Media resource recommendation method, device, system, electronic device and storage medium

Abstract

The present application relates to a media resource recommendation method, device, system, electronic device and storage medium, wherein the media resource recommendation method is applied to a multimedia interactive device, including: obtaining user behavior data of a user group interacting with the multimedia interactive device in real time; obtaining key-value pair samples based on the user behavior data, clustering the key-value pair samples based on a clustering algorithm to obtain classification information of the user group in different time periods, and obtaining a user portrait model of the user group based on the classification information; generating a media resource recommendation result for the user group based on the user portrait model. Through the present application, the problem of low accuracy of media resource recommendation is solved, and high-precision media resource recommendation based on user group portrait is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of media resource recommendation, and in particular to a method, device, system, electronic device and storage medium for media resource recommendation. Background Art

[0002] With the rapid development of Internet Protocol Television (IPTV), the formats and contents of programs are becoming more and more diverse. How to recommend these rich programs to IPTV users according to their preferences has become a hot research area. In related technologies, although there are methods for grouping user groups for watching IPTV, users still need to search manually, which makes the user operation cumbersome and error-prone. In addition, the existing recommendation methods cannot quickly divide the user group portraits so as to quickly and accurately find the media resources recommended for the user groups that meet certain conditions, resulting in low accuracy of media resource recommendation and poor user experience.

[0003] Currently, no effective solution has been proposed for the problem of low accuracy of media resource recommendations in related technologies. Summary of the invention

[0004] Embodiments of the present application provide a media resource recommendation method, apparatus, system, electronic device, and storage medium to at least solve the problem of low accuracy of media resource recommendation in the related art.

[0005] In a first aspect, an embodiment of the present application provides a media resource recommendation method, which is applied to a multimedia interactive device, and the method includes:

[0006] Acquire user behavior data of real-time interaction between a user group and the multimedia interactive device;

[0007] Acquire key-value pair samples according to the user behavior data, cluster the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtain a user portrait model of the user group according to the classification information;

[0008] Generate media resource recommendation results for the user group based on the user portrait model.

[0009] In some embodiments, clustering the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods includes:

[0010] Convert the key-value pair samples into multi-dimensional coordinates, and obtain at least two preset center points in all the multi-dimensional coordinates; wherein the multi-dimensional coordinates include on-demand saturation;

[0011] assigning the multidimensional coordinates to the center point based on the on-demand saturation to generate a first cluster, calculating a first coordinate mean value between all the multidimensional coordinates in the first cluster, and calculating first distance information between the coordinate mean value and the center point;

[0012] In the case where the first distance information is greater than or equal to a preset distance threshold, the multidimensional coordinates are assigned to the first coordinate mean based on the on-demand saturation to generate a second cluster, a second coordinate mean between all the multidimensional coordinates in the second cluster is calculated, and second distance information between the second coordinate mean and the first coordinate mean is calculated;

[0013] When the second distance information is smaller than the distance threshold, the classification information is obtained according to the coordinate average value.

[0014] In some embodiments, assigning the multidimensional coordinates to the first coordinate means based on the on-demand saturation to generate a second cluster comprises:

[0015] When the on-demand saturation is less than a preset saturation threshold, the multi-dimensional coordinates corresponding to the on-demand saturation are used as the elimination points;

[0016] When the rejection points are removed, the multi-dimensional coordinates are reallocated based on the on-demand saturation to generate the second cluster.

[0017] In some embodiments, the multidimensional coordinates further include a user timestamp; and assigning the multidimensional coordinates to the center point based on the on-demand saturation to generate a first cluster includes:

[0018] Acquire preset magnification information, and acquire a target timestamp according to the magnification information and the user timestamp;

[0019] The multi-dimensional coordinates are assigned to the center point based on the target timestamp and the on-demand saturation to generate the first cluster.

[0020] In some embodiments, obtaining a key-value pair sample according to the user behavior data includes:

[0021] Generate corresponding user timestamps according to the user behavior data obtained at different time points;

[0022] Analyze and process all the user behavior data to obtain media viewing information corresponding to each user behavior data;

[0023] The key-value pair sample is acquired according to the user timestamp and the media resource viewing information.

[0024] In some embodiments, generating the media resource recommendation result for the user group according to the user portrait model includes:

[0025] The user behavior data is input into a well-trained user preference model to obtain a user group preference type, and a media resource recommendation result for the user group is generated based on the user group preference type and the user portrait model.

