基于层次隐马尔可夫模型和神经网络的个性化推荐算法

传统推荐系统将推荐准确性作为主要目标,而推荐结果的多样性和个性化有所欠缺。对此,设计一种基于层次隐马尔可夫模型和神经网络的推荐算法。采用层次隐马尔可夫模型建模用户喜好和上下文环境的关系,并通过隐马尔可夫模型预测上下文。设计神经网络结构来解决协同过滤推荐的问题,同时神经网络满足贝叶斯个性化排序的条件,实现对推荐列表的个性化排序。实验结果表明,该算法在保持推荐准确性的前提下,提高了推荐的多样性和个性化。

PERSONALIZED RECOMMENDATION ALGORITHM BASED ON HIDDEN MARKOV MODEL AND NEURAL NETWORKS

Traditional recommendation systems treat recommendation accuracy as the main objective,but they are lack of diversity and personalization.I design a personalized recommendation system based on hidden Markov model and neural networks.It adopted hierarchical hidden Markov to model the relationships between user preference and context,and it predicted the context through hidden Markov model.The neural networks was designed to handle collaborative filtering recommendation problem,meanwhile the neural networks met the conditions of Bayesian personalized ranking to realizepersonalized ranking for recommendation list.The experimental results show that my algorithm improves the diversity and personalization of recommendation while maintaining the accuracy of recommendation.