面向信息茧房的用户画像多样化标签推荐

User Portrait Diversified Label Recommendation for Information Cocoons

个性化推荐算法提高了信息资源推送的精准度,但也带来了信息茧房效应,导致用户信息窄化、认知失衡、价值偏颇、群体极化。文章针对推荐系统中信息茧房导致的负面影响,结合用户画像技术中的多标签分类算法,对推荐系统用户的兴趣标签进行分级,构建多维特征标签体系,由此提出基于用户画像的多样化标签推荐方法,依据所构建的多样化标签体系进行协同过滤推荐,解决推荐系统中因信息茧房所产生的信息推送窄化和信息雷同等问题,在满足精确率的条件下,促进推荐结果的多样化。

The personalized recommendation algorithm improves the accuracy of information resource push,but also brings about the information cocoon effect,which leads to narrowing of user information,cognitive imbalance,value bias,and group polarization.In view of the negative impact caused by the information cocoon in the recommendation system,this paper combines the multi-label classification algorithm in the user portrait technology to classify the interest labels of the users of the recommendation system and build a multi-dimensional feature label system,thus proposes the recommendation method,performs collaborative filtering recommendation based on the constructed diversified label system,solves the problems of narrowing of information push and information mine caused by the information cocoons in the recommendation system,and promotes the diversification of recommendation results under the condition of satisfying the accuracy rate.