The Research For Named Entity Recognition And Relation Extraction In Text

Extracting the semantic relation innamed entities is a criticalstep of information extraction, and it is an important research direction in the field of semantic recognition. With rapid development of Internet, Internet is deeply influencing people's living, studying and working. It has significantmeaning to extract useful structural information fromfree text and Internet websites.By the development of the natural language processing and machine learning, nowadayspeopleare able to extract structural information and even knowledgefrom Internet.This dissertationintroduces the features and wide application of information extraction system, and then analyzesthe characteristics andresearch progress ofnamed entity recognition and relation extraction. Based on the existing researches on information extraction system, this dissertation constructs a named entity recognition and relation extraction system based on Spring and Strutsby using GATE and WordNet.It providesvisual processingof the extracted results.This system has good expansibility and is ease of use, andit can be integratedas a component into other information systems.In addition, this dissertation designs three kinds of algorithms to realize the relation extraction, which are shortly called'relying on verbs','core preposition' and 'possessive case'. Byusingpartofspeech and grammar structure,the system can deal withcomplexconditions like coreference resolution. The experiment results indicate that the proposed algorithm improves the accuracy of named entity recognition and relation extraction. Moreover, as constructed in web framework based on Java, it enables the system ease of transplant.