user tags and content tags in related technologies, thus breaking free from the constraints of the original tags. This not only avoids various problems caused by tags in recommendations, but also ensures that the content preferred by some users in the target user category is more likely to match the target user's preferences, thus guaranteeing the accuracy of content recommendations.

[0051] The following description, with reference to the accompanying drawings and using a server as the content recommendation device, introduces a content recommendation method provided in an embodiment of this application.

[0052] See Figure 2 , Figure 2 A flowchart illustrating a content recommendation method provided in an embodiment of this application. Figure 2 As shown, the content recommendation method includes the following steps:

[0053] S201: Obtain users' content viewing history data from the content platform.

[0054] A content platform is an internet platform that acquires various types of content and distributes it to users. Content refers to the information or experiences that creators present to users; for example, content can include live streams, news, videos, articles, audiobooks, etc.

[0055] Content on content platforms can be divided into two main categories. The first category consists of content with a fixed type, such as videos, audiobooks, and articles. The second category consists of content with a variable type, such as live streams. For content with a variable type, taking live streams as an example, the same live stream might broadcast gaming content one day and singing content the next, making it impossible for fixed tags to comprehensively cover all the content of that live stream. Furthermore, users might watch a live stream regardless of its content, simply because they like the streamer, making it inaccurate to use a variety of tags to represent user characteristics. Therefore, recommending content to target users based on the similarity between user and content tags is not ideal.

[0056] It can be understood that the content in the content platform will consume the viewing time of the user, the more time the user consumes in one content, the higher the interest degree of the user to the content, thereby the user feature of the user preference can be mined based on the consumption time of the user to the content, the user similar to the recommended user preference is found, and the recommended content is determined according to the similar user preference, so that the user label does not need to be concerned, and the content type in the content platform does not need to be concerned, that is, the content label does not need to be concerned, which is suitable for the content with a non-fixed content type in the content platform, avoids the problem caused by the user label and the content label, and improves the content recommendation effect.

[0057] In actual application, the user can log in the content platform through the terminal device, and view different contents through the content platform. The user generates content viewing history data in the process of using the content platform, and the content viewing history data is mainly some behavior data generated by the user in the content platform, such as viewing time, sharing and collecting, and recharging data.

[0058] The server obtains the content viewing history data of the user in the content platform, and executes the following steps to mine the user feature of the user preference according to the content viewing history data, and recommend the content that the user may be interested in.

[0059] S202: determining the time consumption proportion data of the user according to the viewing time of the identified content in the content viewing history data.

[0060] The content viewing history data includes the viewing time corresponding to the user viewing different contents in the content platform, and the time consumption proportion data of the user can be obtained according to the content viewing history data, wherein the time consumption proportion data is used to identify the time consumption distribution of the user viewing different contents through the content platform in the content platform, and the time consumption proportion data of one user can reflect the allocation of the time consumed by the user on the content platform in different contents. For example, the viewing time corresponding to each content viewed by the user in the content platform is divided by the total time of the user viewing all contents in the content platform, so that the time consumption proportion data of the user viewing each content can be obtained.

[0061] Taking the user i and the user j in the scenario shown in Figure 1 As an example, the viewing time of the user i in the A live room is 7 hours, and the viewing time in the B live room is 3 hours, and the corresponding time consumption proportion data is 70% and 30% respectively. The viewing time of the user j in the A live room is 30 hours, and the viewing time in the B live room is 70 hours, and the corresponding time consumption proportion data is 30% and 70% respectively.

[0062] The time consumption proportion data reflects the proportion of the time of the user on the content platform consumed by different contents. The greater the proportion of the time distribution of the user on a content, the more time the user is consumed by the content, and the more interested the user is in the content. Therefore, the time consumption proportion data can be used instead of the user label in the related art as a user feature reflecting user preferences.

[0063] The present application mines the user preference user feature through the time consumption proportion data, which is equivalent to the "objective label of the user". The determination process of the time consumption proportion data does not require manual intervention, avoiding the problem of inaccurate user labels caused by randomly selecting the user's preferences during user registration or accidental errors in the related art. The time consumption proportion data can objectively quantify the user's preference degree for different contents in the content platform.

[0064] The determination process of the time consumption proportion data is further described below with two examples.

[0065] Example 1:

[0066] In this example, the content in the content platform is taken as a live room as an example. As the number of live rooms in the content platform continues to increase, the user's choice of live rooms to view continues to increase, and the user's content viewing history data will gradually increase. The calculation amount of determining the time consumption proportion data of the user according to the content viewing history data is large. In order to reduce the calculation amount of the server, part of the data in the content viewing history data of the user can be selected to determine the time consumption proportion data of the user.

[0067] The live ratings of different live rooms are different. The live rating is an evaluation level of the live room, for example, the live rating can be one or more combinations of the current number of users viewing, the amount of gifts sent by the user, the number of bullet screens, etc. The live rooms in the content platform whose live ratings reach a preset condition are taken as target live rooms, and the time consumption proportion data of the user is determined according to the viewing time of the target live rooms corresponding to the content viewing history data.

[0068] Among them, the target live room is a live room that is more preferred by more users in the content platform. For example, the live rooms with high heat ranking, such as the top 1000 live rooms, are taken as target live rooms. It can be understood that the live rooms with high heat are live rooms that are viewed more by users in the content platform, and the viewing situation of the users in such live rooms is more consistent with the overall viewing rules of the content platform, so the user viewing data of such live rooms can represent the content platform and be used as a basis for content recommendation in the content platform. Moreover, by mining the user preference user feature according to the target live room instead of all live rooms in the content platform, the accuracy of content recommendation can be ensured while reducing the calculation amount of the server.

[0069] Example 2:

[0070] In this example, considering that the amount of content in the content platform is large, the time consumption proportion data of the user for each content is determined according to the content viewing history data, and the number of the time consumption proportion data is large, which may consume a large amount of computing resources in subsequent user classification.

