基于FP＿Growth和Slope＿one的图书推荐

Book Recommendation Based on FP＿Growth and Slope＿one

电子商务的大环境下,人们在电商平台中很难有效地选择出感兴趣的书籍,推荐系统能够解决这一问题。然而将传统的协同过滤算法应用到图书推荐中,存在着数据稀疏、推荐准确率低等问题。针对这些问题,在传统协同过滤算法的基础上,文中提出了一种基于FP_Growth和slope_one的协同过滤算法。首先对数据集中默认评分为0的记录进行重新评分,然后采用基于FP_Growth的矩阵填充算法对数据集进行填充,最后对协同过滤算法中的slope_one预测评分策略进行改进。将该算法应用到Book-Crossing图书数据集中,实验结果表明,改进后的算法推荐效果提升明显。该算法不仅解决了用户对书籍的选择问题,而且能够帮助电商最大程度地提升销售额。

In e-commerce environment,it’s difficult for people to select the books they are interested in effectively on the e-commerceplatform. The recommendation system can cope with this. However,there are some problems such as sparse data and low recommenda-tion accuracy when applying the traditional collaborative filtering algorithm into book recommendation. For this,we propose a collabora-tive filtering algorithm based on FP_Growth and slope_one. Firstly,the scores are re-recorded in data sets when the default score is zero.Then the data set is filled by matrix filling algorithm based on FP_Growth. Finally,the slope_one predictive scoring strategy in collabora-tive filtering algorithm is improved. Application of this algorithm into Book-Crossing data set,the experiment shows that it is significant-ly improved in recommendation effect,not only solving the problem of users’ selection of books,but also helping e-commerce to maxi-mize sales.