会员体验
专利管家(专利管理)
工作空间(专利管理)
风险监控(情报监控)
数据分析(专利分析)
侵权分析(诉讼无效)
联系我们
交流群
官方交流:
QQ群: 891211   
微信请扫码    >>>
现在联系顾问~
热词
    • 9. 发明申请
    • METHODS AND DEVICES FOR CALCULATING RANKING SCORE AND CREATING MODEL, AND PRODUCT RECOMMENDATION SYSTEM
    • 计算分级和创建模型的方法和设备,以及产品推荐系统
    • WO2016058485A3
    • 2016-06-09
    • PCT/CN2015091216
    • 2015-09-30
    • ALIBABA GROUP HOLDING LTDLIU RUILV TAOSUN CHAOYANG ZHIXIONG
    • LIU RUILV TAOSUN CHAOYANG ZHIXIONG
    • G06F17/30
    • G06F17/30
    • A method and device for calculating a ranking score of an object to be evaluated, a method and device for creating a ranking score calculation model, and a product recommendation system. The method for calculating a ranking score of an object to be evaluated comprises: obtaining for an object to be evaluated initial evaluation score data and additional ranking factor values, as well as actual behavior data extracted from past behavior data in an actual interaction behavior system (101); minimizing to an optimal target the difference between the actual ranking distribution obtained on the basis of actual behavior data and the predicted ranking distribution obtained on the basis of a pre-set ranking score calculation model, and solving for the weight coefficients of additional ranking factors in the ranking score calculation model (102); using said ranking score calculation model to calculate the ranking score of the object (103). In the present invention, additional ranking factors may be quickly and conveniently factored into calculations, and ranking scores resulting from said calculations can objectively and accurately predict the ranking of an object.
    • 一种用于计算待评估对象的评分的方法和装置,用于创建排名得分计算模型的方法和装置以及产品推荐系统。 用于计算要评估对象的排名得分的方法包括:获得待评估对象的初始评估分数数据和附加排名因子值,以及从实际交互行为系统中的过去行为数据中提取的实际行为数据( 101); 最小化到根据实际行为数据获得的实际排名分布与基于预先设定的排名分数计算模型获得的预测排名分布之间的差异的最优目标,并求解其中的附加排名因子的权重系数 排名分数计算模型(102); 使用所述排名分数计算模型来计算所述对象(103)的排名得分。 在本发明中,附加排名因素可以快速方便地考虑到计算中,并且由所述计算得出的分数得分可以客观和准确地预测对象的排名。