[0071] Therefore, in order to reduce the consumption of computing resources, the number of time consumption proportion data can be reduced. For example, since the same type of content has similarity, the type of content in the content platform can be determined, the total viewing time of the content belonging to the same type of content is determined according to the content viewing history data, and the time consumption proportion data corresponding to each type of content of the user is determined according to the total viewing time. Thus, the time consumption proportion data of the user for the same type of content is determined according to the content viewing history data, the number of the time consumption proportion data is small, and the subsequent user classification based thereon reduces the amount of calculation. While reducing the amount of calculation, the content viewing history data is also comprehensively considered, and the accuracy of content recommendation is improved.

[0072] S203: classifying the users in the content platform based on the time consumption proportion data to obtain a plurality of user categories.

[0073] As described above, the greater the time consumption distribution proportion of a user for a content, the more interested the user is in the content. The time consumption proportion data can be used as a user feature reflecting the user's preference. If the time consumption distribution proportion of a plurality of users for a content is large, the plurality of users are more interested in the content. Thus, the users in the content platform can be classified according to the time consumption proportion data to obtain a plurality of user categories, and the preferences of the users belonging to the same category are similar.

[0074] It should be noted that, since the total time of different users viewing content in the content platform is different, as shown in the scenario of user i and user j, Figure 1 Although the viewing time of user i in A live room is 7 hours and the viewing time of user j in A live room is 30 hours, user i prefers to view A live room and user j prefers to view B live room, that is, the absolute time length can only reflect that the user has viewed the content, and the relative time length, i.e., the time consumption proportion data, can reflect the degree of user's preference for the content, eliminating the influence of different total time of different users viewing content in the content platform. Thus, compared with classifying the users in the content platform based on the absolute time length, classifying the users based on the time consumption proportion data is more accurate, and the subsequent recommendation of target content to target users is also more accurate.

[0075] The skilled person is not limited to the way of classifying the users based on the time consumption proportion data, and one way of determining the similarity of the users based on the time consumption proportion data and classifying the users based on the similarity of the users is introduced below.

[0076] According to the proportion data in the time-consuming proportion data representing the same content dimension, the user similarity between users in the content platform is determined, and the user similarity can be obtained by the Euclidean distance formula, see formula (1):

[0077] (1)

[0078] wherein, represents the Euclidean distance between user i and user j in the content platform, i and j are integers greater than 0; is an integer greater than 0, indicating the number of contents in the content platform for determining the time-consuming proportion data, such as the number of all contents in the content platform, the number of target live rooms, etc.; represents the time-consuming proportion data of user i viewing content , and represents the time-consuming proportion data of user j viewing content . As shown in the scenario of user i, when content Figure 1 is A live room, when content is B live room, .

[0079] Therefore, the smaller the time-consuming proportion difference of user i and user j for content , the more similar the interest degree of user i and user j for content , further, the smaller the time-consuming proportion difference of user i and user j for the contents in the content platform, the smaller the user similarity of user i and user j, indicating that the preferences of user i and user j are more similar.

[0080] After determining the user similarity, the users in the content platform are classified based on the user similarity, and a plurality of user categories are obtained. The user similarity of users belonging to the same category is close, and the preferences between users are also similar.

[0081] The person skilled in the art does not specifically limit the way of classifying the users in the content platform based on the user similarity, for example, based on the user similarity, the users in the content platform are divided into a plurality of categories by K-means clustering or hierarchical clustering in the clustering method.

[0082] S204: Determine the target user in the content platform.

[0083] ​In actual application, in order to improve the stickiness of users, the content platform can recommend the content that the user is likely to be interested in at any time. For example, after the user logs in the content platform, the content that the user is likely to be interested in is recommended to the user at the position of the homepage. For another example, the content that the user is likely to be interested in is recommended to the user at the position of searching for a certain content. The server obtains the content that the target user is likely to be interested in as the target content when the content platform needs to recommend the content that the target user is interested in.

[0084] It should be noted that S203 and S204 are not necessarily related. For example, S201-S203 can be offline processing, and the server obtains the user categories of multiple users, and then saves each user in the content platform and the user category corresponding to the user. S204 and S205 can be online processing, and when the target user is determined, the target user category to which the target user belongs is obtained from the pre-stored multiple user categories.

[0085] The embodiments of the present application do not specifically limit the manner of obtaining the target user category to which the target user belongs. For example, after the target user logs in the content platform, the server obtains the user identity of the target user, such as platform ID, and then obtains the user category corresponding to the user identity as the target user category according to the corresponding relationship between each user and the user category to which the user belongs which is pre-stored.

[0086] S205: determining the target content recommended to the target user according to the users belonging to the target user category in the content platform.

[0087] After the target user category to which the target user belongs is determined, the users belonging to the target user category in the content platform are similar to the target user in the content preference of the content platform, so that the users belonging to the target user category can be used as the basis for content recommendation, and the target content recommended to the target user is determined. Therefore, it is not necessary to pay attention to the specific content type in the content platform, and it is not necessary to set the corresponding label according to the content type, thereby avoiding the problem that the content label in the related art is difficult to comprehensively cover all possible content types, and improving the effect of content recommendation.

[0088] In addition, the content recommended to the user is determined based on the similarity of the label of the user and the content. Since the label is a fixed concept, the content cannot break out of the label and can only recommend target content to the target user within a limited type range of content. Based on the users belonging to the same target user type as the target user, the content recommended to the target user is not content fixed in a specific label range, but content viewed by a group of users with similar preferences to the target user (users belonging to the target user type). The content may be content that the target user has not seen but is interested in, or content that the target user does not know belongs to their own preferences. Thus, the type range of the recommended content is expanded, the surprise of the user when seeing the recommended content is improved, and the potential interest of the user can be developed.

[0089] For example, user i is the target user, user j is a user belonging to the same target user category as user i, user i likes A content, and user j likes not only A content but also B content. User i has not seen B content, but since user i and user j belong to the same target user category, their preferences for content in the content platform are similar. Therefore, user i may also like B content, but user i does not know that they will like B content, thereby developing the potential interest of user i.

[0090] The present application does not specifically limit the way of determining target content based on users of the same target user category. The following three examples are used as examples for illustration.

