Search

CN-122022943-A - Commodity information recommendation method, device and equipment

CN122022943ACN 122022943 ACN122022943 ACN 122022943ACN-122022943-A

Abstract

The embodiment of the application provides a commodity information recommendation method, device and equipment. The method comprises the steps of obtaining commodity information and social relation information of interacted commodities of a target user, determining commodity information of first non-interacted commodities of the target user and basic grading information of the first non-interacted commodities based on the commodity information of the interacted commodities, determining social influence factors of the first non-interacted commodities based on the social relation information, determining comprehensive grading information of the first non-interacted commodities based on the basic grading information of the first non-interacted commodities and the social influence factors of the first non-interacted commodities, determining ordering probability of the first non-interacted commodities based on the comprehensive grading information of the first non-interacted commodities and a preset prediction model, and generating a commodity information recommendation list based on the ordering probability of the first non-interacted commodities. The method is used for achieving the effect of improving the commodity recommendation information accuracy of the transaction platform.

Inventors

  • ZHU HONGXU

Assignees

  • 北京侠客汇信息技术有限责任公司

Dates

Publication Date
20260512
Application Date
20251229

Claims (10)

  1. 1. A recommendation method of commodity information, comprising: acquiring commodity information and social relation information of interacted commodities of a target user, wherein the social relation information characterizes social friends of the target user; Determining commodity information of a first non-interactive commodity of the target user and basic grading information of the first non-interactive commodity based on commodity information of the interacted commodity; determining a social influence factor of the first non-interacted commodity based on the social relation information; Determining comprehensive scoring information of the first non-interactive commodity based on the basic scoring information of the first non-interactive commodity and the social impact factor of the first non-interactive commodity; Determining the ordering probability of the first non-interactive commodity based on the comprehensive scoring information of the first non-interactive commodity and a preset prediction model; And generating a commodity information recommendation list based on the ordering probability of the first non-interacted commodity.
  2. 2. The method of claim 1, wherein the determining merchandise information for the first non-interacted merchandise of the target user and base scoring information for the first non-interacted merchandise based on merchandise information for the interacted merchandise comprises: determining basic scoring information of the interacted commodity based on commodity information of the interacted commodity; And generating commodity information of the first non-interactive commodity of the target user and the basic grading information of the first non-interactive commodity based on the basic grading information of the interacted commodity and a preset recommendation model.
  3. 3. The method of claim 2, wherein the merchandise information for the interacted merchandise includes interaction behavior and interaction time, wherein the determining the base scoring information for the interacted merchandise based on the merchandise information for the interacted merchandise comprises: determining initial scoring information of the interacted commodity based on interaction behavior in commodity information of the interacted commodity; Determining a time attenuation factor of the interacted commodity based on the interaction time in the commodity information of the interacted commodity; And determining basic scoring information of the interacted commodity based on the initial scoring information of the interacted commodity and a time attenuation factor of the interacted commodity.
  4. 4. The method of claim 1, wherein the social relationship information has an interaction weight between the social friends and the target user, wherein the determining the social impact factor of the first non-interacted merchandise based on the social relationship information comprises: Acquiring basic scoring information of a second non-interactive commodity of the social friend of the target user based on the social relation information; Determining the contribution degree of each social friend to the second non-interactive commodity based on the basic scoring information of the second non-interactive commodity and the interaction weight between the social friend and the target user; and determining a social influence factor of the first non-interactive commodity according to the contribution degree of each social friend to the second non-interactive commodity.
  5. 5. The method of claim 1, wherein the determining composite scoring information for the first non-interacted good based upon the base scoring information for the first non-interacted good and a social impact factor for the first non-interacted good comprises: Acquiring weight parameters corresponding to scene identification information; and based on the weight parameter corresponding to the scene identification information, carrying out weighted fusion on the basic scoring information of the first non-interactive commodity and the social influence factor of the first non-interactive commodity to obtain comprehensive scoring information of the first non-interactive commodity under the scene corresponding to the scene identification information.
  6. 6. The method of claim 1, wherein the determining the ordering probability of the first non-interactive commodity based on the composite score information of the first non-interactive commodity and a preset predictive model comprises: The method comprises the steps of obtaining multi-dimensional characteristic information of a first non-interactive commodity, wherein the multi-dimensional characteristic information comprises at least two of basic scoring information of the first non-interactive commodity, commodity color forming information of the first non-interactive commodity, price characteristic information of the first non-interactive commodity, social influence factors of the first non-interactive commodity, interaction scoring information of a social friend on the first non-interactive commodity, interaction frequency information between a target user and the social friend and scene identification information; and determining the ordering probability of the first non-interactive commodity based on the comprehensive score information of the first non-interactive commodity, the multidimensional feature information and a preset prediction model.
  7. 7. The method of claim 6, wherein generating the item information recommendation list based on the ordering probability of the first non-interacted item comprises: And generating a commodity information recommendation list under a scene corresponding to the scene identification information based on the scene identification information corresponding to the ordering probability of the first non-interacted commodity.
  8. 8. The method according to any one of claims 1-7, further comprising: acquiring scene identification information represented by request parameters of a front-end interface; And displaying a commodity information recommendation list in a scene corresponding to the scene identification information on a front-end interface based on the scene identification information.
  9. 9. A recommendation device for commodity information, comprising: The acquisition module is used for acquiring commodity information and social relation information of interacted commodities of the target user, wherein the social relation information characterizes social friends of the target user; The first determining module is used for determining commodity information of a first non-interactive commodity of the target user and basic grading information of the first non-interactive commodity based on the commodity information of the interacted commodity; A second determining module, configured to determine a social impact factor of the first non-interacted commodity based on the social relationship information; a third determining module, configured to determine comprehensive scoring information of the first non-interactive commodity based on the basic scoring information of the first non-interactive commodity and a social impact factor of the first non-interactive commodity; The fourth determining module is used for determining the ordering probability of the first non-interactive commodity based on the comprehensive scoring information of the first non-interactive commodity and a preset prediction model; and the generation module is used for generating a commodity information recommendation list based on the ordering probability of the first non-interactive commodity.
  10. 10. An electronic device is characterized by comprising a memory and a processor; The memory stores computer-executable instructions; The processor executing computer-executable instructions stored in the memory, causing the processor to perform the method of any one of claims 1-8.

