CN-121980082-A - Recommendation processing method, device and equipment based on artificial intelligence

CN121980082ACN 121980082 ACN121980082 ACN 121980082ACN-121980082-A

Abstract

The invention discloses a recommendation processing method, a device and equipment based on artificial intelligence, and relates to the technical field of artificial intelligence, comprising the steps of synchronously collecting various source data of operation behaviors and materials of a user in real time, and performing cleaning and standardization processing to generate structural characteristic data; the method comprises the steps of carrying out deep analysis on structured data, constructing a user portrait by mining historical behaviors and attributes of users, carrying out tag extraction and attribute analysis on materials to form a content portrait, screening candidate materials interested by the users from mass materials based on multidimensional correlation among the users, the materials and the tags based on the constructed user portrait and the content portrait, and sorting the screened candidate materials interested by the users according to preset rules to ensure that the related materials are preferentially displayed. The invention provides a method capable of integrating recommendation, content and search capability and realizing efficient access, intelligent recommendation, dynamic optimization and comprehensive monitoring.

Inventors

Tan Hongzhao
WANG ZHIGUO
DAI WENCHENG
LI XIAORONG

Assignees

深圳市酷开网络科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260119

Claims (10)

1. An artificial intelligence based recommendation processing method is characterized by comprising the following steps: Synchronously collecting various source data of the operation behaviors and materials of a user in real time, and cleaning and standardizing the source data to generate structural characteristic data; Deep analysis is carried out on the generated structured data, a user portrait is constructed by mining historical behaviors and attributes of the user, and tag extraction and attribute analysis are carried out on materials to form a content portrait which is used for understanding interests of the user and characteristics of the materials; Based on the construction of the user portraits and the content portraits, selecting candidate materials which are interested in the user from mass materials by adopting multidimensional correlation between the user and the materials, between the materials and the labels; And sorting the candidate materials which are screened out and interested by the user according to a preset rule, so as to ensure that the related materials are displayed preferentially.
2. The method of claim 1, wherein the step of sorting the candidate materials selected for interest to the user according to a predetermined rule to ensure preferential display of the related materials comprises: The candidate materials which are interested in the screened out user are subjected to first sorting display according to a preset rule, and the materials are subjected to second sorting display by combining a weight adjustment strategy through a built-in machine learning and deep learning model, so that the most relevant materials are ensured to be displayed preferentially; And (3) splitting the recommended results of the materials according to the similarity from the second sorting display materials, and inserting a strong insertion strategy of the positive feedback related materials to realize fusion optimization display and optimize overall recommendation.
3. The method of claim 1, wherein the step of sorting the candidate materials selected to be of interest to the user according to a predetermined rule to ensure preferential display of the relevant materials further comprises: Continuously monitoring the running state and the user interaction data in real time, detecting and processing abnormality, automatically learning and iterating model parameters based on the incremental data, and assisting in optimizing and adjusting a recommendation strategy by combining a data analysis tool; And providing the recommendation result after optimization and adjustment for external application through a standardized application programming interface API and a software development kit SDK.
4. The recommendation processing method based on artificial intelligence according to claim 1, wherein the step of synchronously collecting various source data of user operation behaviors and materials in real time, and performing cleaning and standardization processing on the source data, and generating structural feature data comprises the steps of: Acquiring user behavior data of browsing records, clicking behaviors and purchase histories of user operation applications in real time and synchronously acquiring material data of commodity details, pictures and prices; cleaning the acquired user behavior data and material data, and performing standardization treatment; And classifying, storing and updating the multi-type materials, extracting the structural characteristics of the commodity, and generating structural characteristic data.
5. The artificial intelligence based recommendation processing method according to claim 1, wherein the step of performing deep analysis on the generated structured data, constructing a user representation by mining user history behavior and attributes, and performing tag extraction and attribute analysis on the material to form a content representation, for understanding user interests and material characteristics comprises: Constructing a user portrait by mining user attributes and historical behaviors; deep analysis is carried out on the generated structured data by constructing a user portrait, interest commodities of a user are trained out, and the strength of interest points of the user is calculated; and carrying out label extraction and attribute analysis on the materials to construct a content portrait so as to understand the interests of the users and the characteristics of the materials.
6. The recommendation processing method based on artificial intelligence according to claim 1, wherein the step of screening candidate materials of interest to a user from a mass of materials based on constructing user portraits and content portraits by using multi-dimensional correlations between users and materials, materials and materials, and users and tags comprises: based on constructing the user portrait and the content portrait, adopting an explicit recall strategy and an implicit recall strategy; recall all relevant merchandise related to the user's interests based on the user-material correlation recall, based on the user's interests identified in the user representation; Detecting material marks browsed by a user based on material-material correlation recall, and recalling other material commodities similar to the material marks; Recall all items with corresponding material tags based on user-tag correlation based on user interest in the material tags; Screening candidate materials interested by a user from mass materials through multi-dimensional recall to generate a candidate list recommendation engine containing preset commodity materials; the recall is to screen candidate content sets interested by the user from a mass database, wherein the explicit expression is based on the interest explicitly expressed by the user, and the implicit expression is to mine the potential interest of the user.
7. The artificial intelligence based recommendation processing method according to claim 1, wherein the step of performing depth analysis on the generated structured data further comprises: Carrying out deep analysis on the generated structured data, and calculating the heat index of each commodity, wherein the heat index comprises short-term click rate, conversion rate acceleration rate and funnel conversion rate change; Predicting the life cycle stage of the current hot commodity according to the real-time heat index and combining the historical burst data; generating a dynamic explosive dynamic weight factor for each commodity according to the predicted life cycle stage; the step of sorting the candidate materials which are screened out and interested by the user according to a preset rule and ensuring that the related materials are displayed preferentially further comprises the steps of; The received dynamic weight factors of the explosion money are fused into a sequencing model, and the explosion money commodities in the rising period and the peak period are sequenced and lifted according to a preset rule according to the weight factors, and the explosion money commodities in the declining period are sequenced and lowered according to the preset rule; The method comprises the steps of executing a differential scattering and forced inserting strategy according to the life cycle stage of the explosive commodity, further reducing the display and removing of the explosive in the declining stage, and carrying out moderate forced inserting on the potential new explosive commodity to accelerate exposure.
8. An artificial intelligence based recommendation processing device, the device comprising: The data access and management module is used for synchronously collecting the operation behaviors of users and various source data of materials in real time, cleaning and standardizing the source data and generating structural characteristic data; The user and content understanding module is used for carrying out deep analysis on the generated structured data, constructing a user portrait by mining the historical behaviors and attributes of the user, and carrying out label extraction and attribute analysis on the materials to form a content portrait for understanding the interests of the user and the characteristics of the materials; the screening module is used for screening candidate materials interested in the user from the mass materials by adopting multidimensional correlation between the user and the materials, between the materials and the labels based on construction of the user portraits and the content portraits; And the recommendation engine module is used for sorting the candidate materials which are screened out and interested by the user according to a preset rule, so as to ensure that the related materials are displayed preferentially.
9. A terminal device comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, the one or more programs comprising steps for performing the method of any of claims 1-7.
10. A computer readable storage medium, on which a computer program is stored which, when being executed by a processor, enables an electronic device to perform the steps of the method according to any one of claims 1-7.

