基于网络浏览隐式反馈信息的偏好判别研究

Preferences Discriminant Analysis Based on Implicit Feedback Information While Web Browsing

网络浏览中用户行为作为隐式反馈信息蕴涵了兴趣和偏好,此信息可以用于个性化服务和推荐,降低人机交互过程中的认知负担。本文基于回归和SVD理论,提出了基于用户隐式反馈行为的情感偏好判别方法,包括粗粒度判别和细粒度判别。在用户行为维度的基础上,通过增加网页内容的类别维度,提高情感偏好粗粒度判别的准确性。同时研究用户-潜在因子,潜在因子-隐式行为矩阵的求取过程,得到用户细粒度的兴趣偏好判别模型。实验表明本文提出的方法对判别用户兴趣偏好信息是有效的。

User behavior in network browsing as implicit feedback information implies the interests and preferences. This information can reduce the cognitive burden for user to get personalized information or services. It is also the basic data of personalized information recommendation. Based on regression and SVD theory, this paper proposes a method of sentiment preference discrimination based on user implicit feedback behavior. It includes coarse-grained mining and fine-grained mining methods. Based on the user behavior dimension, the category dimension of web content is added. It increases the accuracy of sentiment preference coarse-grained mining. At the same time, the user potential factor and the latent factor implicit behavior matrix are studied. The fine-grained preference discriminant model of users is obtained. Experiments show that the proposed method is effective to distinguish the user preference information.