基于三层知识融合模型的个性化商品推荐  

Personalized Merchandise Recommendation Based on Three-Layer Knowledge Fusion Model

作　　者：朱娟[1] 唐晓波[2] 

机构地区：[1]武汉大学信息管理学院 [2]武汉大学信息系统研究中心

出　　处：《图书馆学研究》2017年第5期24-30,共7页Research on Library Science

基　　金：国家自然科学基金资助项目"社会化媒体集成检索与语义分析方法研究"(项目编号:71273194)的研究成果之一

摘　　要：大数据环境为个性化推荐提供了丰富的知识源,但是如何从中获取知识并通过融合转化成实现精准个性化推荐的情报,是亟待解决的问题。从知识科学的角度,以知识融合的三层模型为基础,本文构建了以数据层知识融合、模型层知识融合以及应用层知识融合为一体的个性化推荐模型。该模型融合了大数据环境下的多源信息,构建了基于用户、商品、情境本体的知识库,从消费价值的角度深度分析商品特征,挖掘用户偏好,构建了基于语义—信任—情境融合的用户偏好模型,并利用DS证据理论对传统协同过滤和知识过滤推荐算法进行融合,实现个性化推荐。Big data environment provides a rich source of knowledge for personalized recommendation, but how to get knowledge from it and transform it into intelligence to achieve precise personalized recommendation is a problem to be solved. This paper builds a personalized recommendation model based on the three-layer model of knowledge fusion from the perspective of knowledge science, which has the data-layer of knowledge fusion, the model-layer of knowledge fusion and the application-layer of knowledge fusion. This model integrates multi-source information in large data environment, and constructs a knowledge base based on users, goods and context ontology, which analyzes the characteristics of goods in depth from the point of consumer value theory. Based on the semantic-trust- context fusion user preference model, the personalized recommendation will be achieved by using D-S evidence theory to integrate the traditional collaborative filtering and knowledge filtering recommendation algorithm.

关 键 词：个性化推荐 知识融合 消费价值 用户偏好 

分 类 号：F49[经济管理—产业经济]