[0026] In some embodiments, obtaining user behavior data of the user group interacting with the multimedia interactive device includes:

[0027] Obtaining a video page service in the multimedia interactive device, and obtaining a preset embedding function on the video page service;

[0028] The user behavior data is obtained by utilizing the embedding function.

[0029] In a second aspect, an embodiment of the present application provides a media resource recommendation device, which is applied to a multimedia interactive device, and the device includes: an acquisition module, a classification module, and a generation module;

[0030] The acquisition module is used to acquire user behavior data of real-time interaction between a user group and the multimedia interactive device;

[0031] The classification module is used to obtain key-value pair samples according to the user behavior data, cluster the key-value pair samples according to the clustering algorithm to obtain classification information of the user group in different time periods, and obtain a user portrait model of the user group according to the classification information;

[0032] The generating module is used to generate media resource recommendation results for the user group according to the user portrait model.

[0033] In a third aspect, an embodiment of the present application provides a media resource recommendation system, the system comprising: a multimedia interaction device, a transmission device, and a server device; wherein the multimedia interaction device is connected to the server device via the transmission device;

[0034] The multimedia interaction device is used to generate user behavior data for interaction with a user group;

[0035] The transmission device is used to obtain the user behavior data from the multimedia interaction device and send the user behavior data to the server device;

[0036] The server device is used to execute the media resource recommendation method described in the first aspect above.

[0037] In a fourth aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the media resource recommendation method as described in the first aspect above is implemented.

[0038] In a fifth aspect, an embodiment of the present application provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the media asset recommendation method as described in the first aspect above.

[0039] Compared with the related art, the media resource recommendation method, device, system, electronic device and storage medium provided in the embodiments of the present application are applied to multimedia interactive devices, by obtaining user behavior data of a user group interacting with the multimedia interactive device in real time; obtaining key-value pair samples based on the user behavior data, clustering the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtaining a user portrait model of the user group based on the classification information; generating media resource recommendation results for the user group based on the user portrait model, thereby solving the problem of low accuracy of media resource recommendation and realizing high-precision media resource recommendation based on user group portrait.

[0040] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0042] Figure 1 is an application environment diagram of a media resource recommendation method according to an embodiment of the present application;

[0043] Figure 2 is a flow chart of a media resource recommendation method according to an embodiment of the present application;

[0044] Figure 3 is a flow chart of another media resource recommendation method according to an embodiment of the present application;

[0045] Figure 4 is a structural block diagram of a media resource recommendation device according to an embodiment of the present application;

[0046] Figure 5 is a structural block diagram of a media resource recommendation system according to an embodiment of the present application;

[0047] Figure 6 It is a structural diagram of the inside of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application. In addition, it can also be understood that although the efforts made in this development process may be complex and lengthy, for ordinary technicians in the field related to the contents disclosed in the present application, some changes such as design, manufacturing or production based on the technical contents disclosed in the present application are only conventional technical means, and should not be understood as insufficient contents disclosed in the present application.

[0049] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0050] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to greater than or equal to two. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The terms "first", "second", "third" and the like involved in the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects.

[0051] The media resource recommendation method provided in this embodiment can be applied in the following aspects: Figure 1 In the application scenario shown. Among them, the multimedia interactive device 102 communicates with the server device 104 through the network. The server device 104 obtains key-value pair samples based on the user behavior data of the real-time interaction between the user group and the multimedia interactive device 102, and clusters the key-value pair samples according to the clustering algorithm to obtain the classification information of the user group in different time periods, and then obtains the user portrait model of the user group according to the classification information; finally, the server device 104 generates the media resource recommendation results for the user group according to the user portrait model. Among them, the multimedia interactive device 102 can be but is not limited to various large-screen display devices with multimedia interactive functions such as various televisions and home theaters, and the server device 104 can be implemented with an independent server or a server cluster composed of multiple servers.

[0052] This embodiment provides a media resource recommendation method, which is applied to a multimedia interactive device. Figure 2 is a flow chart of a media resource recommendation method according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:

[0053] Step S210, obtaining user behavior data of real-time interaction between a user group and the multimedia interactive device.

[0054] The multimedia interactive device may be a television or a home theater, and the multimedia interactive device may be deployed in application scenarios such as homes and elderly activity centers. The user group refers to multiple users who interact with and watch movies with the multimedia interactive device in the application scenario. During the interaction between the user group and the multimedia interactive device, the processor may collect user behavior data generated by user on-demand; it should be noted that the user behavior data shall include information such as user account, on-demand time, asset name, asset age, and on-demand duration.

