Mining mobile users’ interests through cellular network browsing profiles

Abstract

Mining mobile users’ interest is very important for numerous of commercial applications such as product recommendation, personalized advertisement, precision marketing, etc. In this paper, we proposed a novel clustering approach for semantic mining from cellular network browsing profiles based on the topic model. We treat each URL as a word and the user’s browsing history as a document, and adopt the Latent Dirichlet Allocation (LDA) model to represent the web browsing interest of mobile users. We further used K-means to cluster the users into several groups according to their topic similarities, and apply a feature ranking approach to explain the sematic meaning of the clustering results. The performance of the proposed approach is verified on a dataset from a telecom operator, which explains users’ interests well in the clusters.