Research On The Extraction Method Of Attribute Relationship In Specific Domain Entities

Entity relation extraction is an important part of information extraction,it can identify the semantic relationships between entities,as the basis of Semantic Web entity relation extraction has been widely used in automatic answering system,information retrieval,Machine Translation etc.Relation extraction is an important basis for constructing knowledge map.With the change of semantic web,the construction of knowledge map becomes more and more important.Attribute relationship among domain entities is one of the most important parts of constructing knowledge map.For a specific domain,the domain entity attribute relationship is of great significance to the semantic relationship between the entities.In this paper,we focus on the extraction of the entity attribute relationship in Chinese domain;(1)Based on the Distant Supervision(distance supervision)domain entity attribute relation extraction method,the entity attribute relation is obtained based on domain knowledge base and related text set.Relationship type:instance attribute value.In the field of tourism in the "scenic spots" this type,take out and "attractions"this type of related attributes,such as ticket prices,area,etc.Distance supervision is through the process of knowledge base is mapped to the text focus on the realization of relation extraction,i.e.if an entity to appear in the knowledge base,the text contains the two entities of all sentences extracted from these sentences,feature extraction,classifier training.This paper first constructs a small Chinese tourism field knowledge base,predefined attributes and relation instances in the knowledge base,and the corresponding knowledge base from Baidu encyclopedia or Wikipedia web page crawled text set.In this paper,we use a variety of features to improve the performance of classifier and improve the performance of classification.(2)Extraction of domain entity attributes based on convolutional neural network.In this paper,we propose a convolution neural network to extract the attributes of domain entities,in which the attribute type is the instance attribute,attribute value,instance attribute value.The use of syntactic features,automatic learning representation entity attribute relationship convolutional neural network text feature level features and entities in the sentence,to form a feature vector,added to the convolutional neural network,classification model training entity attribute relationship.The experimental results show that the proposed method can effectively improve the extraction of entity attributes.