结合服务相似性的Bandit智能推荐模型

BANDIT INTELLIGENT RECOMMENDATION MODELCOMBINED WITH SERVICE SIMILARITY

近年来,在商品推荐系统的研究日趋成熟的基础上,考虑服务性能的Web服务推荐技术日益受到学者和技术人员的关注。但是,已有的Web推荐算法多由集中式的服务器运行,而并非用户终端本身。它们不适用于当需求频繁变化的场合,且无法由终端设备独立地实现推荐。针对无线移动环境下Web服务分散在各个终端的问题,提出结合服务相似性的Bandit智能推荐模型,使终端设备能独立、快速地对变化场景进行推荐。实验结果分析表明推荐的服务具有潜在的较高评分,并且在推荐系统的大多数评价指标上拥有较好性能。

In recent years,with the development of commodity recommendation system,people pay more attention to Web service recommendation technology with its performance.However,the existing Web recommendation algorithms mostly run on the centralized servers rather than users terminals.They are not suitable for the situation where users needs change frequently and cannot run independently on the users equipment.Aiming at the problem that Web services are distributed in different terminals in the wireless mobile environment,we propose a Bandit intelligent recommendation model combined with service similarity,which enables terminal devices to independently and quickly recommend with changing scenarios.The experiment analysis results show that the recommended services based on our model are more likely to get a higher score and this method has better performance in most evaluation indicators of the recommendation system.