Information pushing method and device based on user interest portrait, equipment and medium

Abstract

The invention relates to the technical field of natural languages, and provides an information pushing method, device and equipment based on a user interest portrait and a computer readable storage medium. Historical structured behavior data and historical unstructured behavior data are analyzed; explicit interest preference information is obtained based on the historical structured behavior data, and implicit interest preference information is obtained based on the historical unstructured behavior data. Through two dimensions of dominant interest preference information and implicit interest preference information, comprehensive and diversified interest analysis is performed on the user, so that a more accurate interest recognition analysis result is provided, and a target interest portrait of the target user is obtained. And through the user interest portrait, personalized and targeted push information conforming to the user interest is screened out, so that the user experience is further improved.

Description

Technical Field

[0001] The present invention relates to the field of natural language, and in particular to a method, device, computer equipment and computer-readable storage medium for information push based on user interest profiles. Background Art

[0002] With the development of big data, various Internet service platforms aim to provide users with convenient information content services by providing a variety of personalized services. In the development and operation of personalized services, determining the user's user profile is an essential step. User profile refers to a labeled user model abstracted from the user's attributes, user preferences, living habits, user behavior and other interest data information. That is to label the user, and the label is a highly refined feature identifier obtained by analyzing the user information. By labeling, some highly generalized and easy-to-understand features can be used to describe the user, which makes it easier for the machine to understand the user and facilitates computer processing. However, the current information push model is usually used to identify common entity categories (such as names, place names, organization names, etc.), but cannot accurately identify entity information in specific fields (such as movie types, singers, actors, etc.), resulting in the current information push model's low recognition accuracy when processing entertainment texts and low information push accuracy.

[0003] Therefore, how to solve the low accuracy of information push in the current information push model has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] The main purpose of the present invention is to provide an information push method, device, equipment and computer-readable storage medium based on user interest portraits, aiming to solve the current technical problems of low efficiency in processing reconciliation differences and low accuracy of processing results.

[0005] To achieve the above object, the present invention provides an information push method based on user interest portrait, the information push method comprising the following steps:

[0006] Acquire historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data;

[0007] Determining the explicit interest preference information of the target user based on the historical structured behavior data, and determining the implicit interest preference information of the target user based on the historical unstructured behavior data;

[0008] Based on the explicit interest preference information and the implicit interest preference information, generating a target interest portrait of the target user;

[0009] Based on the target interest portrait, the recommendation information corresponding to the target user is filtered out, so as to push personalized content to the target user based on the recommendation information.

[0010] In addition, to achieve the above-mentioned purpose, the present invention also provides an information push device based on user interest portrait, and the information push device based on user interest portrait includes:

[0011] A behavior data acquisition module, used to acquire historical behavior data of a target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data;

[0012] A preference information determination module, used to determine the explicit interest preference information of the target user based on the historical structured behavior data, and to determine the implicit interest preference information of the target user based on the historical unstructured behavior data;

[0013] An interest portrait generation module, used to generate a target interest portrait of the target user based on the explicit interest preference information and the implicit interest preference information;

[0014] The recommendation information push module is used to filter out the recommendation information corresponding to the target user based on the target interest portrait, so as to push personalized content to the target user based on the recommendation information.

[0015] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer device, which includes a processor, a memory, and an information push program based on user interest portraits stored on the memory and executable by the processor, wherein when the information push program based on user interest portraits is executed by the processor, the steps of the information push method based on user interest portraits as described above are implemented.

[0016] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which is stored an information push program based on user interest portraits, wherein when the information push program based on user interest portraits is executed by a processor, the steps of the information push method based on user interest portraits as described above are implemented.

