Mining Users’ Important Locations and Semantics on Cellular Network Data

With the development of mobile communication technology, mobile phones play a more important role in people's daily life. The user's location and its semantic are very important to Location Based Services (LBS), and this inspires a tremendous amount of research effort on analyzing large-scale trajectory data to mine these informations in the last decade. The existing researches have achieved good results, but there are still some challenges, such as the limitation on low quality data. In this paper, firstly, we propose a novel algorithm to mine meaningful points on cellular network data, which can reduce the processing time apparently by conditional filtering original data. Then, we adopt a density based clustering method together with a naive Bayes classifier to complete the important location discovery and semantic analysis. Theoretical analysis and extensive experiments on real dataset show that: 1) The precision rate and recall rate of identifying important locations have reached more than 96%, which are improved by nearly 3% comparing with the existing algorithms. Meanwhile, the precision rate and recall rate of semantic analysis are also increased by more than 4%. 2) When applying to low quality data, the proposed approach is also robust, efficient and has a good performance.