基于情境语义推理的O2O移动推荐系统研究  

Study on O2O Mobile Recommendation System Based on Contextual Semantics Inference

作　　者：艾丹祥[1] 张玉峰[2] 左晖[3] 杨君[1] 

机构地区：[1]广东工业大学管理学院,广州510520 [2]武汉大学信息管理学院,武汉430072 [3]广东工业大学经济与贸易学院,广州510520

出　　处：《情报杂志》2015年第8期182-189,共8页Journal of Intelligence

基　　金：国家自然科学基金面上项目"基于动态数据挖掘的物流信息智能分析研究"(编号:71373197);国家社会科学基金青年项目"移动网络环境下情景敏感的个性化知识推荐机制研究"(编号:11CTQ020)

摘　　要：移动网络的发展使O2O电商模式的线下业务对象推荐受到关注。首先,针对O2O推荐的情境敏感性,探讨适用于推荐的情境语义建模方法,并以移动餐饮推荐为应用背景,设计情境本体的两层结构模型并研究其实例化。其次,提出基于情境本体的推荐规则生成方法和基于规则推理的推荐算法。最终,融合各方法设计开发O2O移动推荐系统并进行运行实验,对推理效率和推荐准确性进行跟踪评估。实验结果表明新方法和系统能更好适用于个性化O2O推荐,具有可行性。With the development of mobile network, the recommendation of offline business objects in 020 mobile e-commerce pattern has gained much concern. Firstly, concerning the existence of context sensitivity in 020 mobile recommendation, the contextual semantics modeling method suitable for the recommendation is discussed, and taking the mobile dining recommendation as the application back- ground, a two-layer contextual ontology model is designed and its instantiation is studied. Secondly, an automatic rule generation method based on contextual ontology and a recommendation algorithm based on rule inference are proposed. Finally, by fusing all the methods, the 020 mobile recommendation system is designed and developed to carry on running experiments in which the inference efficiency and the recommendation accuracy are evaluated. The result indicates that the new methods and system are feasible and more suitable for person- alized 020 recommendation.

关 键 词：移动推荐 情境本体 情境语义推理 O2O 

分 类 号：G350[文化科学—情报学]