[0091] Example 1:

[0092] The category proportion data of the content viewed by the user can reflect the category of the content. For example, there are 10 users who view the content, of which 8 users belong to user category A and 2 users belong to user category B. The proportion of user category A is high, so the content is more likely to be liked by users of user category A. Therefore, the proportion of different content viewed by users belonging to the same target user category as the target user can be used as a basis for determining target content in order to improve the accuracy of the target content recommended to the target user.

[0093] First, the category proportion data of the content in the content platform is obtained. The category proportion data can more accurately reflect the proportion of users belonging to the target user category who view the content. Thus, when determining the target user, the target content determined based on the category proportion data is more in line with the current actual content viewing situation in the content platform. This approach can adapt to the situation where the content in the content platform may have different display information at different times, such as the live streaming room (content) of the live streaming platform (content platform) having different information even at different times of the same day or different days.

[0094] In a possible implementation, the category proportion data is category proportion data when the target user is determined, and is used to identify a proportion of users belonging to a target user category among users viewing the content. For example, there are 10 users viewing a content, and 8 users belong to a target user category A. The category proportion data is 80%.

[0095] It can be understood that in the content platform, the number of users viewing the same content changes in real time, and the category proportion data of the users viewing the content also changes in real time. For example, when a content adds a viewing user, the number corresponding to the user category to which the viewing user belongs is increased by 1, and when a content reduces a viewing user, the number corresponding to the user category to which the viewing user belongs is reduced by 1. For example, there are 10 users viewing a content, and 8 users belong to a user category A. The category proportion data is 80%. If a new user i views the content, and the user i belongs to the user category A, at this time, there are 9 users belonging to the user category A, and the category proportion data changes from 80% to 82%. Similarly, if a user j no longer views the content, and the user j belongs to the user category A, at this time, there are 7 users belonging to the user category A, and the category proportion data changes from 80% to 78%. Thus, the category proportion data of each content in the content platform can be obtained in real time, so as to determine the target content recommended for the target user by using the category proportion data.

[0096] As described above, the server does not recommend the target content to all users in the content platform. For example, when a user does not log in to the content platform, the server does not recommend the target content to the user. Thus, the category proportion data of the content in the content platform can be obtained while the target user is determined, and the category proportion data reflects the category proportion data corresponding to the target user in the content platform, so as to ensure the accuracy of determining the target content based on the category proportion data.

[0097] After obtaining the category proportion data, the content whose category proportion data meets a preset condition can be determined as a to-be-determined content, and the target content recommended to the target user is determined according to the to-be-determined content, so as to further filter the content preferred by the target user by using the category proportion data. For example, the target user category proportion is sorted, and the content corresponding to the target user category proportion is higher, and the possibility that the content is preferred by the user belonging to the target user category is greater. Thus, the content determined by the top n target user category proportions can be determined as the to-be-determined content, and the target content is determined according to the to-be-determined content. A person skilled in the art can set the value of n according to actual conditions. For example, the value of n can be set to be between 20 and 50.

[0098] For example, the number of users viewing A content is 100, and the number of users viewing A content belonging to the target user category is 20, so the category proportion data of A content is 20%. The number of users viewing B content is 10, and the number of users viewing B content belonging to the target user category is 8, so the category proportion data of B content is 80%. The preset condition is to determine the content with the highest ranking in the category proportion data as the pending content, so B content is determined as the pending content, and B content can also be recommended to the target user as the target content. As can be seen, although the number of users corresponding to A content is higher than the number of users corresponding to B content, the target user category proportion of viewing B content is higher, and B content has greater appeal to users belonging to the target user category than A content, and B content may be more in line with the preferences of the target user.

[0099] Thus, the category proportion data can be used as a basis for determining the target content, so as to further improve the accuracy of recommending the target content to the target user.

[0100] Example II:

[0101] In this example, the content in the content platform is taken as a live room. When the target user enters the first live room in the content platform, for example, the target user logs in to the content platform and enters the live room that he often watches. The first user belonging to the target user category with the target user is obtained in the first live room, wherein the first user and the target user not only belong to the same target user category, but also view the same live room in the same time period.

[0102] It should be noted that the content of the same live room broadcasted in different time periods may be different, and there may be a user i and a user j who both like the same live room but like slightly different content. For example, the live room broadcasts content related to "second dimension", and user i and user j both like to view the live room, but the live room plays X animation content in the morning and Y animation content in the afternoon, user i likes to view the live room in the morning, and user j likes to view the live room in the afternoon, and user i and user j like different animations.

[0103] Thus, by determining the first user belonging to the same target user category with the target user and viewing the same live room in the same time period, the possibility that the first user likes the same content as the target user is increased again, and the time consumption proportion data of the first user is used as a recommendation basis to determine the preferences of the first user for different live rooms in the content platform, and the target content recommended to the target user is determined through the preferences of the first user, thereby improving the accuracy of target content recommendation. For example, the preferences of the first user for different live rooms in the content platform are ranked, and the top n live rooms are recommended to the target user for selection.

[0104] Example three:

[0105] The users in the content platform that belong to the target user category and are currently online are obtained, and the target content recommended for the target user is determined according to the number of different contents viewed by the above-mentioned users. For example, there are 31 users in the content platform that belong to the target user category and are currently online, and they are respectively viewing A content and B content, wherein 30 users are viewing A content and 1 user is viewing B content, and thus A content can be recommended to the target user as the target content.

[0106] As can be seen from the above technical solutions, by obtaining the content viewing history data of the users in the content platform, the time length of the users viewing different contents on the content platform can be determined, and the time consumption ratio data of the users is obtained according to the content viewing history data, which is used to identify the time consumption distribution of the users viewing different contents, i.e., the proportion of the time of the users on the content platform consumed by different contents. Since the consumption of time can objectively and quantitatively reflect the preference degree of the users for different contents in the content platform, the time consumption ratio data can represent the content preference of the users from the time dimension, which can replace the user tags in the related technology as the user features reflecting the user preference. The users are classified by the time consumption ratio data to obtain a plurality of user categories, and the users in the same user category obtained in this way have similar preferences for contents. When recommending the target content for the target user, the target user category to which the target user belongs is determined based on the foregoing user classification. Since the users in the content platform that belong to the target user category have similar content preferences, the users in the target user category can be used as the basis for content recommendation, and this kind of recommendation based on objective data such as time consumption ratio data does not need to consider various user tags and content tags in the related technology, thereby breaking the shackles brought by the original tags, avoiding various problems of tags in recommendation, and ensuring the accuracy of content recommendation.

