Product recommendation method and device fusing manufacturer similarity

Product recommendation method and device fusing manufacturer similarity

  • CN 105,761,122 B
  • Filed: 04/29/2016
  • Issued: 09/08/2020
  • Est. Priority Date: 04/29/2016
  • Status: Active Grant
First Claim
Patent Images

1. A product recommendation method fusing manufacturer similarity is characterized by comprising the following steps:

  • step (1);

    acquiring relationship information and generating a product-enterprise relationship network topology with products and enterprises as nodes;

    the relationship information comprises product-product relationship, product-enterprise relationship and enterprise-enterprise relationship;

    the enterprise-to-enterprise relationship is a supplier-to-manufacturer relationship;

    step (2);

    acquiring product and enterprise information and storing the product and enterprise information into a Neo4j database according to a product-enterprise relationship network topology;

    calculating the proportion of a common supplier between any two manufacturers in the total suppliers of the two manufacturers in a Neo4j graph database to obtain the business similarity of enterprises;

    the steps of calculating the business similarity of the enterprise in the graph database are as follows;

    1) removing manufacturers with the number of suppliers less than 30, and setting the manufacturer label meeting the requirement as '"'"'OK'"'"';

    2) acquiring id of each enterprise with the label of '"'"'OK'"'"', and carrying out

         3) -6);

    3) taking a manufacturer P with the serial number of pid, and calculating the number pnum of the suppliers owned by the manufacturer P;

    4) according to the manufacturer P, taking each other manufacturer Q with the common supplier and obtaining the number SHARE of the common suppliers;

    for each manufacturer Q, proceed

         5) -6);

    5) calculating the number qnum of owned suppliers of the manufacturer according to the relation between enterprises and products of the manufacturer Q;

    6) SIM for calculating business similarity of enterprise P and enterprise QpqAnd establishing an edge pointing to enterprise Q from enterprise P, wherein the relationship type is Simiar, the similarity attribute name is SIM, and the value is SIMpq

    7) Finally constructing a new graph S, wherein any two manufacturers with a common supplier are connected by a Similar edge, and the attribute SIM of the edge is the service similarity;

    and (3);

    acquiring characteristic information of a user, characteristic information of a product and attention and evaluation degree information of the user to the product in a preset time period, and determining a recommended product set consisting of a plurality of products according to the information;

    according to the product-enterprise relationship network topology, associating the recommended product set with manufacturers, and storing the associated manufacturers to a manufacturer set;

    and (4);

    selecting one manufacturer from the manufacturer set, screening the manufacturers with the enterprise business similarity of the manufacturers larger than or equal to a preset enterprise business similarity threshold value, and storing the manufacturers in a recommended manufacturer set;

    screening out recommended products associated with a recommended manufacturer set from the recommended product set according to the product-enterprise relationship network topology;

    the process of obtaining the product information in the step (2) is as follows;

    crawling product information stored in a product database;

    cleaning the crawled product information to obtain product information meeting preset requirements;

    the process of acquiring the enterprise information in the step (2) is as follows;

    crawling enterprise information stored in an enterprise database;

    cleaning the crawled enterprise information to obtain enterprise information meeting preset requirements;

    the user in the step (3) represents the attention information of the product in a preset time period by adopting any numerical value between 0 and 1;

    and (4) the user in the step (3) represents the evaluation degree information of the product in a preset time period by adopting any numerical value between 0 and 1.

View all claims
    ×
    ×

    Thank you for your feedback

    ×
    ×