基于知识图谱的女装相似款匹配及搭配推荐  

Similar Matching and Collocation Recommendation for Women s Wear Based on Knowledge Graph

作　　者：张理想 李沛儒 夏明[1,2] ZHANG Li-xiang;LI Pei-ru;XIA Ming(College of Fashion and Art Design,Donghua University,Shanghai 200051,China;Key Laboratory of Clothing Design and Technology of Ministry of Education,Donghua University,Shanghai 200051,China)

机构地区：[1]东华大学服装与艺术设计学院,上海200051 [2]东华大学现代服装设计与技术教育部重点实验室,上海200051

出　　处：《北京服装学院学报（自然科学版）》2023年第1期41-49,共9页Journal of Beijing Institute of Fashion Technology：Natural Science Edition

基　　金：中央高校基本科研业务费专项资金资助项目(2232020G-08)。

摘　　要：为使用户在线购物时能及时获取心仪服装单品的搭配方案,对服装相似款匹配和搭配推荐的方法进行了研究。首先通过构建知识图谱有效整合和关联与服装搭配相关领域的知识;其次,基于Jaccard算法对知识图谱中的相似款式进行筛选和匹配,通过历史样本中相似款搭配方案的数据重用为用户提供搭配灵感和决策信息;最后,通过问卷调查验证基于知识图谱的女装搭配方法的合理性和可行性。结果表明:人为判定相似款与基于Jaccard算法判定相似款一致,且相比随机搭配,人为更倾向于Jaccard算法推荐的搭配方案。In order to enable users to timely obtain the matching scheme of their favorite clothing items when shopping online,the paper studied the methods of clothing similarity matching and collocation recommendation.Firstly,the domain knowledge related to clothing collocation was effectively integrated and correlated by constructing the knowledge graph.Secondly,based on the Jaccard algorithm,similar styles in the knowledge graph were screened and matched to provide users with collocation inspiration and decision information through the data reuse of similar collocation schemes in historical samples.Finally,the rationality and feasibility of collocation method based on knowledge graph were verified by questionnaire.The results show that artificially judged similar styles are consistent with similar styles based on Jaccard algorithm,and people prefer the matching scheme recommended by Jaccard algorithm to random collocation.

关 键 词：本体 知识图谱 女装搭配 Jaccard算法 相似款匹配 

分 类 号：TS941.17[轻工技术与工程—服装设计与工程]