[0017] The present application provides an information push method based on user interest portrait, the method obtains the historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data; determines the explicit interest preference information of the target user based on the historical structured behavior data, and determines the implicit interest preference information of the target user based on the historical unstructured behavior data; generates the target interest portrait of the target user based on the explicit interest preference information and the implicit interest preference information; based on the target interest portrait, screens out the recommended information corresponding to the target user, so as to push personalized content to the target user based on the recommended information. Through the above-mentioned method, the present application analyzes the historical structured behavior data and the historical unstructured behavior data respectively, and obtains the explicit interest preference information based on the historical structured behavior data, and obtains the implicit interest preference information based on the historical unstructured behavior data. Through the two dimensions of explicit interest preference information and implicit interest preference information, a comprehensive and diversified interest analysis of the user is performed, thereby providing a more accurate interest identification analysis result, and obtaining the target interest portrait of the target user. Then, through user interest portraits, personalized and targeted push information that matches the user's interests is screened out, thereby further improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a first embodiment of the information push method based on user interest portraits of the present invention;

[0019] Figure 2 This is a flow chart of a second embodiment of the information push method based on user interest portraits of the present invention;

[0020] Figure 3 It is a flowchart of a third embodiment of the information push method based on user interest portrait of the present invention;

[0021] Figure 4 A schematic diagram of an information push device based on user interest portraits according to the present invention;

[0022] Figure 5 The figure is a schematic diagram of the hardware structure of a computer device involved in an embodiment of the present invention.

[0023] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0026] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0027] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0028] The information push method based on user interest portrait involved in the embodiment of the present invention is mainly applied to computer equipment, which can be a PC, a portable computer, a mobile terminal and other equipment with display and processing functions.

[0029] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0030] Reference Figure 1 , Figure 1 This is a flow chart of the first embodiment of the information push method based on user interest profile of the present invention.

[0031] like Figure 1 As shown, an embodiment of the present invention provides an information push method based on user interest portraits, and the information push method based on user interest portraits includes steps S10 to S40.

[0032] Step S10: Acquire historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data;

[0033] At present, there are some applications in the market for entertainment interest analysis, especially in recommendation systems, which analyze users' historical behavior data and infer users' interest preferences based on the analysis results. However, most of the above analysis technologies rely on the analysis of structured data, and their processing capabilities for unstructured text are relatively limited. There is a lack of effective natural language processing tools to analyze the rich information in unstructured text, so it is impossible to fully explore the user's interest points expressed in unstructured text.

[0034] In order to solve the above problems, this embodiment analyzes the historical structured behavior data and the historical unstructured behavior data respectively, and obtains explicit interest preference information based on the historical structured behavior data, and obtains implicit interest preference information based on the historical unstructured behavior data. Through the two dimensions of explicit interest preference information and implicit interest preference information, a comprehensive and diversified interest analysis of users is performed, thereby providing more accurate interest identification analysis results and obtaining the target interest portrait of the target user. Then, through the user interest portrait, personalized and targeted push information that meets the user's interests is screened out, thereby further improving the user experience.

[0035] Specifically, the target users' historical structured behavior data, i.e., browsing history, click counts, ratings, etc. on the application or platform, are obtained respectively, and the target users' historical unstructured behavior data, such as user comments on social media, movie reviews, song reviews, etc. are obtained.

[0036] Step S20: determining the explicit interest preference information of the target user based on the historical structured behavior data, and determining the implicit interest preference information of the target user based on the historical unstructured behavior data;

[0037] In this embodiment, explicit interest preference information is determined through analysis of historical structured behavior data, such as the user's movie preferences, TV series preferences, music type preferences, favorite singers and actors, and data types with high views, high clicks, and high scores in the browsing history, etc., to mine the user's explicit interest preference information. Then, the natural language model is used to perform intent recognition and key information extraction on historical unstructured behavior data such as user comments on social media, movie reviews, and song reviews, to mine the user's implicit interest preference information.

[0038] Step S30: generating a target interest portrait of the target user based on the explicit interest preference information and the implicit interest preference information;

[0039] In this embodiment, the user interest portrait can be comprehensively analyzed according to user attributes and user behavior attributes. User attributes generally include natural attributes, social relationships, location information, device attributes, interests and hobbies, etc.; user behavior attributes include interest preference categories, active frequency, product preferences, product drivers, behavioral habits, product consumption, etc.

