以穿衣搭配数据为基础的协同过滤算法改进  

Improvement of collaborative filtering algorithm based on clothing matching data

作　　者：李圆 王孝东 于淼 LI Yuan;WANG Xiaodong;YU Miao(College of Textile and Clothing,Qingdao University,Qingdao 266071,Shandong,China;College of Fashion and Design,Donghua University,Shanghai 200051,China)

机构地区：[1]青岛大学纺织服装学院,山东青岛266071 [2]东华大学服装与艺术设计学院,上海200051

出　　处：《纺织高校基础科学学报》2023年第2期93-100,共8页Basic Sciences Journal of Textile Universities

基　　金：国家自然科学基金(52073151);山东省自然科学基金(ZR2019PEE022);中国纺织工业联合会科技指导性项目(2018078)。

摘　　要：为了提高服装推荐方法的推荐精度,解决当前服装推荐算法不成熟,智能度、精确度较低的问题。以阿里云的天池大赛——淘宝穿衣搭配挑战提供的数据为基础,参考基于项目的协同过滤推荐算法的原理,提出了一种服装产品相似度加权计算的方法。该方法将商品相似度细分为类别相似度、商品相似度、特征相似度,分别计算之后作加权处理,以改进相似度计算。本文算法与其他算法相比,在推荐精度方面有一定的提高,并通过仿真实验验证了算法的有效性。In order to improve the recommendation accuracy of clothing recommendation methods and solve the problems of current immature clothing recommendation algorithms with low intelligence and accuracy.Based on the data provided by the Alibaba Cloud Tianchi Competition:Taobao Clothing Matching Challenge,and referring to the principle of project-based collaborative filtering recommendation algorithm,a clothing product similarity weighting calculation method was proposed.This method dissolved commodity similarity into category similarity,commodity similarity and feature similarity,calculated them respectively and then dealt with them weighted to improve similarity calculation.Compared with other algorithms,the proposed algorithm has a certain improvement in recommendation accuracy,and the effectiveness of the proposed algorithm is verified by simulation experiments.

关 键 词：穿衣搭配 阿里云天池大赛 协同过滤 相似度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]