[0055] Step S220, obtaining key-value pair samples according to the user behavior data, clustering the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtaining a user portrait model of the user group according to the classification information.

[0056] In some embodiments, the above-mentioned acquisition of key-value pair samples according to the user behavior data includes the following steps: generating corresponding user timestamps according to the user behavior data acquired at different time points; parsing and processing all the user behavior data to obtain the media asset viewing information corresponding to each of the user behavior data; and acquiring the key-value pair samples according to the user timestamps and the media asset viewing information. Among them, the user timestamp refers to the interaction time point corresponding to the user behavior data; the media asset viewing information includes information such as the asset type and asset tag viewed by the user. Specifically, the unique user ID corresponding to the user account in the above-mentioned user behavior data can be composed of a series of primary keys Key, and the Value corresponding to the Key is determined according to the actual user interaction situation; for example, considering that the film and television asset category, the film and television asset age in the film and television assets usually have a strong relationship with the user age, the data form of the key-value pair can be set to (user ID, (film and television asset category, film and television asset age)); or, the number of film and television on-demand and the on-demand duration of film and television assets also have a certain relationship with different types of users, so the data form of the above-mentioned key-value pair samples can be set to (user ID, (asset category, asset age, on-demand number, on-demand duration)).

[0057] Among them, based on the above key-value pair samples, a clustering algorithm can be used to calculate the classification information of the above user groups in different time periods. The clustering algorithm can be a k-means clustering algorithm, and its steps are to randomly select K key-value pairs as initial cluster centers, and then calculate the distance between each key-value pair sample and each initial cluster center, and assign each key-value pair sample to the initial cluster center closest to it; the initial cluster center and the key-value pair samples assigned to them represent a cluster, and each time a sample is assigned, the cluster center of the cluster is recalculated according to the existing key-value pairs in the cluster; this process will be repeated until no (or the minimum number) of key-value pair samples are reallocated to different clusters, or no (or the minimum number) of cluster centers change again, thereby finally generating the target cluster, taking each target cluster as a class, and then obtaining the above classification information. Alternatively, the clustering algorithm can also be DBSCAN (Density-Based Spatial Clustering of Applications with Noise), which is a relatively representative density-based clustering algorithm; the steps are to start with any key-value pair sample, find all nearby samples within the scanning radius (including the scanning radius) of the key-value pair sample, if the number of nearby samples is greater than or equal to the minimum number, then the key-value pair sample and its nearby samples form a cluster, and the key-value pair sample is marked as visited; then recursively, all samples in the cluster that are not marked as visited are processed in the same way, thereby expanding the cluster; finally, the target cluster is generated, and then the above-mentioned classification information is generated.

[0058] It can be understood that, since in application scenarios such as the same family or activity center, user groups usually interact with multimedia interactive devices through the same user ID, this embodiment divides the user group based on different time periods to obtain classification information; the classification information refers to the clustering values ​​obtained according to each key-value pair sample using a clustering algorithm, and the user group can be classified according to the size of the clustering value. For example, if the clustering value range is 0 to 10, then 0 to 3 represents children, 4 to 7 represents middle-aged people, and 8 to 10 represents the elderly; then a user portrait model of the user group can be constructed based on the above classification information. For example, when the classification information indicates that the clustering value range corresponding to the time period from 7 to 9 o'clock on the same day is 8 to 10, the clustering value range corresponding to the time period from 11 to 13 o'clock is 4 to 7, and the clustering value range corresponding to the time period from 17 to 18 o'clock is 0 to 3, then the user portrait model constructed based on the classification information is for the elderly user group from 7 to 9 o'clock on the same day, for the middle-aged user group from 11 to 13 o'clock, and for children from 17 to 18 o'clock.

[0059] Step S230: generating media resource recommendation results for the user group based on the user portrait model.

[0060] Among them, after obtaining the above-mentioned user portrait model, the above-mentioned processor can generate media asset recommendation results corresponding to different time periods based on the user portrait model, so as to recommend corresponding film and television assets on the above-mentioned multimedia interactive device in different time periods based on the media asset recommendation results, so that users in the above-mentioned different time periods can obtain their respective recommended assets through the recommendation algorithm, that is, different time periods can obtain recommended assets corresponding to user characteristics of multiple time periods. It should be noted that in the process of real-time interaction between the above-mentioned user group and the multimedia interactive device, since each on-demand behavior of the user will affect the center point of the clustering algorithm, the generated user portrait model can also change in real time, and finally the film and television assets recommended in different time periods will be continuously adjusted, thereby improving the user interaction experience.