[0040] Among them, natural attributes are the user's own status information attributes, such as name, age, height, telephone number, occupation, etc.; social relations are the sum of all social relations generated by people on the basis of practical activities, such as: economic status, family status, social status, political religion, geographical location, values, etc.; location information is the geographical location of the user, such as the country, region, city, etc.; device attributes are the information of the mobile phone model, computer model, and other electronic equipment used by the user; interest preference categories are people's inner psychological state and external behavioral expression of non-material objects, which is an instinctive preference from the heart and has no necessary relationship with material, such as: interest preference, color preference, etc.; product category is the category of entertainment, information, and tools that the product used belongs to; active frequency is the frequency of using the product; product preference is the favorite way of using the product and the usage scenario; product drive is the reason and behavior that drives the use of the product; behavioral habits are the activity and loyalty to a certain behavioral preference, website, etc.; product consumption is based on consumption cycle, consumption amount, consumption amount, consumption interval, such as shopping type and consumption level.

[0041] Based on the explicit interest preference information and the network behavior data, user preference data, transaction data, and life cycle corresponding to the implicit interest preference information, users and behavior variables are matched one by one, thereby identifying the differentiated behavior patterns of users:

[0042] Behavior: What the user did, that is, what was the user's behavioral path, the frequency and workload of a certain action, whether there were unnecessary steps, or areas that could be automated, to determine the user's behavior and motivation;

[0043] Attitude: The user's attitude and emotions towards the product's field and the technology it uses, and the behavior related to the product;

[0044] Capabilities: the user’s education and training, learning ability, whether something is easy for the user to do, and whether the user is more willing to perform a certain behavior;

[0045] Motivation: The reason why users get involved in the product area, what makes users decide, what hinders users, and what attracts users to check;

[0046] Skills: The user's skills relevant to viewing the data.

[0047] Based on the behavioral variables of each user, and then based on the static information of each user (such as age, geographical location, relative income and professional title, etc.), the portrait details are added to generate the target interest portrait.

[0048] Step S40: based on the target interest portrait, filter out the recommendation information corresponding to the target user, so as to push personalized content to the target user based on the recommendation information.

[0049] In this embodiment, based on the target interest portrait, products or viewing information that are in line with the target user's interests are screened out, thereby achieving personalized content push for the target user.

[0050] It can be understood that the solution provided in this embodiment is not only applicable to entertainment content recommendation and interest analysis, but can also be applied to application scenarios such as advertising recommendation, news push, and social media analysis.

[0051] This embodiment provides an information push method based on user interest portrait, the method obtains the historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data; determines the explicit interest preference information of the target user based on the historical structured behavior data, and determines the implicit interest preference information of the target user based on the historical unstructured behavior data; generates the target interest portrait of the target user based on the explicit interest preference information and the implicit interest preference information; based on the target interest portrait, screens out the recommended information corresponding to the target user, so as to push personalized content to the target user based on the recommended information. Through the above-mentioned method, the present application analyzes the historical structured behavior data and the historical unstructured behavior data respectively, and obtains the explicit interest preference information based on the historical structured behavior data, and obtains the implicit interest preference information based on the historical unstructured behavior data. Through the two dimensions of explicit interest preference information and implicit interest preference information, a comprehensive and diversified interest analysis is performed on the user, thereby providing a more accurate interest identification analysis result and obtaining the target interest portrait of the target user. Then, through user interest portraits, personalized and targeted push information that matches the user's interests is screened out, thereby further improving the user experience.

[0052] Reference Figure 2 , Figure 2 This is a flow chart of the second embodiment of the information push method based on user interest portrait of the present invention.

