面向在线社区用户的群体推荐算法研究

结合现有两种主要群体推荐算法的优势, 建立新的算法框架, 并引入差异度因素对模型进行优化。另外, 考虑到在线社区用户的特点, 定义互动度指标来描述群体成员间的互动程度, 通过分析其与推荐精度之间的关系, 探讨互动度对群体推荐的影响。选取豆瓣网数据进行实验, 并与传统方法进行比较, 结果表明, 融入差异度的算法具有更好的推荐效果, 且有效的互动机制能够保证较高的推荐精度。

Group recommendation algorithm for online community users

By combining advantages of the most two prevalent group recommending methods, this paper built a new algorithm frame and introduced the disagreement factor to perfect the model. In addition, considering the specialty of online groups, it defined a variable to describe the interaction frequency among group members, and evaluated its effect on recommending results by analyzing its relationship with the recommending precision index. It used the data of Douban to test the efficacy of this algorithm. Results show that the algorithm considering disagreement factor obtains better recommendation effect, and that a well-designed interaction mechanism contributes to improving recommending precision.