Design And Realization Of Domain Specific Knowledge Base Extraction Syste

The support of the knowledge base is essential in the field of natural language processing,and the automatic or semi-automatic construction of an domain-specific knowledge base is the key to the success of many domain-related applications.This thesis gives a complete domain knowledge base automatic extraction system design and implementation.The main tasks of this thesis include:(1)A complete domain knowledge base automatic extraction system is analyzed and designed.The system takes a given general knowledge base as the knowledge source and a set of text from the same domain as input,analyzes the requirements of system,and gives the complete design of the system based on the requirements.(2)The description of obtaining domain information according to the text set given in the same domain is one of the core elements of automatic extraction system.This thesis uses the entity and inter-entity relationship extraction technology to obtain the "seed" entity diagram of domain-related knowledge from the text of the same field given and uses it as the basis for the extraction of the domain knowledge base.(3)This thesis proposes a extraction method that integrates prior knowledge and context-sensitiveness.The known entity is regarded as a priori knowledge,the relationship between entities is regarded as correlation,and the domain knowledge base is extracted by coding the entity and relationship information.This thesis verifies the automatic extraction system and finds that the method given can greatly reduce the size of the general knowledge base,selected entity link tasks to verify the effectiveness,shown that the entity link model that already in place on the "smaller" domain knowledge base can still perform well.