Semantic Approach to Discover Topic over Mail Data
Text sequences or Time stamped texts, are everpresent in real-world applications. Multiple text sequences are frequently connected to each other by distributing common topics. The correspondence between these sequences provides more significant and comprehensive clues for topic mining than those from every individual sequence. However, it is non retrieval to explore the equivalence with the existence of asynchronism among multiple sequences, i.e., documents from different sequences about the same topic may have different time stamps. In this paper, we properly addressed the problem and suggested a new algorithm based on the generative topic model. The proposed algorithm consists of two alternate steps: the first step retrieves common data from multiple sequences based on the arranged time stamps provided by the second step; the second step arranges the time stamps of the documents according to the time distribution of the topics found by the first step. We accomplish these two steps simultaneously and after number retrievals a monotonic convergence of our objective function can be extracted. The effectiveness and advantage of our approach were justified through extensive practical studies on two real data sets consisting of six research paper repositories and two news article feeds, respectively