[0053] Based on the above Figure 1In the embodiment shown, in this embodiment, the historical unstructured behavior data includes at least one of user comment data, audio and video preference data, movie review data, and song evaluation data, and the historical structured behavior data includes at least one of information browsing records, information click counts, or rating data. Step S20 specifically includes:

[0054] Step S21, identifying the proper name entity information in the user comment data, audio and video preference data, movie review data, and song evaluation data through a target named entity recognition model to obtain a proper name entity recognition result;

[0055] Step S22, based on the information relationship extraction model and the entity recognition result of the proper name, unstructured text analysis is performed on the user comment data, audio and video preference data, movie review and song evaluation data to obtain the implicit interest preference information;

[0056] Step S23, based on the interest preference prediction model, perform interest preference prediction on the information browsing records, information click times and rating data to generate the explicit interest preference information.

[0057] Currently, there is a lack of dedicated entity recognition models in the entertainment field. For example, the general named entity recognition (NER) model is usually used to identify common entity categories (such as names of people, places, and organizations), but cannot accurately identify entity information in specific fields (such as movie types, singers, and actors). Therefore, the target interest analysis model not only has great deficiencies in processing user entertainment interest recognition in unstructured texts, but also has low accuracy when processing entertainment texts, making it difficult to effectively identify users' real interests.

[0058] In order to solve the above problems, this embodiment provides a dedicated entity recognition model (i.e., a target named entity recognition model) in terms of accuracy, targeting, and model customization, so that the system can more accurately identify users' entertainment interests from unstructured texts, reducing the false positive rate and missed positive rate.

[0059] Specifically, training samples are generated by crawling Douban movie, TV series, song data and other data. The spaCy and BERT models are trained with the training samples to generate a target named entity recognition (NER) model dedicated to the entertainment field. This model identifies entity information of multiple dimensions related to the user's entertainment interests from unstructured text, including movie types, TV types, song types, singers, actors, etc. The target named entity recognition model is used to conduct in-depth analysis of the unstructured behavior data generated by users, thereby mining more potential user interests, especially the implicit interest data expressed by users on social media, comment areas and other platforms, thereby providing more comprehensive interest analysis results.

[0060] After obtaining the entity recognition result of the proper name through the target named entity recognition model, based on the information relationship extraction model, combined with the entity recognition result of the proper name, the unstructured behavioral data such as the user comment data, audio and video preference data, movie review and song evaluation data are subjected to intent recognition and key information extraction to obtain the implicit interest preference information.

[0061] Based on the interest preference prediction model, explicit interest mining is performed on the information browsing records, information click times and rating data to predict explicit interest preference information.

[0062] Through the above method, this embodiment accurately identifies entity information of multiple dimensions related to the user's entertainment interests from historical structured behavior data and historical unstructured behavior data, thereby improving the accuracy and pertinence of interest identification. In addition, this embodiment not only identifies a single entertainment interest dimension, but also analyzes multiple entertainment interest dimensions at the same time, such as the user's movie preferences, TV series preferences, music type preferences, favorite singers and actors, etc. Through the analysis of multi-dimensional interest data, a comprehensive interest portrait of the user is generated, and personalized recommendations and content push are made to the user more accurately through the comprehensive interest portrait.

[0063] Reference Figure 3 , Figure 3 This is a flow chart of the third embodiment of the information push method based on user interest portrait of the present invention.

[0064] Based on the above Figure 2 In the embodiment shown, in this embodiment, before step S21, the following steps are further included:

[0065] Step S01, crawling entertainment data related to the target user based on a web crawler, wherein the entertainment data includes at least one of movie related information, TV series related information, short video related information, song information, singer information and actor information;

[0066] Step S02: training an initial language processing model based on the entertainment data to generate the target named entity recognition model, so as to perform entity recognition on proper names in the entertainment data through the target named entity recognition model.

