A Method of Public Opinion Relation Extraction Combined With Relationship Enhanced Fusion Model
Aiming at the problems of various types of public opinion information relations and huge state space,and the problems of relation-ship overlap and single model can not completely identify all triples in the relationship extraction task of public opinion information,a method of public opinion relation extraction based on relationship enhancement fusion model was proposed.First,the public opinion information ob-tained from the Chinese Internet is preliminarily processed to obtain a preliminary relationship table;Then,entity types are introduced to en-hance the relationship table obtained;Finally,the enhanced relationship table is input into the fusion model as a priori feature to improve the accuracy of relationship classification.Combining the recognition results of the two models,the problem that a single model cannot fully recog-nize all triples is solved.The experimental results indicate that,compared with the single model without relationship enhancement,the F1 val-ue of this method is increased by 5.4%.
国家科技期刊平台
NSTL
万方数据
