Research On Chinese Feature-based Semantic Relation Extraction Between Named Entities

Semantic Relation Extraction(SRE) identifies and classifies the relationship between two entities from corpus.It plays a critical role in Information Extraction.At present, Chinese Feature-based Semantic Relation Extraction takes an importance in machine leaning method,the key of the method is how to get different kinds of effective word,entity,semantic feature information,and to combine these features reasonable,so that we can describe the local and whole object's features integral.This paper proposes a feature-combine method based on basic features to solve the relation extraction problems,it mainly includes extracting Chinese basic features and combined features from Chinese Corpus.In this paper,we adopt ACE2005 Chinese corpus for the experimental data.First,after data preprocess,we investigate the word,entity and syntax features from segmented documents independently in feature-based relation extraction using Support Vector Machines.The F-measure of relation detection,major type relation classify and subtypes relations classify are 62.78/57.08/55.77 respectively.Second we get combined features by combining word,entity and basic phrase with each other.It shows that combined features improve the F-measure by 8.5/3.4/3.8 in relation detection,major types classify and subtype respectively,so combined method is a reasonable method to improve the result of relation extraction.In brief,basic features and combined features have high performance in distinction and improve the performance in Chinese Feature-based Semantic Relation Extraction between Named Entities.