[0061] Through the above steps S210 to S230, the user behavior data of the user group is clustered by a clustering algorithm, so as to enrich the user group portrait based on the clustering result, thereby subdividing the user group into corresponding categories, and being able to make detailed recommendations for the user group based on the divided categories, thereby solving the problem of low accuracy of media resource recommendations and realizing a high-precision media resource recommendation method based on user group portraits.

[0062] In some embodiments, a media asset recommendation method is provided. Figure 3 is a flowchart of another media resource recommendation method according to an embodiment of the present application. Figure 3 As shown, the process includes Figure 2 Step S210 and step S230 in the embodiment further include the following steps:

[0063] Step S310, obtaining a key-value pair sample according to the user behavior data, converting the key-value pair sample into a multi-dimensional coordinate, and obtaining at least two preset center points of all the multi-dimensional coordinates; wherein the multi-dimensional coordinates include the on-demand saturation.

[0064] Among them, after obtaining the above key-value pair samples, the key-value pair samples can be used as multidimensional coordinates for calculation. For example, a user ID has two on-demand behaviors. The first time, he watched a comedy asset at 6 o'clock, the asset age was 2008, and the on-demand saturation was 70%. The second time, he watched a comedy asset at 12 o'clock, the asset age was 2021, and the on-demand saturation was 80%; then the key-value pair samples corresponding to the two user behavior data can be converted into two multidimensional coordinates, namely (6, 12008, 0.7) and (12, 12021, 0.8). It should be noted that 12008 and 12021 in the above multidimensional coordinates are used to represent asset tags; the asset tag refers to the mapping value between the asset classification and the asset age. For example, if the asset classification of an asset is comedy and the asset age is 2008, then the mapped asset tag is 12008; the asset classification of another asset is also comedy, but the asset age is 2021, then the mapped asset tag is 12021, and so on. The above-mentioned on-demand saturation refers to the ratio of the user's on-demand duration for a certain asset to the total duration of the asset when the user on-demands the asset. For example, if the total duration of a movie is 40 minutes and the user's on-demand duration is 30 minutes, the on-demand saturation is 75%. Before clustering the multidimensional coordinates, K center points can be pre-set, and K are randomly selected from each multidimensional coordinate as the center points, where K is a positive integer greater than 1. For example, the staff can set 3 center points, that is, the user groups corresponding to each set-top box are divided into 3 categories of users.

[0065] Step S320, based on the on-demand saturation, the multidimensional coordinate is assigned to the center point to generate a first cluster, the first coordinate mean between all the multidimensional coordinates in the first cluster is calculated, and the first distance information between the coordinate mean and the center point is calculated; when the first distance information is greater than or equal to a preset distance threshold, the multidimensional coordinate is assigned to the first coordinate mean based on the on-demand saturation to generate a second cluster, the second coordinate mean between all the multidimensional coordinates in the second cluster is calculated, and the second distance information between the second coordinate mean and the first coordinate mean is calculated.

[0066] Specifically, the Euclidean distance calculation formula can be used to sequentially calculate the distance between each multidimensional coordinate containing the on-demand saturation and each center point, and the above clustering algorithm can be used to assign each multidimensional coordinate to the center point with the smallest distance to generate a cluster. Then, the first distance information between the mean of the first coordinate in the cluster and the corresponding center point is compared. If the first distance information is greater than or equal to the distance threshold, it means that the distance between the new centroid and the original centroid is large, and the algorithm still needs to be re-circulated for calculation, that is, the cluster center of the cluster is recalculated according to the existing multidimensional coordinates in the cluster; the preset threshold can be set to 5.

[0067] Step S330, when the second distance information is less than the distance threshold, obtain the classification information according to the coordinate mean, and obtain the user portrait model of the user group according to the classification information; wherein the coordinate mean refers to the average value of all the multidimensional coordinates in the cluster.

[0068] Specifically, when the second distance message is smaller than the distance threshold, it indicates that the distance between the new centroid and the original centroid is small, that is, the position of the recalculated particle does not change much, and the clustering algorithm converges in the region, thereby taking the center of the finally formed clustering cluster as the corresponding category in the above classification information.

