Abstract: In view of the context awareness feature of the mobile user’s behavior sequence, this paper proposes a behavior transfer pattern recommendation algorithm MPRC. This algorithm uses Apriori to filter the data of user’s historical behavior in order to obtain a frequent pattern with length 2, and then converts these data into the decision tables for using rule extraction method of rough set to process the decision tables to mine context awareness transfer pattern. Finally, MPRC uses pattern matching method and contextual similarity calculation to rank the pattern and to make recommendation. Experimental results show that this algorithm is more effective and more accurate in user behavior pattern mining and recommendation in mobile environment.

Key words: context awareness, behavior transfer pattern, contextual similarity

摘要： 针对移动用户行为序列的情景感知特性，提出一种基于情景感知的行为转移模式推荐算法MPRC。该算法首先采用Apriori对用户历史行为数据进行长度为2的频繁模式的挖掘过滤，然后将过滤后的行为数据转换成决策表，采用粗糙集规则提取对决策表进行处理，挖掘情景转移模式，最后通过模式匹配及情景相似性计算进行推荐排序。实验结果证明了该算法在移动环境下的模式挖掘及推荐方面的有效性和较高的准确性。

关键词: 情景感知, 行为转移模式, 情景相似性