[0067] Furthermore, the training of the initial language processing model based on the entertainment data to generate the target named entity recognition model specifically includes:

[0068] Acquire the entertainment data and entity annotation data corresponding to the entertainment data;

[0069] Based on the entertainment data and the entity annotation data, performing entity extraction training on the initial language processing model to extract entity words corresponding to the entertainment data;

[0070] The entity words corresponding to the entertainment data are compared with the entity standard data, and according to the comparison result, the initial language processing model is iteratively trained until the target named entity recognition model is generated.

[0071] Furthermore, the training of the initial language processing model according to the comparison result specifically includes:

[0072] In the case where there is an abnormal comparison result, determining the entity word of the abnormal comparison result, and determining the abnormal type of the entity word of the abnormal comparison result;

[0073] Based on the abnormal type of the entity word, the entertainment data is supplemented and updated, or the entertainment data corresponding to the entity word of the abnormal result is deleted.

[0074] Furthermore, the training of the initial language processing model according to the comparison result specifically includes:

[0075] In the absence of abnormal comparison results, the training of the initial language processing model is stopped and the target named entity recognition model is generated.

[0076] In this embodiment, the movie-related information, TV series-related information, short video-related information, song information, singer information and actor information browsed by the target user are crawled based on a web crawler. Then, according to a preset data format, such as movie+type, TV series+type or song+singer, the data is extracted to obtain entertainment data as model training data. Based on the entertainment data, an initial language processing model (such as a spaCy or BERT model) is trained to generate the target named entity recognition model, so as to perform entity recognition on the proper names in the entertainment data through the target named entity recognition model.

[0077] For example:

[0078] Bidding words: refers to the keywords obtained through bidding in search engine optimization (SEO) to improve the ranking of the website in the search engine results;

[0079] Digging for traffic: Optimizing website content and structure by analyzing user needs and behaviors to attract more visits;

[0080] Registration big data and WHOIS big data: involving the registration information and related data analysis of Internet domain names, used to analyze user needs and optimize online marketing strategies;

[0081] High-frequency words, question words, and industry roots: used to reflect hot topics and user concerns in the industry;

[0082] Controlling comments: On social media, through actions such as likes, replies, and reports, the direction of the comment area is controlled, so that positive comments become popular and negative comments are suppressed;

[0083] Raw photos: Unretouched photos of celebrities, usually taken by fans or fans, reflecting the real appearance of the celebrities;

[0084] Fanjie: refers to the manager and operator of an idol or star's "support station", who is mainly responsible for posting relevant information and activity photos of the idol;

[0085] Fan circle: The abbreviation of fan circle, which is a transliteration of the English word "fans" and refers to the fan group;

[0086] Killing part: In a singing group, it refers to the most impressive or most popular singing part of a song.

[0087] When the entertainment data is crawled, the entity data in the entertainment data is annotated to obtain entity annotated data. Based on the entertainment data and its entity annotated data, the initial language processing model is trained for entity extraction. That is, the entity words corresponding to the entertainment data are extracted through the model in training. The entity words corresponding to the entertainment data are compared with the entity standard data, and according to the comparison results, the initial language processing model is iteratively trained, that is:

[0088] In the case where there is an abnormal comparison result, determining the entity word of the abnormal comparison result, and determining the abnormal type of the entity word of the abnormal comparison result;

[0089] Based on the abnormal type of the entity word, the entertainment data is supplemented and updated, or the entertainment data corresponding to the entity word of the abnormal result is deleted.

[0090] In the absence of abnormal comparison results, the training of the initial language processing model is stopped and the target named entity recognition model is generated.

[0091] Therefore, when it is determined that the existing entity words are not extracted, the data and the entities to be extracted are used as supplementary training data; when the entity words are not fully extracted or extracted incorrectly, the erroneous training data are eliminated; if there are ambiguous entity words that are extracted inaccurately, the training data is supplemented based on the ambiguous entity words, and then the model is iteratively trained until the target named entity recognition model is generated.