[0107] It should be noted that although the content viewing history data is real-time changing, the user in the content platform is classified each time the content viewing history data changes, and a more accurate user category can be obtained, but the content interested by the user in a period of time is similar, the user category obtained by calculation each time changes little, and the number of content viewing history data is large, and if the user category is calculated in real time, a large amount of computing resources will be consumed. Based on this, in order to reduce the amount of calculation under the premise of ensuring that the user category is divided more accurately, the content viewing history data can be stored offline, and based on a preset period or a recommendation condition, the content viewing history data is obtained, and when it is determined that the content viewing history data is updated, S202 is executed again. The recommendation condition is a factor affecting the data amount of the content viewing history data. For example, the recommendation condition can be that the content viewing history data reaches 20 million, the content viewing history data increases by 1,000, etc., and the preset period can be one day, one week, etc.

[0108] After the user logs in the content platform, the server will receive the request sent by the user, and the request can be preprocessed. The following introduces three preprocessing methods, and it should be noted that the three methods can be used alone or in combination.

[0109] Method one: request verification.

[0110] According to the parameters included in the request, such as user ID list user_ids, QUA parameter, client version number and the like, the request is verified to ensure the legality of the request. For malicious or illegal requests, they can be directly shielded or blacklisted, and legal requests can continue to be reported for subsequent analysis and processing. For example, it is verified whether the user ID list user_ids is empty to ensure that the request comes from the user of the content platform. The QUA parameter is a parameter for identifying client information, for example, in the form of key-value pair, connected by & between key-value, and the server can give adaptive content according to the QUA information. For example, it is verified whether the QUA parameter is correct so that the push of target content can be realized.

[0111] Method two: frequency limitation.

[0112] According to the number of requests, the target user is determined, and by limiting the frequency of requests, the server can be prevented from crashing. For example, after receiving a login request of a user logging into a content platform, it is determined whether the current number of requests is within the maximum load capacity of the server, i.e., whether the number of login requests obtained within a preset period of time is less than a preset threshold. If yes, it means that the server can process the current number of login requests, and the user sending the request is determined as the target user in the content platform. If no, it means that the current number of login requests exceeds the load of the server. In order to ensure the reliability of the server, the login request exceeding the preset threshold is discarded, and the target content is not recommended to the user sending the request, so as to avoid the server resources being exhausted by the overloaded login request and the problem of server crash, so that the server is always in a usable state, and the reliability of the server is ensured.

[0113] Method three: request preprocessing.

[0114] The request is processed so that the request is a valid request and subsequent steps can be processed. For example, missing or illegal data items in the request are removed so that subsequent steps can identify the data structure in the request. For another example, the number of users viewing each content in the content platform is obtained so as to obtain the category proportion data of the content in the content platform, for example, all users in the content platform are divided into three user categories A, B and C, and when a user belonging to category A views a content, the number of viewers of the content is increased by 1, so that the category proportion data of different contents in the content platform can be obtained in real time.

[0115] In order to meet the operation needs of the content platform, intervention rules can be added when recommending target contents to target users, and contents meeting the intervention rules are inserted into the target contents recommended to the target users, so that the content platform can recommend some contents meeting the operation needs to the target users. For example, the contents meeting the operation needs can be the contents of advertisers.

[0116] In order to meet the management needs of the content platform, filtering rules can be added when recommending target contents to target users, and contents meeting the filtering rules are deleted from the target contents recommended to the target users, so that the target contents recommended to the target users by the content platform can meet the management requirements of the content platform. For example, in the over the top (OTT) video service scenario based on open Internet, there is a concept of license control. For some contents that have not passed the audit of a certain license (there are multiple license parties, and only some of them pass the audit), the users belonging to the license parties that have not passed the audit are not recommended.

[0117] Next, the embodiments of the present application will be described in detail with reference to the accompanying drawings. Figure 3 and Figure 4The content recommendation method provided by the embodiments of the present application is described by taking the content platform as a live broadcast platform and the content as a live broadcast room as an example. Referring to Figure 3 FIG. 3 is a schematic diagram of an application scenario embodiment of a content recommendation method provided by the embodiments of the present application.

[0118] S301: The client is triggered to log in to the live broadcast platform.

[0119] The live broadcast platform can be in the form of a client and installed in the user's mobile phone. The user logs in to the live broadcast platform by opening the live broadcast platform client in the mobile phone. At this time, the server receives a login request sent by the live broadcast platform for the user to log in. The request at least includes the user ID, so that the user category corresponding to the user can be obtained in the database through the user ID.

[0120] S302: The server pre-processes the request.

[0121] The request pre-processing includes three steps, which are request verification by using the aforementioned manner one, frequency limitation by using the aforementioned manner two, and data pre-processing by using manner three, which will not be described here.

[0122] S303: The server obtains the category proportion data of the live broadcast room in the live broadcast platform.

[0123] As can be known from the data pre-processing step S302, the category proportion data is known data, which can be obtained from the user classification category database 300. The user classification category database 300 stores the time consumption proportion data of the user. Through the category proportion data, the number of users who are viewing the corresponding live broadcast room belonging to the target user category can be obtained.

[0124] S304: The server determines the target user in the live broadcast platform.

[0125] For example, after the request of the user logging in to the live broadcast platform is pre-processed, it is determined that the user is a target user. The target user is a target user category in multiple user categories, which can be obtained from the user classification category database 300. The user classification category database 300 is used to store the corresponding relationship between the user and the user category.

[0126] It should be noted that the server will pre-classify all users in the live broadcast platform into multiple user categories, as described in S310-S330. According to the user and the user category to which the user belongs determined by the server in advance, the target user category to which the target user belongs can be determined from the multiple user categories.

