基于情境感知的餐饮O2O推荐系统研究

Research on Catering O2O Recommendation System Based on Context Awareness

餐饮O2O推荐具有情境敏感性,而普适计算和移动互联网的发展为更全面、更实时的情境信息的获取提供了基础,也使得对情境与推荐对象进行知识表示和推理成为提高推荐质量的关键。充分考虑移动商务活动中情境对用户需求的影响,设计了基于情境感知的领域本体模型结构并研究模型的实例化,通过规则推理实现餐饮O2O推荐。在此基础上,提出基于关联分析的情境规则生成方法,根据用户的历史行为挖掘情境与推荐对象的通用关联模式。并通过基于内容推荐的用户兴趣模型与菜品特征模型来表示个人对菜品的特殊兴趣偏好,构建了基于情境和基于内容相融合的混合推荐系统。实验结果表明,该方法有效解决了基于内容推荐的用户冷启动问题,并可以提高餐饮O2O推荐的准确性。

Catering O2O recommendation is context sensitive,and the development of pervasive computing and mobile Internet provides the basis for the acquisition of more comprehensive and real-time context information,and makes the knowledge representation and reasoning of context and recommendation objects become the key to improve recommendation quality.We fully consider the influence of context on user demand in M-Commerce activities,design the domain ontology model structure based on context awareness and study the instantiation of the model,realizing catering O2O recommendation through rule-based reasoning.On this basis,we propose a context rule generation method based on correlation analysis to mine general association patterns between context and recommendation objects according to the historical behavior of users.Moreover,the special interests for dishes of individuals are expressed by user interest models and dish feature models of content-based recommendation.A hybrid recommendation system of the combination of context-based and content-based recommendation methods is constructed.Experiment shows that this method can solve the user cold start problem of content-based recommendation and improve the accuracy of catering O2O recommendation.