[0069] Through the above steps S310 to S330, multiple center points are selected as initial clustering centers, and the classification information obtained by clustering is determined based on the above-mentioned on-demand saturation, so that the weight occupied by the on-demand duration is higher, avoiding the inability to accurately judge the preference of the user group when the number of on-demand times is large but the on-demand time is short, thereby effectively improving the accuracy of media resource recommendations.

[0070] In some embodiments, the multidimensional coordinates also include a user timestamp; the assigning of the multidimensional coordinates to the center point based on the on-demand saturation to generate the first cluster also includes the following steps: obtaining preset amplification information, and obtaining a target timestamp based on the amplification information and the user timestamp; assigning the multidimensional coordinates to the center point based on the target timestamp and the on-demand saturation to generate the first cluster.

[0071] Among them, in order to avoid the influence of randomly selecting a preset center point on the clustering result, the magnification factor to be magnified can be pre-set and stored as the above-mentioned magnification information; it can be understood that the magnification information can be continuously debugged through testing to improve accuracy. The above-mentioned processor can use the magnification information to magnify the above-mentioned user timestamp to the above-mentioned target timestamp before clustering, and then generate the above-mentioned first clustering cluster. For example, in a day, the user timestamp takes a value of 0 to 23h. When the above-mentioned magnification information indicates a magnification of 1000 times, the coordinate values ​​of the user timestamp on the time axis can be magnified 1000 times. Then, on the multi-dimensional coordinates, the distance between each time point is large, which is conducive to improving the accuracy of the clustering algorithm and further improving the accuracy of media resource recommendations.

[0072] In some embodiments, the assigning of the multi-dimensional coordinates to the center point based on the video-on-demand saturation to generate the first cluster further includes the following steps:

[0073] Step S322, when the on-demand saturation is less than the preset saturation threshold, the multidimensional coordinates corresponding to the on-demand saturation are used as the elimination points; the multidimensional coordinates are reallocated based on the on-demand saturation to generate the second cluster. The saturation threshold can be set in advance, for example, set to 0.3; when the on-demand saturation included in a certain multidimensional coordinate is lower than the saturation threshold, it means that the user's on-demand time is short at this time, and the corresponding multidimensional coordinates belong to the outliers in the clustering process, so further filtering can be performed on the abnormal outliers, and the calculated multidimensional coordinates with low on-demand saturation are used as the elimination points for elimination; then the remaining multidimensional coordinates after elimination are reallocated to generate a new cluster, thereby realizing the elimination and adjustment of abnormal points in the clustering process, and further improving the accuracy of media recommendation.

[0074] In some embodiments, the above-mentioned generation of the media resource recommendation result of the user group according to the user portrait model includes the following steps: inputting the user behavior data into a well-trained user preference model to obtain the user group preference type, and generating the media resource recommendation result of the user group according to the user group preference type and the user portrait model. Among them, the above-mentioned user preference model can be trained based on the historical user behavior data obtained in the historical time period based on the recommendation algorithm such as the alternating least squares (ALS) algorithm. Then the user behavior data in the above-mentioned real-time interaction process is input into the user preference model, and the above-mentioned user group preference type can be output to obtain the above-mentioned user group preference type, and then the media resource recommendation result of the above-mentioned user group is generated. For example, if the user group preference type calculated by the user preference model is action movies and cartoons, the user group preference type and the above-mentioned user portrait model are comprehensively analyzed to generate a media resource recommendation result of automatically recommending action movies to the user group from 7 to 9 o'clock, and recommending cartoons to the user group from 17 to 18 o'clock. Through the above-mentioned embodiment, the media resource recommendation result is generated in combination with the user group preference type and the user portrait model, so that the accuracy and efficiency of the media recommendation processing are higher.

[0075] In some embodiments, the above step S220 also includes the following steps: obtaining the video page service in the multimedia interactive device, and obtaining the preset tracking function on the video page service; using the tracking function, obtaining the user behavior data. Specifically, the tracking function is used to capture, process and send the user behavior data of the video page service interaction of the multimedia interactive device. The tracking method can send the user's behavior to the above processor in the form of a log, thereby realizing the data collection stage for the user behavior data.