[0127] S305: The server determines the top n live broadcast rooms of the target user category in the category proportion data as the to-be-determined live broadcast rooms.

[0128] From the foregoing, the user belonging to the same target user category as the target user has similar preferences, and the target user determined by the user belonging to the target user category is also more interested. At the same time, the more people viewing the live room, the higher the popularity of the live room, so the top n live rooms of the target user category are filtered out from the live rooms determined by the target user category, which further improves the accuracy of the recommendation.

[0129] For example, when recommending a live room to a target user i, the user category of the target user i belongs to A, and then n live rooms are selected from the live room category proportion data according to the A category user from high to low as the live room to be determined.

[0130] S306: The server determines whether the live room to be determined meets the filtering rule.

[0131] The live room meeting the filtering rule is deleted from the live room to be determined, and is not recommended.

[0132] S307: The server inserts the live room meeting the intervention rule into the live room to be determined.

[0133] The live room meeting the intervention rule can be further inserted into the live room to be determined for recommendation to the target user. The present application embodiment does not specifically limit the order of execution of S306 and S307, for example, S306 and S307 can be executed simultaneously.

[0134] S308: The server determines the target live room recommended to the target user according to the live room to be determined.

[0135] Finally, the live room determined by S305-S307 is recommended to the target user, and the target live room is recommended to the user through the client.

[0136] It should be noted that the live room can be further recommended to the user after S308, for example, if it is determined that the target user enters a first live room in the content platform, a first user belonging to the target user category in the first live room is obtained.

[0137] When the target user enters the first live room in the recommended live room, the first user viewing the first live room at this time and belonging to the target user category as the target user is obtained.

[0138] According to the time consumption proportion data of the first user, a target live room recommended to the target user in the first live room is determined, and the target live room is displayed on the interface of the live platform client.

[0139] The server pre-divides users in the live broadcast platform into multiple user categories is described below. It should be noted that S310-S330 can be offline processed.

[0140] S310: The server calculates the time-consuming proportion data of the user.

[0141] The content viewing history data of the user in the content platform can be obtained from the user live broadcast database, and the time-consuming proportion data of the user is determined according to the viewing time length of the identified content in the content viewing history data. For example, the content viewing history data can be obtained based on a preset period or a recommendation condition. When it is determined that the content viewing history data is updated, the following steps are executed. According to the content viewing history data, the time-consuming proportion data of the user can be determined, and two ways are introduced below, which correspond to way one and way two respectively. Either way can obtain the time-consuming proportion data of the user.

[0142] Way one:

[0143] Firstly, the target live broadcast room in the content platform is determined.

[0144] The live broadcast room whose live broadcast rating reaches a preset condition is determined as the target live broadcast room in the live broadcast platform. For example, the live broadcast rooms with top k heat ranks in the live broadcast platform can be determined as the target live broadcast room.

[0145] Secondly, the time-consuming proportion data of the user is determined according to the viewing time length of the target live broadcast room corresponding to the content viewing history data, and the obtained time-consuming proportion data is stored in the user feature database 100.

[0146] Way two:

[0147] Firstly, the type of the live broadcast room in the live broadcast platform is determined.

[0148] For example, the live broadcast room can be divided into game live broadcast room, talent live broadcast room, life live broadcast room, etc.

[0149] Then, the total viewing time length of the live broadcast rooms belonging to the same type is determined according to the content viewing history data.

[0150] The viewing time lengths corresponding to the live broadcast rooms belonging to the same type are summed up. For example, there are m game live broadcast rooms in the live broadcast platform, and the viewing time lengths corresponding to the m game live broadcast rooms in the content viewing history data are summed up to obtain the total viewing time length of the game live broadcast room.

[0151] Finally, the time-consuming proportion data of the user is determined according to the total viewing time length, and the obtained time-consuming proportion data is stored in the user feature database 200.

[0152] S320: The server calculates the user similarity.

[0153] Obtain the time consumption proportion data of the user from the user feature database, and according to the proportion data in the same content dimension in the time consumption proportion data, obtain the user similarity between users in the live broadcast platform by using formula (1).

[0154] S330: The server divides the user categories.

[0155] Classify the users in the live broadcast platform based on the user similarity, obtain multiple user categories, and store the corresponding relationship between the user similarity, the user and the user category to which the user belongs in the user classification category database 300.

[0156] The Euclidean distance between users can be obtained by formula (1), and the Euclidean distance size can reflect whether the users are similar. The users in the live broadcast platform are divided into multiple user categories by the K-means clustering algorithm, and each user and the user category to which the user belongs are saved in the database.

[0157] Referring to Figure 4 , the figure is a schematic diagram of a content recommendation system provided by an embodiment of the present application. In Figure 4 , the input module 401, the recommendation module 402, the intervention module 403 and the output module 404 are included.

[0158] The input module 401 is configured to receive the live broadcast data generated by the live broadcast platform, such as content viewing history data, user time consumption proportion data, user ID, etc., pre-process the live broadcast data, and execute S302 in the method. Figure 3

[0159] The recommendation module 402 is configured to determine the live room to be recommended for the target user, execute S303-S305 and S310-S330 in the method. Figure 3

[0160] The intervention module 403 is configured to insert and delete the live room to be recommended, and execute S306-S307 in the method. Figure 3

[0161] The output module 404 is configured to output the recommendation result of the target live room, and execute S308 in the method. Figure 3

[0162] For the content recommendation method provided in the above embodiment, an embodiment of the present application further provides a content recommendation device.

[0163] Referring to Figure 5 , the figure is a schematic diagram of a content recommendation device provided by an embodiment of the present application. As Figure 5 ​​​​As shown, the content recommendation apparatus 500 comprises an acquisition unit 501, a first determination unit 502, a classification unit 503, a second determination unit 504, and a third determination unit 505.

[0164] The acquisition unit 501 is configured to acquire content viewing history data of a user in a content platform.

[0165] The first determination unit 502 is configured to determine time consumption proportion data of the user according to a viewing duration of an identified content in the content viewing history data, the time consumption proportion data being used to identify a time consumption distribution of the user in viewing different contents in the content platform.

