Cloud computing based cold start item recommending method and device

Cloud computing based cold start item recommending method and device

  • CN 104,615,741 A
  • Filed: 02/12/2015
  • Published: 05/13/2015
  • Est. Priority Date: 02/12/2015
  • Status: Active Application
First Claim
Patent Images

1. , based on a method for the cold start-up project recommendation of cloud computing, it is characterized in that, comprise the following steps:

  • S1, registration new projects, and obtain the item attribute that this project has;

    The classification of S2, default new projects, described classification comprises " Like " and " Dislike ";

    The rating matrix of S3, preset term attribute, comprise the matrix number P (u that just marks, a), negative scoring matrix number N (u, a) and overall score matrix number T (u, a), wherein, P (u, a), N (u, a) and T (u, each element a) respectively representative of consumer to the just scoring number of existing project projects attribute, negative scoring number and overall score number;

    S4, import existing project from cloud server, comprise user, user in described existing project to the score value of existing project, and the information of item attribute belonging to existing project;

    The score data information of rating matrix counting user of S5, the existing project imported according to step S4 and described item attribute, be specially;

    travel through the score value of all users to existing project, and whether judge user to the score value of existing project higher than the score value preset, if, just will be marked matrix number P (u, a) with overall score matrix number T (u, a) corresponding element adds one, if not, scoring matrix number N (u will be born, a) and overall score matrix number T (u, corresponding element a) adds one;

    S6, to calculate the probability P (u comprising cold start-up project i in project that user u likes according to described score data information and bayesian algorithm, Like|i), with the probability P (u, Dislike|i) comprising cold start-up project i in the project that user u does not like;

    S7, according to formula;

    D(u,Like|i)=P(u,Like|i)P(u,Like|i)+P(u,Dislike|i)Be normalized P (u, Like|i), D (u, Like|i) value normalized obtained is by descending sort, and the N number of user choosing rank forward carries out this project recommendation, and wherein N is positive integer.

View all claims
    ×
    ×

    Thank you for your feedback

    ×
    ×