摘要
大数据背景下,一般使用推荐算法获取目标用户。基于此,通过对各类推荐算法进行分析,比较各类算法的优缺点,并针对各类算法的特点和不足,提出一种混合推荐算法。首先,为解决算法初期的冷启动现象,将主题模型与协同过滤算法相结合,生成用户偏好概率预测矩阵;其次,为改善用户过少造成的稀疏性问题,采用聚类算法填充评分矩阵;最后,为进一步提高推荐精确度,改进各项权重参数,生成融合主题模型和协同过滤推荐算法的混合推荐方法。
In the context of big data,recommendation algorithms are generally used to obtain target users.Based on this,by analyzing various recommendation algorithms,comparing their advantages and disadvantages,and aiming at the characteristics and disadvantages of various algorithms,a hybrid recommendation algorithm is proposed.Firstly,the topic model is combined with the collaborative filtering algorithm to generate the user preference probability prediction matrix.Secondly,a clustering algorithm is used to fill the scoring matrix.Finally,a hybrid recommendation method combining the theme model and collaborative filtering recommendation algorithm was generated.
作者
赵棣
ZHAO Di(Shandong University of Engineering and Vocational Technology,Jinan Shandong 250200,China)
出处
《信息与电脑》
2023年第5期81-83,共3页
Information & Computer
关键词
主题模型
协同过滤
混合推荐
聚类算法
theme model
collaborative filtering
mixed recommendation
clustering algorithm
