People Relation Extraction of Chinese Microblog Based on SVMDT-RFC

People relation extraction is a significant topic in information extraction field. While in traditional study, the feature of extraction lexical and semantic was attached importance to, and the function of classifier was neglected, furthermore, there is great difference between microblog language materials and that of tradition. When it mentioned traditional classification algorithm, its low correctness and the inaccuracy to identification of fuzzy sample become the reason of being used little. In this paper, the traditional classification algorithm was improved. Using SVMDT-Random Forest and we designed, the fuzzy sample classifying ability increased, which remedied the shortcomings of SVM and Random Forest effectively. By testing the microblog language materials, the result indicated that this method can improve the performance of people relation extraction.