[0166] The classification unit 503 is configured to classify the user in the content platform based on the time consumption proportion data, to obtain a plurality of user categories.

[0167] The second determination unit 504 is configured to determine a target user in the content platform, the target user belonging to a target user category in the plurality of user categories.

[0168] The third determination unit 505 is configured to determine target content recommended to the target user according to the user belonging to the target user category in the content platform.

[0169] As a possible implementation manner, the content in the content platform is a live room, and the first determination unit 502 is configured to:

[0170] determine a target live room in the content platform, the target live room being a live room whose live rating reaches a preset condition;

[0171] determine the time consumption proportion data of the user according to a viewing duration of the target live room in the content viewing history data.

[0172] As a possible implementation manner, the first determination unit 502 is configured to:

[0173] determine a type of content in the content platform;

[0174] determine a total viewing duration of the content of the same type according to the content viewing history data;

[0175] determine the time consumption proportion data of the user according to the total viewing duration.

[0176] As a possible implementation manner, the third determination unit 505 is configured to:

[0177] acquire category proportion data of the content in the content platform, the category proportion data being used to identify a proportion of the user belonging to the target user category in viewing the content.

[0178] determine the content whose category proportion data meets a preset condition as a pending content;

[0179] determine the target content recommended to the target user according to the pending content.

[0180] As a possible implementation manner, the content in the content platform is a live room, and the third determining unit 505 is configured to:

[0181] if it is determined that the target user enters a first live room in the content platform, obtain a first user belonging to the target user category in the first live room;

[0182] determine a target live room recommended to the target user in the first live room according to the time consumption proportion data of the first user.

[0183] As a possible implementation manner, the classifying unit 503 is configured to:

[0184] determine a user similarity between users in the content platform according to the proportion data in the time consumption proportion data and identifying the same content dimension;

[0185] classify the users in the content platform based on the user similarity to obtain a plurality of user categories.

[0186] As a possible implementation manner, the apparatus 500 is further configured to:

[0187] obtain the content viewing history data based on a preset period or a recommendation condition;

[0188] when it is determined that the content viewing history data is updated, perform the step of determining the time consumption proportion data of the user according to the viewing duration of the content identified in the content viewing history data.

[0189] As a possible implementation manner, the second determining unit 504 is configured to:

[0190] receive a login request of a user logging in the content platform;

[0191] if the number of login requests obtained in a preset period is less than a preset threshold, determine the user corresponding to the login request as the target user in the content platform.

[0192] The content recommendation device provided by the above embodiments can determine the time length of the user viewing different contents on the content platform by obtaining the content viewing history data of the user in the content platform, and obtain the time consumption proportion data of the user according to the content viewing history data. The time consumption proportion data is used to identify the time consumption distribution of the user viewing different contents, that is, the proportion of the time of the user on the content platform consumed by different contents. Since the consumption of time can objectively and quantitatively reflect the preference degree of the user for different contents in the content platform, the time consumption proportion data can represent the content preference of the user from the time dimension, and can replace the user label in the related art as a user feature reflecting the user preference. The user is classified by the time consumption proportion data to obtain a plurality of user categories. The users in the same user category obtained in this way have similar preferences for contents. When recommending target contents for a target user, the target user category to which the target user belongs can be determined based on the foregoing user classification. Since the content preferences of the users belonging to the target user category in the content platform are relatively similar, the users of the target user category can be used as the basis for content recommendation. This recommendation method based on objective data such as time consumption proportion data does not need to consider various user labels and content labels in the related art, thereby breaking the constraints of the original labels. Not only various problems of the labels in the recommendation are avoided, but also the content preferences of some users in the target user category are more likely to meet the preferences of the target user, which can ensure the accuracy of the content recommendation.

[0193] The foregoing content recommendation device can be a computer device, which can be a server and can also be a terminal device. The computer device provided by the embodiments of the present application will be introduced from the perspective of hardware embodiment. Wherein, Figure 6 Fig. 1 shows a structural schematic diagram of a server, Figure 7 Fig. 2 shows a structural schematic diagram of a terminal device.

[0194] Referring to Figure 6 , Figure 6Fig. 1 shows a schematic diagram of a server according to an embodiment of the present application. The server 1400 can have a great difference due to different configurations or performances, and can include one or more central processing units (CPU) 1422 (e.g., one or more processors) and a memory 1432, one or more storage media 1430 (e.g., one or more mass storage devices) storing applications 1442 or data 1444. The memory 1432 and the storage media 1430 can be temporary or persistent storage. The programs stored in the storage media 1430 can include one or more modules (not shown in the figure), each of which can include a series of instructions operated in the server. Further, the central processing unit 1422 can be configured to communicate with the storage media 1430 and execute the series of instructions operated in the storage media 1430 on the server 1400.

[0195] The server 1400 can further include one or more power supplies 1426, one or more wired or wireless network interfaces 1450, one or more input / output interfaces 1458, and / or one or more operating systems 1441, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0196] The steps performed by the server in the above embodiments can be based on the Figure 6 The server structure shown in the figure.

[0197] The CPU 1422 is configured to perform the following steps:

[0198] Obtain content viewing history data of a user in a content platform;

[0199] According to the viewing duration of the identified content in the content viewing history data, determine time-consuming proportion data of the user, the time-consuming proportion data being used to identify the time-consuming distribution of the user in viewing different contents in the content platform;

[0200] Classify the users in the content platform based on the time-consuming proportion data to obtain a plurality of user categories;

[0201] Determine a target user in the content platform, the target user belonging to a target user category in the plurality of user categories;

[0202] According to the users belonging to the target user category in the content platform, determine target content recommended to the target user.

[0203] Optionally, the CPU 1422 can also execute the method steps of any of the specific implementation manners of the content recommendation method in the embodiments of the present application.

