Research And Implementation Of Personal Attribute Slot Filling In Chinese

Personal attribute represent the most basic personal information of a person,such as "name"?"gender","work unit" etc.Personal attribute slot filling is to extract attribute information of the given name from the free text,which is easily obtained,and to fill the value into the corresponding attribute slot.With the development and popularization of Internet,there is more and more personal text information on the Internet.How to find and extract attribute information from text has great significance for constructing knowledge graph,event tracking,name disambiguation and so on.This thesis explores a variety of effective methods to extract and fill personal attribute slot,based on chinese free texts.And a personal attribute mining system is constructed based on this.It contains a few aspects of the specific work:(1)Implemented a Chinese personal attribute slot filling algorithm based on Bootstrapping.It is a method of rule extraction and filling attribute slots through rules,which are generated using Bootstrapping model.This algorithm is based on a small number of artificial rules,and automatically generates a large number of attribute extraction rules through iteration.Finally it uses those rules to recognize and extraction personal attribute and filling it to the corresponding attribute slot.(2)Implemented a semi supervised attributes slot filling algorithm based on dependency parsing.The method obtains the relationship between persons,attribute candidate words and trigger words through the dependency parsing,and obtains the personal attribute by exploring the attribute trigger words.The experimental results show that this method has better performance of slot filling.(3)Implemented the Chinese personal attribute slot filling algorithms based on Long Short-Term Memory(LSTM)and Convolutional Neural Network(CNN)model respectively.Finally,the advantages and disadvantages between the supervised model and the semi supervised model in the attribute slot filling task are compared.(4)A Chinese personal attribute mining system is implemented based on the exploration of Chinese personal attribute slot filling models.