Research On Multi Type Named Entity Extraction Based On The Complex Characteristics Of The Knowledge Of Food Safety Events

Food safety is related to agriculture,nural areas,Peasants and it is an important research topic of the "three rural".Food safety incidents has always been one of the most focus attention.Although different news media reported many incidents of food safety,but sorting out events,building the corpus for food safety incident and achieving extraction of multiple types of named entity extraction are still in the preliminary inquiry stage.This paper collected and organized food safety incidents on tihe network from the news media which is from 2005 to 2015 for the buliding of food safety corpus.At the same tine,it researched and analysised these food safety incidents with the knowledge of natural language and machine learning to extract food safety incidents corpus' named entity.In this paper,it used the CRFs machine learning model and the complex characteristics of knowledge.By analysising and researching part of speech,word boundary characteristics,we classified the marked food safety incidents corpxus and determined feature template for applicating CRFs model in the experiments.By dividing the whole corpus according to certain percentage,determining the performance of the model and analyzing the results with the corresponding evaluation.Then,completing named entity extraction of complex knowledge and many types of food safety incidents from crawling CNKI corpus.The paper achieved the automatic generation of time series evolution on extraction of time named entity by delving into this type of entity.By analyzing the food safety incidents feature of time named entity in the event expression and commonly used words in the event expression,this paper implemented the named entity time conversion time-specific expression and every word or the formation of a time corresponding to each event expression.