[0204] Referring to Figure 7 , Figure 7 A structural schematic diagram of a terminal device provided in the embodiments of the present application is shown. A block diagram of part of the structure of a smart phone related to the terminal device provided in the embodiments of the present application is shown, referring to Figure 7 The smart phone includes a radio frequency (RF) circuit 1510, a memory 1520, an input unit 1530, a display unit 1540, a sensor 1550, an audio circuit 1560, a wireless fidelity (WiFi) module 1570, a processor 1580, and a power supply 1590, and the like. Those skilled in the art can understand that Figure 7 The structure of the smart phone shown in the embodiments of the present application does not constitute a limitation on the smart phone, and can include more or fewer components than shown, or combine certain components, or different arrangement of components.

[0205] The specific components of the smart phone will be introduced below Figure 7

[0206] The RF circuit 1510 can be used for receiving and sending signals in the process of information or call, in particular, receiving the downlink information of the base station and processing by the processor 1580; in addition, sending the uplink data to the base station. Usually, the RF circuit 1510 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, and the like. In addition, the RF circuit 1510 can also communicate with the network and other devices through wireless communication. The above wireless communication can use any communication standard or protocol, including but not limited to global system for mobile communication (GSM), general packet radio service (GPRS), code division multiple access (CDMA), wideband code division multiple access (WCDMA), long term evolution (LTE), email, short message service (SMS), and the like.

[0207] ​The memory 1520 can be used to store software programs and modules, and the processor 1580 implements various function applications and data processing of the smart phone by running the software programs and modules stored in the memory 1520. The memory 1520 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data created according to the use of the smart phone (such as audio data, a phone book, etc.), and the like. In addition, the memory 1520 can include a high-speed random access memory, and can also include a non-volatile memory, for example, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state memory device.

[0208] The input unit 1530 can be used to receive inputted digital or character information, and to generate key signal input related to the user settings and function control of the smart phone. Specifically, the input unit 1530 can include a touch panel 1531 and other input devices 1532. The touch panel 1531, also called a touch screen, can collect a user's touch operation (such as the user's operation on or near the touch panel 1531 using a finger, a stylus, or any suitable object or accessory) and drive the corresponding connection device according to the pre-set program. Optionally, the touch panel 1531 can include two parts of a touch detection device and a touch controller. Among them, the touch detection device detects the touch position of the user and detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, and converts it into touch coordinates, and then sends it to the processor 1580, and can also receive the command from the processor 1580 and execute it. In addition, the touch panel 1531 can be implemented in various types such as a resistive type, a capacitive type, an infrared type, and a surface acoustic wave type. In addition to the touch panel 1531, the input unit 1530 can also include other input devices 1532. Specifically, the other input devices 1532 can include one or more of a physical keyboard, a function key (such as a volume control key, an on-off key, etc.), a trackball, a mouse, a joystick, and the like.

[0209] The display unit 1540 can be used to display information input by a user or information provided to the user as well as various menus of the smart phone. The display unit 1540 can include a display panel 1541, which can be configured in the form of a Liquid Crystal Display (LCD), an Organic Light-Emitting Diode (OLED), or the like, optionally. Further, a touch panel 1531 can cover the display panel 1541, which, when detecting a touch operation thereon or in the vicinity thereof, transmits to the processor 1580 to determine the type of the touch event, and then the processor 1580 provides corresponding visual output on the display panel 1541 according to the type of the touch event. Although in the above description, the touch panel 1531 and the display panel 1541 are implemented as two independent components to realize the input and output functions of the smart phone, in some embodiments, the touch panel 1531 and the display panel 1541 can be integrated to realize the input and output functions of the smart phone. Figure 7

[0210] The smart phone can further include at least one sensor 1550, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor can include an ambient light sensor and a proximity sensor, wherein the ambient light sensor can adjust the brightness of the display panel 1541 according to the brightness of ambient light, and the proximity sensor can turn off the display panel 1541 and / or the backlight when the smart phone is moved to the ear. As one of the motion sensors, the accelerometer sensor can detect the magnitude of acceleration in each direction (generally three axes), and when at rest, it can detect the magnitude and direction of gravity, which can be used for applications of identifying the posture of the smart phone (such as switching between landscape and portrait screens, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, knocking), and the like. As for other sensors that the smart phone can be configured, such as a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, and the like, they will not be described here.

[0211] The audio circuit 1560, the speaker 1561, and the microphone 1562 can provide an audio interface between the user and the smart phone. The audio circuit 1560 can convert the received audio data into an electrical signal and transmit it to the speaker 1561, which converts the electrical signal into a sound signal for output. On the other hand, the microphone 1562 converts the collected sound signal into an electrical signal, which is received by the audio circuit 1560 and converted into audio data. The audio data is output to the processor 1580 for processing, and then transmitted to another smart phone via the RF circuit 1510, or output to the memory 1520 for further processing.

[0212] ​WiFi belongs to short distance wireless transmission technology, and the smart phone can help users send and receive emails, browse web pages and access streaming media through the WiFi module 1570, which provides wireless broadband Internet access for users. Although Figure 7 The WiFi module 1570 is shown, but it is understood that it is not a necessary component of the smart phone and can be omitted as needed without changing the essence of the application.

[0213] The processor 1580 is the control center of the smart phone, which connects all parts of the smart phone through various interfaces and lines, and performs various functions of the smart phone and processes data by running or executing software programs and / or modules stored in the memory 1520 and calling data stored in the memory 1520. Optionally, the processor 1580 can include one or more processing units; preferably, the processor 1580 can integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface and application program, etc., and the modem processor mainly processes wireless communication. It is understood that the above-mentioned modem processor can also not be integrated into the processor 1580.

[0214] The smart phone also includes a power supply 1590 (such as a battery) for powering various components, and the power supply can be logically connected to the processor 1580 through a power management system, so that the power management system can realize functions such as charging, discharging and power consumption management.

[0215] Although not shown, the smart phone can also include a camera, a Bluetooth module, etc., which will not be described here.

[0216] In the embodiments of the present application, the memory 1520 included in the smart phone can store program codes and transmit the program codes to the processor.

[0217] The processor 1580 included in the smart phone can execute the content recommendation method provided by the above-mentioned embodiments according to the instructions in the program codes.

