Research And Integration Of Knowledge Acquisition System Based On Meta-Search

With the rapid development of the Internet,it has been an important platform and explicit knowledge base.However,the tools of existing knowledge acquisition and agents are aim to different areas and with weak flexible.To solve this problem,this thesis proposes a flexible platform for knowledge acquisition,which including many methods and techniques such as knowledge acquisition,analysis,classification,and system integration etc.Focusing on obtaining explicit knowledge from the Internet and insufficiency of existing methods and tools,corresponding solutions are proposed.The research work can be divided into four parts,knowledge acquisition from the Internet,cluster analysis and evaluation search result,the process to the semi-structured pages and implement the system.The research work of this thesis as follows:(1) In the process of knowledge acquisition,the traditional search engine is low coverage and lack of personalized service.By using meta search engine,given field knowledge bases are formed,combined with concept expansion and keywords optimization,the coverage ratio is improved.For the result process,a call method based on given field is proposed,and the weight of result pages index quantification according with weight(2) In knowledge processing stage,Meta searching result are clustered.An approach of clustering algorithm based on given field is proposed.By using vector space model and singular value decomposition,the search results can be assigned to the corresponding class themes,and the retrieval and processing is facilitate.In this part,the model of text representation and feature extraction are explained in detail..The advantage and practice of the clustering algorithm is introduced at last,and its application in the knowledge acquisition system is proper.(3) In the processing of result pages,we propose a content extraction method based on node identity analysis,which is combined with the meta data of meta search engine,plus the wrapper methods,achieve the requirement of accuracy and precision.Finally we design and implement the system,and test the knowledge system using bases of different fields,and evaluate the returned structured data and the effectiveness of clustering result,and the result is effective.