Mining Work Focus Using News Topic Modeling

In the era of information explosion, the Internet has become one of very important channel for information dissemination. The Internet has become huge and complex information storage over the time. Simple navigation and search no longer meet the needs of people. People also want to be able to mine some useful information from these data and provide the evaluation or summary of the past and guidance for future development. Therefore, the hot work mining system based on news topic modeling presented in this paper is valuable.The main part of hot work mining is the assessment of the number and importance of certain kind activities. So we design this automated hot work mining system based on text mining of news reports on the government departments.In this system, we propose and improve two key algorithms. First, this paper proposes a new position recognition method based on statistical and rule. Through the methods like role tagging we can automatically identify the position of people appearing in news reports. And we do the tests on the corpus of CDPF news article, and results show the great effectiveness of the proposed method. Second, we improve the existing multi-label classification algorithm process, introduced the PLSA with label info in our process. The algorithm analysis and experimental results show that the improved method is better than the original process in efficiency and accuracy.Based on these two algorithms, we design and implement the whole mining system. And this system has integrated into the data mining and analysis system on persons with disabilities. The system has provided valuable supporting information.