Research Of Named Entity Relation Extraction Method Based On Bootstrapping

Information Extraction (IE) is an important field in language information automatic processing, and Named Entity (NE) relation extraction plays an important role in Information Extraction. It is related to Information Retrieval (IR), Question Answer (QA) system and Information Filtration, as a basic research it is of great significance for automatic Summarization, Machine Translation, Content Understanding, Linguistic Environment Production, Text Categorization and the construction of digital library.Mainly, there are two approaches for extracting the relations between the named entities. They are Knowledge Engineering Approach and Automatic Training Approach. The knowledge engineering approach has the relative good effect, but it also has obvious disadvantages. The developing of knowledge engineering approach is extremely expensive, also it is not flexible. More and more scholars start to devote to the automatic training approach research.This thesis, aiming at the present situation, discusses the named entity relation extraction deeply. The main work of this thesis is to extract the relations between named entities automatically based on automatic training approach using statistical methods, The main issues of the paper are as follows:(1) Proposed a named entity relation extraction method based on bootstrapping and it can be used to extract the relations between named entities from large corpora,(2) Used the Latent semantic analysis based on vector space model and feature extraction method to obtain the information for relation extraction.By doing these, the performance of named entity relation extraction was enhanced greatly.The method we proposed has reached average Recall of 72.9% and average Precision of 69.3% in open corpora, which validates the efficiency of the method.