基于用户空间位置评分云模型的Web服务协同过滤推荐算法

Web服务作为无形的产品,不具备真实环境下的空间地理位置坐标,针对服务推荐中无法衡量用户群体与Web服务之间的距离位置关系,造成用户相似度计算失衡,导致推荐不准确等问题,提出了基于用户空间位置评分云模型的Web服务协同过滤推荐算法.首先基于用户群体的行为数据量化Web服务的热度区域,通过空间位置量化评分描述用户对于Web服务的兴趣偏好;其次利用云模型来描述每个用户空间行为评分的整体特征,设计了云模型间相似贴近度的计算方法,基于该方法提出了一种用户差异程度系数评估算法,并作为调控系数优化了皮尔森相似度量;最后通过协同过滤找出用户感兴趣的Web服务.实验结果表明该算法使得用户行为偏好的区域划分更加精确,在推荐准确率上明显提高,为基于位置的Web服务推荐提供新颖的方案.

Collaborative filtering recommendation algorithm for web services based on user-space location score cloud model

Web services, as invisible products, don''t have spatial geographic coordinates in the real environment. Aiming at the inability to measure the distance position relationship between the user group and the Web service in the services recommendation, caused user similarity calculations to be out of balance, leading to inaccurate recommendations and other issues, this paper proposed a collaborative filtering recommendation algorithm for Web services based on user-space location rating cloud model. Firstly, based on the behavior data of user groups, it quantified the hot areas of Web services and described the user''s interest and preference for Web services through the spatial location quantitative score. Secondly, it used the cloud model to describe the overall characteristics of each user''s spatial behavior score, and designed the calculation method of similar closeness between cloud models. Based on this method, this paper proposed a user difference degree coefficient evaluation algorithm, and optimized the Pearson similarity measure as a control coefficient. Finally, it found out the Web services that users were interested in through collaborative filtering. Experimental results show that the algorithm makes the regional division of user behavior preferences more accurate, the recommendation accuracy rate is significantly improved, and it provides a novel solution for location-based Web service recommendation.