[0076] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0077] This embodiment also provides a media resource recommendation device, which is applied to a multimedia interactive device. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the terms "module", "unit", "sub-unit", etc. can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0078] Figure 4 is a structural block diagram of a media resource recommendation device according to an embodiment of the present application, such as Figure 4 As shown, the device includes: an acquisition module 42, a classification module 44 and a generation module 46; the acquisition module 42 is used to obtain user behavior data of real-time interaction between a user group and the multimedia interactive device; the classification module 44 is used to obtain key-value pair samples based on the user behavior data, cluster the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtain a user portrait model of the user group according to the classification information; the generation module 46 is used to generate media resource recommendation results for the user group according to the user portrait model.

[0079] Through the above embodiment, the above classification module 44 clusters the user behavior data of the user group through a clustering algorithm, so as to enrich the user group portrait based on the clustering results, thereby subdividing the user group into corresponding categories, and can make detailed recommendations for the user group based on the divided categories, thereby solving the problem of low accuracy of media resource recommendations and realizing a high-precision media resource recommendation device based on user group portraits.

[0080] In some embodiments, the classification module 44 is further used to convert the key-value pair sample into a multidimensional coordinate and obtain at least two preset center points of all the multidimensional coordinates; wherein the multidimensional coordinates include on-demand saturation; the classification module 44 assigns the multidimensional coordinate to the center point based on the on-demand saturation to generate a first cluster, calculates the first coordinate mean between all the multidimensional coordinates in the first cluster, and calculates the first distance information between the coordinate mean and the center point; if the first distance information is greater than or equal to a preset distance threshold, the classification module 44 assigns the multidimensional coordinate to the first coordinate mean based on the on-demand saturation to generate a second cluster, calculates the second coordinate mean between all the multidimensional coordinates in the second cluster, and calculates the second distance information between the second coordinate mean and the first coordinate mean; if the second distance information is less than the distance threshold, the classification module 44 obtains the classification information according to the coordinate mean; wherein the coordinate mean refers to the average value of all the multidimensional coordinates in the cluster.

[0081] In some embodiments, the classification module 44 is further used to use the multidimensional coordinates corresponding to the on-demand saturation as a rejection point when the on-demand saturation is less than a preset saturation threshold; when taking out the rejection point, the classification module 44 reallocates the multidimensional coordinates based on the on-demand saturation to generate the second cluster.

[0082] In some embodiments, the multidimensional coordinates also include a user timestamp; the classification module 44 is also used to obtain preset magnification information, and obtain a target timestamp based on the magnification information and the user timestamp; the classification module 44 assigns the multidimensional coordinates to the center point based on the target timestamp and the on-demand saturation to generate the first cluster.

[0083] In some embodiments, the classification module 44 is further used to generate a corresponding user timestamp based on the user behavior data obtained at different time points; the classification module 44 parses and processes all the user behavior data to obtain the media viewing information corresponding to each user behavior data; the classification module 44 obtains the key-value pair sample based on the user timestamp and the media viewing information.

[0084] In some of the embodiments, the generation module 46 is further used to input the user behavior data into a well-trained user preference model to obtain a user group preference type, and generate a media resource recommendation result for the user group based on the user group preference type and the user portrait model.

[0085] In some embodiments, the acquisition module 42 is also used to acquire the video page service in the multimedia interactive device and acquire the preset embedding function on the video page service; the acquisition module 42 acquires the user behavior data using the embedding function.

[0086] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0087] This embodiment also provides a media resource recommendation system. Figure 5 is a structural block diagram of a media resource recommendation system according to an embodiment of the present application, such as Figure 5 As shown, the system includes: a multimedia interaction device 102, a transmission device 52 and a server device 104; wherein the multimedia interaction device 104 is connected to the server device via the transmission device; the multimedia interaction device 104 is used to generate user behavior data for interacting with a user group; the transmission device 52 is used to obtain the user behavior data from the multimedia interaction device and send the user behavior data to the server device; the server device 104 is used to obtain the user behavior data; the server device 104 obtains key-value pair samples based on the user behavior data, clusters the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtains a user portrait model of the user group based on the classification information; the server device 104 generates media resource recommendation results for the user group based on the user portrait model.

[0088] Through the above embodiment, the above server device 104 clusters the user behavior data of the user group through a clustering algorithm, so as to enrich the user group portrait based on the clustering result, thereby subdividing the user group into corresponding categories, and can make detailed recommendations for the user group based on the divided categories, thereby solving the problem of low accuracy of media resource recommendations and realizing a high-precision media resource recommendation system based on user group portraits.

