基于RBM模型的豆瓣小组推荐系统设计与实现

Design and Implementation of the Recommendation System for Douban Group based on RBM Model

将受限玻尔兹曼机(restricted boltzmann machine,RBM)模型应用于推荐领域已成为一个很有意义的研究方向。针对豆瓣小组,设计实现了一个基于RBM模型的推荐系统,该系统由数据层、模型层、评测层3部分组成。数据层通过选取"豆瓣达人"数据,一定程度上解决了数据稀疏问题。模型层利用对比散度(contrastive divergence,CD)算法进行学习。实验结果表明,在豆瓣小组数据集上,RBM模型相较传统协同过滤算法具有更好的推荐效果。

Restricted Boltzmann Machine for recommendation has become one of the significant researches.In this paper,a recommendation system for Douban Group based on RBM model is designed and implemented.The system consists of three layers： data layer,model layer and evaluation layer.The data layer can solve the problem of data sparsity to a certain extent by selecting the data of ＂the Douban expert＂.The experimental results show that the RBM model rivals the traditional collaborative filtering algorithm by providing a better recommendation effect on the data set of the Douban Group.