[0092] Through the above-mentioned method, this embodiment continuously trains and optimizes the target named entity recognition model by continuously acquiring new training data, so that it can adapt to the ever-changing user interests and newly emerging entertainment content. Through this iterative optimization training mechanism, the recognition accuracy and practicality of the target named entity recognition model in long-term applications are ensured. The information push method based on user interest portrait provided by this embodiment includes but is not limited to social media analysis, entertainment content recommendation, and other fields. Therefore, this embodiment solves the shortcomings of traditional information push methods in unstructured text processing, and provides a more intelligent and accurate user entertainment interest identification and analysis method, which can not only analyze the user's single interest point, but also cover multiple dimensions of interest point data at the same time, and provide more comprehensive and diversified interest analysis results. Thus, through accurate interest identification and analysis, users are provided with more personalized and targeted content recommendations to improve user experience and satisfaction. And it is continuously optimized with the changes in user behavior and entertainment content, so that the analysis results always remain efficient and accurate.

[0093] See also Figure 4 , Figure 4 This is a functional module diagram of an information push device based on user interest portrait provided in an embodiment of the present application.

[0094] like Figure 4 As shown, the information push device 500 based on user interest portrait includes:

[0095] A behavior data acquisition module 10 is used to acquire the historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data;

[0096] A preference information determination module 20, configured to determine the explicit interest preference information of the target user based on the historical structured behavior data, and to determine the implicit interest preference information of the target user based on the historical unstructured behavior data;

[0097] An interest portrait generating module 30, configured to generate a target interest portrait of the target user based on the explicit interest preference information and the implicit interest preference information;

[0098] The recommendation information push module 40 is used to filter out the recommendation information corresponding to the target user based on the target interest portrait, so as to push personalized content to the target user based on the recommendation information.

[0099] Furthermore, the historical unstructured behavior data includes at least one of user comment data, audio and video preference data, movie review data, and song evaluation data. The preference information determination module 20 specifically includes:

[0100] A proper name recognition unit, used to recognize proper name entity information in the user comment data, audio and video preference data, movie review data, and song evaluation data through a target named entity recognition model to obtain a proper name entity recognition result;

[0101] The implicit preference determination unit is used to perform unstructured text analysis on the user comment data, audio and video preference data, movie review and song evaluation data based on the information relationship extraction model and the entity recognition result of the proper name to obtain the implicit interest preference information.

[0102] Furthermore, the preference information determination module 20 further includes:

[0103] An entertainment data crawling unit, configured to crawl entertainment data related to the target user based on a web crawler, wherein the entertainment data includes at least one of movie-related information, TV series-related information, short video-related information, song information, singer information, and actor information;

[0104] The recognition model training unit is used to train the initial language processing model based on the entertainment data to generate the target named entity recognition model, so as to perform entity recognition on the proper names in the entertainment data through the target named entity recognition model.

[0105] Furthermore, the recognition model training unit specifically includes:

[0106] An entertainment data standard subunit, used to obtain the entertainment data and entity annotation data corresponding to the entertainment data;

[0107] An entity word extraction subunit, configured to perform entity extraction training on the initial language processing model based on the entertainment data and the entity annotation data, and extract entity words corresponding to the entertainment data;

[0108] The recognition model training subunit is used to compare the entity words corresponding to the entertainment data with the entity standard data, and iteratively train the initial language processing model according to the comparison result until the target named entity recognition model is generated.

[0109] Furthermore, the recognition model training subunit is also used for:

[0110] In the case where there is an abnormal comparison result, determining the entity word of the abnormal comparison result, and determining the abnormal type of the entity word of the abnormal comparison result;

[0111] Based on the abnormal type of the entity word, the entertainment data is supplemented and updated, or the entertainment data corresponding to the entity word of the abnormal result is deleted.

[0112] Furthermore, the recognition model training subunit is also used for:

[0113] In the absence of abnormal comparison results, the training of the initial language processing model is stopped and the target named entity recognition model is generated.