[0218] The embodiments of the present application also provide a computer readable storage medium for storing a computer program, which is used to execute the content recommendation method provided by the above-mentioned embodiments.

[0219] The embodiments of the present application also provide a computer program product or computer program, which includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute the content recommendation method provided in various optional implementation manners of the above-mentioned aspects.

[0220] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program performs the steps of the above-mentioned method embodiments when executed; and the foregoing storage medium can be at least one of the following: read-only memory (English: read-only memory, abbreviation: ROM), RAM, magnetic disk or optical disk and various storage medium capable of storing program codes.

[0221] It should be noted that each of the embodiments in the specification is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment focuses on the difference from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, they are described more simply, and the relevant parts can be referred to the part of the description of the method embodiments. The above-described device and system embodiments are only illustrative, and the units described as separate components can be or can not be physically separated, and the components shown as units can be or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiments according to the actual needs. Those skilled in the art can understand and implement without creative labor.

[0222] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any skilled person in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A content recommendation method characterized by, The method comprises: obtaining content viewing history data of a user in a content platform; determining time consumption proportion data of the user according to a viewing duration of identified content in the content viewing history data, the time consumption proportion data being used to identify a time consumption distribution of the user in viewing different content through the content platform; classifying users in the content platform based on the time consumption proportion data to obtain a plurality of user categories; wherein a similarity of preferences of users belonging to a same user category satisfies a similarity threshold; determining a target user in the content platform, the target user belonging to a target user category in the plurality of user categories; determining target content recommended to the target user according to users belonging to the target user category in the content platform; The determination of the target content recommended to the target user according to the users belonging to the target user category in the content platform comprises: obtaining category proportion data of content in the content platform, the category proportion data being used to identify a proportion of users belonging to the target user category in users viewing the content; determining content for which the category proportion data meets a preset condition as pending content; determining the target content recommended to the target user according to the pending content.

2. The method of claim 1, wherein, The content in the content platform is a live room, and the determination of the time consumption proportion data of the user according to the viewing duration of the identified content in the content viewing history data comprises: determining a target live room in the content platform, the target live room being a live room whose live rating meets a preset condition; determining the time consumption proportion data of the user according to a viewing duration of the target live room in the content viewing history data.

3. The method of claim 1, wherein, The determination of the time consumption proportion data of the user according to the viewing duration of the identified content in the content viewing history data comprises: determining a type of content in the content platform; determining a total viewing duration of content belonging to a same type according to the content viewing history data; determining the time consumption proportion data of the user according to the total viewing duration.

4. The method according to any one of claims 1 to 3, characterized in that, If the content in the content platform is a live room, the determination of the target content recommended to the target user according to the users belonging to the target user category in the content platform further comprises: if it is determined that the target user enters a first live room in the content platform, obtaining a first user belonging to the target user category in the first live room; determining a target live room recommended to the target user in the first live room according to the time consumption proportion data of the first user.

5. The method of claim 1, wherein, The classification of the users in the content platform based on the time consumption proportion data to obtain a plurality of user categories comprises: determining a user similarity between users in the content platform according to proportion data in the time consumption proportion data identifying a same content dimension; classifying the users in the content platform based on the user similarity to obtain a plurality of user categories.

6. The method of claim 1, wherein, The method further comprises: obtaining the content viewing history data based on a preset period or a recommendation condition; When the content viewing history data is determined to be updated, the step of determining the time consumption proportion data of the user according to the viewing duration of the identified content in the content viewing history data is performed.

7. The method of claim 1, wherein, The target user in the content platform is determined, including: receiving a login request of a user logging into the content platform; if the number of login requests obtained within a preset period of time is less than a preset threshold, the user corresponding to the login request is determined as the target user in the content platform.

8. A content recommendation apparatus characterized by comprising: The device includes an acquisition unit, a first determination unit, a classification unit, a second determination unit, and a third determination unit. The acquisition unit is configured to acquire content viewing history data of a user in a content platform. The first determination unit is configured to determine time consumption proportion data of the user according to a viewing duration of an identified content in the content viewing history data, the time consumption proportion data being used to identify a time consumption distribution of the user in the content platform for viewing different contents through the content platform. The classification unit is configured to classify the users in the content platform based on the time consumption proportion data to obtain a plurality of user categories; wherein a similarity of preferences of users belonging to a same user category satisfies a similarity threshold. The second determination unit is configured to determine a target user in the content platform, the target user belonging to a target user category in the plurality of user categories. The third determination unit is configured to determine target content recommended to the target user according to users belonging to the target user category in the content platform. The third determination unit is configured to: acquire category proportion data of contents in the content platform, the category proportion data being used to identify a proportion of users belonging to the target user category among users viewing the contents; determine contents for which the category proportion data meets a preset condition as pending contents; determine the target content recommended to the target user according to the pending contents.

9. The apparatus of claim 8, wherein, The content in the content platform is a live room, and the first determination unit is configured to: determine a target live room in the content platform, the target live room being a live room whose live rating meets a preset condition; determine the time consumption proportion data of the user according to a viewing duration of the target live room in the content viewing history data.

10. The apparatus of claim 8, wherein, The first determination unit is configured to: determine a type of contents in the content platform; determine a total viewing duration of contents belonging to a same type according to the content viewing history data; determine the time consumption proportion data of the user according to the total viewing duration.

11. The apparatus of any of claims 8-10, wherein, The content in the content platform is a live room, and the third determination unit is configured to: if it is determined that the target user enters a first live room in the content platform, acquire a first user belonging to the target user category in the first live room; determine a target live room recommended to the target user in the first live room according to the time consumption proportion data of the first user.

12. A computer device, comprising: The device includes a processor and a memory: The memory is configured to store program code and transmit the program code to the processor; The processor is configured to execute the method according to the instructions in the program code.

13. A computer-readable storage medium, characterized in that, The computer readable storage medium is configured to store a computer program, and the computer program is loaded and executed by the processor to implement the method in any one of claims 1-7.

14. A computer program product, characterised in that, The computer program product comprises instructions which, when executed on a computer device, cause the computer device to perform the method in any one of claims 1-7.

Images