[0089] In some of the embodiments, the server device 104 is further used to convert the key-value pair sample into a multidimensional coordinate, and obtain at least two preset center points of all the multidimensional coordinates; wherein the multidimensional coordinates include on-demand saturation; the server device 104 assigns the multidimensional coordinate to the center point based on the on-demand saturation to generate a first cluster, calculates the first coordinate mean between all the multidimensional coordinates in the first cluster, and calculates the first distance information between the coordinate mean and the center point; if the first distance information is greater than or equal to a preset distance threshold, the server device 104 assigns the multidimensional coordinate to the first coordinate mean based on the on-demand saturation to generate a second cluster, calculates the second coordinate mean between all the multidimensional coordinates in the second cluster, and calculates the second distance information between the second coordinate mean and the first coordinate mean; if the second distance information is less than the distance threshold, the server device 104 obtains the classification information according to the coordinate mean; wherein the coordinate mean refers to the average value of all the multidimensional coordinates in the cluster.

[0090] In some of the embodiments, the server device 104 is further configured to use the multidimensional coordinates corresponding to the on-demand saturation as a rejection point when the on-demand saturation is less than a preset saturation threshold; when taking out the rejection point, the server device 104 reallocates the multidimensional coordinates based on the on-demand saturation to generate the second cluster.

[0091] In some embodiments, the multidimensional coordinates also include a user timestamp; the server device 104 is further used to obtain preset amplification information, and obtain a target timestamp based on the amplification information and the user timestamp; the server device 104 assigns the multidimensional coordinates to the center point based on the target timestamp and the on-demand saturation to generate the first cluster.

[0092] In some of the embodiments, the server device 104 is further used to generate a corresponding user timestamp based on the user behavior data obtained at different time points; the server device 104 parses and processes all the user behavior data to obtain the media viewing information corresponding to each user behavior data; the server device 104 obtains the key-value pair sample based on the user timestamp and the media viewing information.

[0093] In some of the embodiments, the server device 104 is further used to input the user behavior data into a well-trained user preference model to obtain a user group preference type, and generate a media resource recommendation result for the user group based on the user group preference type and the user portrait model.

[0094] In some of the embodiments, the server device 104 is also used to obtain the video page service in the multimedia interactive device and obtain the preset embedding function on the video page service; the server device 104 uses the embedding function to obtain the user behavior data.

[0095] In some embodiments, a computer device is provided, which may be a server. Figure 6 is a structural diagram of the inside of a computer device according to an embodiment of the present application, such as Figure 6 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store user portrait models. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above-mentioned media resource recommendation method is implemented.

[0096] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0097] This embodiment further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0098] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0099] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0100] S1, obtaining user behavior data of real-time interaction between a user group and the multimedia interactive device.

[0101] S2, obtaining key-value pair samples according to the user behavior data, clustering the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtaining a user portrait model of the user group according to the classification information.

[0102] S3: Generate media resource recommendation results for the user group based on the user portrait model.

[0103] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0104] In addition, in combination with the media resource recommendation method in the above embodiment, the present application embodiment can provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, any one of the media resource recommendation methods in the above embodiment is implemented.

[0105] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0106] Those skilled in the art should understand that the technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0107] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A media resource recommendation method, characterized in that: Applied to a multimedia interactive device, the method comprises: Acquire user behavior data of real-time interaction between a user group and the multimedia interactive device; Acquire key-value pair samples according to the user behavior data, cluster the key-value pair samples according to a clustering algorithm to obtain classification information of the user group in different time periods, and obtain a user portrait model of the user group according to the classification information; wherein, acquiring key-value pair samples according to the user behavior data includes: generating corresponding user timestamps according to the user behavior data acquired at different time points; parsing and processing all the user behavior data to obtain media viewing information corresponding to each of the user behavior data; acquiring the key-value pair samples according to the user timestamp and the media viewing information; clustering the key-value pair samples according to the clustering algorithm to obtain classification information of the user group in different time periods includes: converting the key-value pair samples into multidimensional coordinates, and acquiring at least one of all the multidimensional coordinates. two preset center points; wherein the multidimensional coordinates include on-demand saturation; based on the on-demand saturation, the multidimensional coordinates are assigned to the center point to generate a first cluster, a first coordinate mean between all the multidimensional coordinates in the first cluster is calculated, and first distance information between the coordinate mean and the center point is calculated; when the first distance information is greater than or equal to a preset distance threshold, the multidimensional coordinates are assigned to the first coordinate mean to generate a second cluster based on the on-demand saturation, a second coordinate mean between all the multidimensional coordinates in the second cluster is calculated, and second distance information between the second coordinate mean and the first coordinate mean is calculated; when the second distance information is less than the distance threshold, the classification information is obtained according to the coordinate mean; Generate media resource recommendation results for the user group based on the user portrait model.