[0114] Furthermore, the historical structured behavior data includes at least one of information browsing records, information click times or rating data, and the preference information determination module 20 further includes:

[0115] The explicit preference determination unit is used to perform interest preference prediction on the information browsing records, information click times and rating data based on the interest preference prediction model to generate the explicit interest preference information.

[0116] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0117] The above-mentioned device can be implemented in the form of a computer program. Figure 5 Runs on the computer device shown.

[0118] See also Figure 5 , Figure 5 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a server.

[0119] See also Figure 5 The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0120] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any information push method based on the user interest profile.

[0121] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0122] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any information push method based on the user interest portrait.

[0123] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 5The 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.

[0124] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0125] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0126] Acquire historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data;

[0127] Determining the explicit interest preference information of the target user based on the historical structured behavior data, and determining the implicit interest preference information of the target user based on the historical unstructured behavior data;

[0128] Based on the explicit interest preference information and the implicit interest preference information, generating a target interest portrait of the target user;

[0129] Based on the target interest portrait, the recommendation information corresponding to the target user is filtered out, so as to push personalized content to the target user based on the recommendation information.

[0130] In one embodiment, the historical unstructured behavior data includes at least one of user comment data, audio and video preference data, movie review data, and song evaluation data, and the processor is used to run a computer program stored in the memory to implement the following steps:

[0131] Using a target named entity recognition model, identifying the proper name entity information in the user comment data, video preference data, movie review data, and song evaluation data, and obtaining a proper name entity recognition result;

[0132] Based on the information relationship extraction model and the entity recognition results of the proper names, unstructured text analysis is performed on the user review data, audio and video preference data, movie review and song evaluation data to obtain the implicit interest preference information.

[0133] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0134] Crawling entertainment data related to the target user based on a web crawler, wherein the entertainment data includes at least one of movie-related information, TV series-related information, short video-related information, song information, singer information, and actor information;

[0135] The initial language processing model is trained based on the entertainment data to generate the target named entity recognition model, so as to perform entity recognition on the proper names in the entertainment data through the target named entity recognition model.

[0136] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0137] Acquire the entertainment data and entity annotation data corresponding to the entertainment data;

[0138] Based on the entertainment data and the entity annotation data, performing entity extraction training on the initial language processing model to extract entity words corresponding to the entertainment data;

[0139] The entity words corresponding to the entertainment data are compared with the entity standard data, and according to the comparison result, the initial language processing model is iteratively trained until the target named entity recognition model is generated.

[0140] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0141] In the case where there is an abnormal comparison result, determining the entity word of the abnormal comparison result, and determining the abnormal type of the entity word of the abnormal comparison result;

[0142] Based on the abnormal type of the entity word, the entertainment data is supplemented and updated, or the entertainment data corresponding to the entity word of the abnormal result is deleted.

[0143] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0144] In the absence of abnormal comparison results, the training of the initial language processing model is stopped and the target named entity recognition model is generated.

[0145] In one embodiment, the historical structured behavior data includes at least one of information browsing records, information click counts, or rating data, and the processor is used to run a computer program stored in the memory to implement the following steps:

[0146] Based on the interest preference prediction model, interest preference prediction is performed on the information browsing records, information click times and rating data to generate the explicit interest preference information.

[0147] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any one of the information push methods based on user interest portraits provided in the embodiments of the present application.

[0148] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (Smart Medi a Card, SMC), a secure digital (Secure Digital, SD) card, a flash memory card (Flash Card), etc., equipped on the computer device.

[0149] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for pushing information based on user interest portrait, characterized in that: The information push method comprises the following steps: Acquire historical behavior data of the target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data; Determining the explicit interest preference information of the target user based on the historical structured behavior data, and determining the implicit interest preference information of the target user based on the historical unstructured behavior data; Based on the explicit interest preference information and the implicit interest preference information, generating a target interest portrait of the target user; Based on the target interest portrait, the recommendation information corresponding to the target user is filtered out, so as to push personalized content to the target user based on the recommendation information.

