Dynamic Logic Topic Model And Visualization

News is one of the most fundamental information source for social research or business decision making. With the development of information technology, traditional paper media is transforming to an electronic type and the amount of news is growing explosively. Nowadays, the amount of information provided by news is impossible for reader to read and what really matters is the topics lies inside those words. News topic is the process that extract the essential information out of the text. It is the key of understanding news text quickly and with that technology, we can gather information faster and make wiser decisions.In our paper, we implemented an automated news topic analytic system. Recently, data mining and natural language processing technology has made a lot of progress, this enables the possibility of exploring the underlying topics in news possible. On the other hand, the overload of information motivates the improvement of visualization technology. Our system is based on the former two motivations and proposed a creative method for modeling the topics of news. Overall, we solved following problems:1. How to generate topics for news in a human sensible way. We improved the old method by incorporating temporal factor and human knowledge base.2. How to visualizing news text.3. How to jointly design news topic visualization system. We proposed and implemented a framework that can help user better understand and explore the news corpus.Finally, we used news data for topic modeling experiment and visualization. We proved that our algorithm outperforms the traditional data in human interoperability. Overall, our model and design is a very good solution for news topic modeling and visualization.