Research On Text Mining Of Doctor-Patient Question Answering Community

With the rapid development of the internet, more and more people are involved in the exchange of information. Doctors and patients can communicate with each other about health problems online in the question answering community for patients and doctors. The community keeps a large amount of question answering information, including knowledge of medical record, medical knowledge and medication health. Its excavation can provide valuable reference for medical research, business operations and so on. The traditional methods work mostly through questionnaires or manual analysis approach to research and exploration, but when faced with the growing mass of information, the shortcomings of traditional methods have become increasingly prominent. This paper adopts a combination of statistical and machine learning methods for drug name recognition and sentiment analysis in the question answering community for patients and doctors, building the hybrid approach over hierarchical structure based on machine learning methods.Aiming at the problem of colloquial corpus and non-standard drug entities in Chinese online health community, this paper uses conditional random fields model combined with maximum matching algorithm to improve the drug named entity recognition, besides adds approximate matching and search engines to normalize drug name entity. Then identify the corpus sentiment analysis of pharmaceutical research. First, it take the 2 times 2 classification methods for sentiment analysis, using support vector machine model for separate subjective and objective, choosing the n-gram, stylistic and emotional words characteristics. After that using the dictionary matching method to split subjective emotion polarity, in which the method of identify new sentiment word is used to supplement medical emotional words. Finally, the real data are checked in experiments to test the feasibility and rationality of the method in this paper.Due to the fact that the current study on Chinese online health community is still at a preliminary stage, this paper combines drug entity recognition with sentiment analysis extracting infromation about medicine research, and mining the results with emotional information in online health community provides an important reference value for medication-condition and disease-difference studies, meanwhile providing a new ideas for the age of big data of medical information mining.