Abstract: Batch service recommendation has a great application prospect in Service Computing area. However, how to adapt to the frequent changes of the large-scale online user demand to dynamically recommend the most stable service collection is still a challenging technical problem. This paper presents a new recommendation algorithm of services, which uses backtracking to choose determine the users which have some service target. The impact of time variation on user online state, to discover the best service set which can satisfy most user demand stably with changing time, through the dynamic competition optimization of services with the users. To verify the rationality and validity of the proposed algorithm, a series of experiments is maken on real data set from WS-DREAM. Experimental results show that the proposed algorithm can effectively help product providers find the most stably service collection with changing time, thus achieving the maximization of interests.

Key words: changing time, most stably demand, service recommendation, personalization service, reverse Top-k query, knapsack problem

摘要： 服务的批量推荐在服务计算领域具有巨大的应用前景。然而，针对动态变化的大规模在线用户，如何实现时序范围内最具稳定性服务集合批量推荐，仍然是一个极具挑战的技术问题。提出一种新的服务集推荐算法，它采用回溯法挑选出满足潜在用户需求的服务集，并着重考虑用户在线状态的实时变化，通过服务集与用户集的动态竞争优化，最终挖掘出时序范围内稳定满足最多用户需求的服务集。为验证提出算法的合理性和有效性，利用WS-DREAM的真实数据集，进行了一系列实验。实验结果表明，提出的算法能够有效发现系统中时序范围最稳定满足用户需求的服务集合，从而达到利益最大化。

关键词: 时序范围, 最具稳定性, 服务集推荐, 个性化服务, 反向Top-k查询, 背包问题