摘要
针对移动个性化推荐问题,通过将用户的情景信息引入到协同过滤推荐过程,提出一种结合情景和协同过滤的移动推荐算法。该算法首先通过情景相似度的计算来获得用户当前情景的近似情景集;并对"用户-项目-情景"三维模型采用情景预过滤方法进行降维,得到传统协同过滤"用户-项目"二维模型,然后结合Slope one算法进行项目的偏好预测和推荐。实验表明,该算法与传统协同过滤、Slope one算法相比,具有更高的推荐精确度。
Towards the problem of personalized recommendation in mobile network, presented a collaborative filtering algorithm based on context similarity of users by incorporating users＇ context information into collaborative filtering recommendation process. The algorithm calculates firstly context similarities to construct a set of similar contexts related to the current context of the user. Using context pre-filtering recommentation method, the “user- item-context”3D model is reduced to the “user-item”2D model. Finally, it predicts the unknown user preferences and generates recommendations based on Slope one algorithm. Experimental results indicate that this algorithm achieve better recommendation accuracy than the traditional CF algorithm and Slope one algorithm.
出处
《科学技术与工程》
北大核心
2014年第8期49-52,64,共5页
Science Technology and Engineering
基金
国家自然科学基金项目(61075053)
河北省自然科学基金项目(F2013402031)资助
关键词
情景
相似度计算
协同过滤
移动推荐
context similarity measure collaborative filtering mobile recommendation