2. The information push method based on user interest portrait according to claim 1, characterized in that: The historical unstructured behavior data includes at least one of user comment data, audio and video preference data, movie review data, and song evaluation data. The step of determining the implicit interest preference information of the target user based on the historical unstructured behavior data specifically includes: Using a target named entity recognition model, identifying the proper name entity information in the user comment data, video preference data, movie review data, and song evaluation data, and obtaining a proper name entity recognition result; Based on the information relationship extraction model and the entity recognition results of the proper names, unstructured text analysis is performed on the user review data, audio and video preference data, movie review and song evaluation data to obtain the implicit interest preference information.

3. The information push method based on user interest portrait according to claim 2, characterized in that: The method further includes identifying the proper name entity information in the user comment data, video preference data, movie review data, and song evaluation data by using the target named entity recognition model, and obtaining the proper name entity recognition result. Crawling entertainment data related to the target user based on a web crawler, wherein the entertainment data includes at least one of movie-related information, TV series-related information, short video-related information, song information, singer information, and actor information; The initial language processing model is trained based on the entertainment data to generate the target named entity recognition model, so as to perform entity recognition on the proper names in the entertainment data through the target named entity recognition model.

4. The information push method based on user interest portrait according to claim 3, characterized in that: The training of the initial language processing model based on the entertainment data to generate the target named entity recognition model specifically includes: Acquire the entertainment data and entity annotation data corresponding to the entertainment data; Based on the entertainment data and the entity annotation data, performing entity extraction training on the initial language processing model to extract entity words corresponding to the entertainment data; The entity words corresponding to the entertainment data are compared with the entity standard data, and according to the comparison result, the initial language processing model is iteratively trained until the target named entity recognition model is generated.

5. The information push method based on user interest portrait according to claim 4, characterized in that: The training of the initial language processing model according to the comparison result specifically includes: In the case where there is an abnormal comparison result, determining the entity word of the abnormal comparison result, and determining the abnormal type of the entity word of the abnormal comparison result; Based on the abnormal type of the entity word, the entertainment data is supplemented and updated, or the entertainment data corresponding to the entity word of the abnormal result is deleted.

6. The information push method based on user interest portrait according to claim 4, characterized in that: The training of the initial language processing model according to the comparison result specifically includes: In the absence of abnormal comparison results, the training of the initial language processing model is stopped and the target named entity recognition model is generated.

7. The information push method based on user interest profile according to any one of claims 1 to 6, characterized in that: The historical structured behavior data includes at least one of information browsing records, information click times or rating data, and the determining of the explicit interest preference information of the target user based on the historical structured behavior data specifically includes: Based on the interest preference prediction model, interest preference prediction is performed on the information browsing records, information click times and rating data to generate the explicit interest preference information.

8. An information push device based on user interest portrait, characterized in that: The information push device based on user interest portrait includes: A behavior data acquisition module, used to acquire historical behavior data of a target user, wherein the historical behavior data includes historical structured behavior data and historical unstructured behavior data; A preference information determination module, used to determine the explicit interest preference information of the target user based on the historical structured behavior data, and to determine the implicit interest preference information of the target user based on the historical unstructured behavior data; An interest portrait generation module, used to generate a target interest portrait of the target user based on the explicit interest preference information and the implicit interest preference information; The recommendation information push module is used to filter out the recommendation information corresponding to the target user based on the target interest portrait, so as to push personalized content to the target user based on the recommendation information.

9. A computer device, characterized in that: The computer device includes a processor, a memory, and an information push program based on user interest portraits stored in the memory and executable by the processor, wherein when the information push program based on user interest portraits is executed by the processor, the steps of the information push method based on user interest portraits as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an information push program based on user interest portraits, wherein when the information push program based on user interest portraits is executed by a processor, the steps of the information push method based on user interest portraits as described in any one of claims 1 to 7 are implemented.

Images