改进的协同过滤混合算法在电影系统中的应用

Application of Improved Collaborative Filtering Hybrid Algorithm in Film System

摘要: 随着现代科技的发展，在当前信息过载的互联网时代，根据用户历史行为数据进行物品推荐已经成为当前研究的热点。近年来，推荐系统不断更新迭代，大多围绕准确性问题进行研究和改进。但是，多样性、新颖性以及长尾物品推荐等方面也同等重要。通过对协同过滤推荐系统的研究分析，引入多样性因子，提出一种基于聚类的混合协同过滤推荐算法，通过在Movielens数据集上验证了基于聚类的混合协同过滤算法较传统的协同过滤在提高了多样性的同时，准确性也得到改善，满足用户个性化需求。

Abstract: With the development of modern science and technology, in the current Internet era of information overload, item recommendation based on user historical behavior data has become the focus of current research. In recent years, the recommendation system has been continuously updated and iterated, and most of them focus on the accuracy problem. However, diversity, novelty and long tail recommendation are equally important. Through the research and analysis of collaborative filtering recommendation system, the diversity factor is introduced, and a hybrid collaborative filtering recommendation algorithm based on clustering is proposed. It is verified on Movielens data set that the hybrid collaborative filtering algorithm based on clustering improves the diversity and accuracy compared with the traditional collaborative filtering, so as to meet the personalized needs of users.