2. The media resource recommendation method according to claim 1, characterized in that: The allocating the multidimensional coordinates to the first coordinate mean based on the on-demand saturation to generate a second cluster comprises: When the on-demand saturation is less than a preset saturation threshold, the multi-dimensional coordinates corresponding to the on-demand saturation are used as the elimination points; When the rejection points are removed, the multi-dimensional coordinates are reallocated based on the on-demand saturation to generate the second cluster.

3. The media resource recommendation method according to claim 1, characterized in that: The multidimensional coordinates further include a user timestamp; and assigning the multidimensional coordinates to the central point based on the on-demand saturation to generate a first cluster includes: Acquire preset magnification information, and acquire a target timestamp according to the magnification information and the user timestamp; The multi-dimensional coordinates are assigned to the center point based on the target timestamp and the on-demand saturation to generate the first cluster.

4. The media resource recommendation method according to claim 1, characterized in that: Generating the media resource recommendation result for the user group according to the user portrait model includes: The user behavior data is input into a well-trained user preference model to obtain a user group preference type, and a media resource recommendation result for the user group is generated based on the user group preference type and the user portrait model.

5. The media resource recommendation method according to any one of claims 1 to 3, characterized in that: The acquiring of user behavior data of the interaction between the user group and the multimedia interactive device comprises: Obtaining a video page service in the multimedia interactive device, and obtaining a preset embedding function on the video page service; The user behavior data is obtained by utilizing the embedding function.

6. A media resource recommendation device, characterized in that: Applied to multimedia interactive equipment, the device comprises: an acquisition module, a classification module and a generation module; The acquisition module is used to acquire user behavior data of real-time interaction between a user group and the multimedia interactive device; The classification module is used to obtain key-value pair samples according to the user behavior data, cluster the key-value pair samples according to the clustering algorithm to obtain the classification information of the user group in different time periods, and obtain the user portrait model of the user group according to the classification information; wherein, obtaining the key-value pair samples according to the user behavior data includes: generating corresponding user timestamps according to the user behavior data obtained at different time points; parsing and processing all the user behavior data to obtain the media viewing information corresponding to each of the user behavior data; obtaining the key-value pair samples according to the user timestamp and the media viewing information; clustering the key-value pair samples according to the clustering algorithm to obtain the classification information of the user group in different time periods includes: converting the key-value pair samples into multi-dimensional coordinates, and obtaining all the multi-dimensional coordinates. at least two preset center points are marked; wherein the multidimensional coordinates include on-demand saturation; based on the on-demand saturation, the multidimensional coordinates are assigned to the center point to generate a first cluster, a first coordinate mean between all the multidimensional coordinates in the first cluster is calculated, and first distance information between the coordinate mean and the center point is calculated; when the first distance information is greater than or equal to a preset distance threshold, the multidimensional coordinates are assigned to the first coordinate mean to generate a second cluster based on the on-demand saturation, a second coordinate mean between all the multidimensional coordinates in the second cluster is calculated, and second distance information between the second coordinate mean and the first coordinate mean is calculated; when the second distance information is less than the distance threshold, the classification information is obtained according to the coordinate mean; The generating module is used to generate media resource recommendation results for the user group according to the user portrait model.

7. A media resource recommendation system, characterized in that: The system comprises: a multimedia interactive device, a transmission device and a server device; wherein the multimedia interactive device is connected to the server device via the transmission device; The multimedia interaction device is used to generate user behavior data for interaction with a user group; The transmission device is used to obtain the user behavior data from the multimedia interaction device and send the user behavior data to the server device; The server device is used to execute the media asset recommendation method according to any one of claims 1 to 5.

8. An electronic device comprising a memory and a processor, characterized in that: The memory stores a computer program, and the processor is configured to run the computer program to execute the media asset recommendation method according to any one of claims 1 to 5.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the media resource recommendation method according to any one of claims 1 to 5 when running.

Images