Description

Commodity information recommendation method, device and equipment Technical Field The present application relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, and a device for recommending merchandise information. Background The social secondhand transaction platform is a composite electronic commerce scene integrating commodity transaction and social interaction, and a user can issue or purchase secondhand commodities on the platform and can interact with friends or potential buyers through social behaviors such as praise, comment, private letter and the like. The transaction platform can conduct personalized commodity recommendation to the user based on the behavior data (such as browsing, collecting, purchasing and ordering) and the social behavior data (such as praise, comment and private letter) of the user, so that the user searching cost is reduced, and the transaction conversion efficiency is improved. In the related technology, collaborative filtering algorithm based on user-commodity interaction is mostly adopted, and recommendation is generated by relying on historical interaction data of a user, but the method ignores key influences of friend preference and interaction endorsement on secondhand commodity purchasing decision, so that recommendation results are disjointed from actual transaction demands of the user, and accuracy is low. Disclosure of Invention The commodity information recommending method, device and equipment provided by the embodiment of the application are used for achieving the effect of improving the commodity recommending information accuracy of the transaction platform. In a first aspect, an embodiment of the present application provides a method for recommending merchandise information, including: acquiring commodity information and social relation information of interacted commodities of a target user, wherein the social relation information characterizes social friends of the target user; Determining commodity information of a first non-interactive commodity of the target user and basic grading information of the first non-interactive commodity based on commodity information of the interacted commodity; determining a social influence factor of the first non-interacted commodity based on the social relation information; Determining comprehensive scoring information of the first non-interactive commodity based on the basic scoring information of the first non-interactive commodity and the social impact factor of the first non-interactive commodity; Determining the ordering probability of the first non-interactive commodity based on the comprehensive scoring information of the first non-interactive commodity and a preset prediction model; And generating a commodity information recommendation list based on the ordering probability of the first non-interacted commodity. In one possible implementation manner, the determining, based on the commodity information of the interacted commodity, commodity information of a first non-interacted commodity of the target user and basic scoring information of the first non-interacted commodity includes: determining basic scoring information of the interacted commodity based on commodity information of the interacted commodity; And generating commodity information of the first non-interactive commodity of the target user and the basic grading information of the first non-interactive commodity based on the basic grading information of the interacted commodity and a preset recommendation model. In one possible implementation, the commodity information of the interacted commodity comprises interaction behavior and interaction time, and the determining the basic scoring information of the interacted commodity based on the commodity information of the interacted commodity comprises the following steps: determining initial scoring information of the interacted commodity based on interaction behavior in commodity information of the interacted commodity; Determining a time attenuation factor of the interacted commodity based on the interaction time in the commodity information of the interacted commodity; And determining basic scoring information of the interacted commodity based on the initial scoring information of the interacted commodity and a time attenuation factor of the interacted commodity. In one possible implementation, the social relationship information includes a social relationship between a social friend and a target user, and the determining the social influence factor of the first non-interacted commodity based on the social relationship information includes: Acquiring basic scoring information of a second non-interactive commodity of the social friend of the target user based on the social relation information; Determining the contribution degree of each social friend to the second non-interactive commodity based on the basic scoring information of the second non-interactive