Research and Application of Personalized Modeling Based on Individual Interest in Mining

Baocheng Huang,Guang Yu
DOI: https://doi.org/10.1155/2014/514295
2014-01-01
Abstract and Applied Analysis
Abstract:Weibo services, provided by the service providers, is simple and changeless. The research based on the content of microblog reflects the user's personalized features. The method has important significance to improve user satisfaction and expand the scale of users. First, the interest classification problem called multiclass classification algorithm is proposed based on improving support vector machine of binary tree. Second, an improved model of mixed interest based on implicit feedback is proposed. This method is based on the shortcomings of the establishment of the interest model and the drift strategy in update phase among existing users. The improved model is applied to the user modeling of personalization, improving the authenticity and accuracy of the personalized modeling.
What problem does this paper attempt to address?