Description

Recommendation processing method, device and equipment based on artificial intelligence Technical Field The invention relates to the technical field of artificial intelligence, in particular to a recommendation processing method, device, terminal equipment and storage medium based on artificial intelligence. Background With the rapid development of the mobile internet, enterprise self-built independent operation APP has become a trend. However, most enterprises lack professional experience in private domain traffic operation, so that the existing recommendation system has the problems of insufficient individuation, low user viscosity, complex system access and maintenance and the like. In particular, the existing recommendation system depends on a single recommendation mode, so that the data access flexibility is poor, the model iteration efficiency is low, the full link monitoring capability is lacked, and the requirements of private domain flow fine operation on high efficiency, intelligence and dynamic optimization are difficult to meet. Accordingly, there is a need for improvement and development in the art. Disclosure of Invention In order to solve the technical problems, the invention provides a recommendation processing method, a recommendation processing device, terminal equipment and a storage medium based on artificial intelligence. The technical scheme of the application is as follows: An artificial intelligence based recommendation processing method, comprising: Synchronously collecting various source data of the operation behaviors and materials of a user in real time, and cleaning and standardizing the source data to generate structural characteristic data; Deep analysis is carried out on the generated structured data, a user portrait is constructed by mining historical behaviors and attributes of the user, and tag extraction and attribute analysis are carried out on materials to form a content portrait which is used for understanding interests of the user and characteristics of the materials; Based on the construction of the user portraits and the content portraits, selecting candidate materials which are interested in the user from mass materials by adopting multidimensional correlation between the user and the materials, between the materials and the labels; And sorting the candidate materials which are screened out and interested by the user according to a preset rule, so as to ensure that the related materials are displayed preferentially. According to the recommendation processing method based on artificial intelligence, wherein the step of sorting the candidate materials which are screened out and interested by the user according to a preset rule to ensure that the related materials are displayed preferentially comprises the following steps: The candidate materials which are interested in the screened out user are subjected to first sorting display according to a preset rule, and the materials are subjected to second sorting display by combining a weight adjustment strategy through a built-in machine learning and deep learning model, so that the most relevant materials are ensured to be displayed preferentially; And (3) splitting the recommended results of the materials according to the similarity from the second sorting display materials, and inserting a strong insertion strategy of the positive feedback related materials to realize fusion optimization display and optimize overall recommendation. According to the recommendation processing method based on artificial intelligence, the steps of sorting the candidate materials which are screened out and interested by the user according to preset rules and ensuring that related materials are displayed preferentially further comprise: Continuously monitoring the running state and the user interaction data in real time, detecting and processing abnormality, automatically learning and iterating model parameters based on the incremental data, and assisting in optimizing and adjusting a recommendation strategy by combining a data analysis tool; And providing the recommendation result after optimization and adjustment for external application through a standardized application programming interface API and a software development kit SDK. According to the recommendation processing method based on artificial intelligence, the steps of synchronously collecting various source data of operation behaviors and materials of a user in real time, cleaning and standardizing the source data, and generating structural feature data comprise the following steps: Acquiring user behavior data of browsing records, clicking behaviors and purchase histories of user operation applications in real time and synchronously acquiring material data of commodity details, pictures and prices; cleaning the acquired user behavior data and material data, and performing standardization treatment; And classifying, storing and updating the multi-type materials, extracting the stru