The Research On Conceptual Graph Indexing Of Book Abstract In Book Retrieval

At present people always use book retrieval to get book information from the large quantity of book resource. Book retrieval is becoming a hot research field in information retrieval, and it is widely used in digital library and bookstore on internet. So far almost all the book retrieval systems use keywords based retrieving methods, whose core is matching the critical character mechanically. One of the existing problems is the low recall and precision, which leads to the unsatisfactory performance. Therefore we start to research on conceptual graph based book retrieval model. This book retrieval model indexes both book information and user queries into conceptual graph, and find retrieval result of good quality through conceptual graph matching. This way can makeup the defects of keywords based retrieval effectively.The research work of this paper is a part of the research on conceptual graph based book retrieval model. This paper mainly researches on conceptual graph indexing of book abstract, and tries to solve the problem of book information indexing in conceptual graph based book retrieval model. Our work is as following:(1) Define the indexing form for conceptual graph of book abstractThis paper first constructs a canonical relation type set for conceptual graph indexing, and then extends the concept node in conceptual graph. As a result the indexing form for conceptual graph is more suitable for automatic conceptual graph indexing.(2) Propose an automatic conceptual graph indexing method for book abstract.Using syntactic and semantic knowledge on texts in specific domain, this paper combines shallow parsing technology with layered syntactic and semantic analysis to implement automatic conceptual graph indexing for book abstract. Shallow parsing recognizes basic noun phrases in sentence and labels semantic classes for them; Using syntactic and semantic templates, layered syntactic and semantic analysis first performs skeleton parsing to index top conceptual graph for the sentence, and then performs chuck parsing to index conceptual graphs for chunks. Consequently the whole conceptual graph for the sentence will be generated.(3) Experiment was carried out to evaluate our automatic conceptual graph indexing method for book abstract. The accuracy of conceptual graph indexing is 66.1%, and the accuracy coefficient is 76.7%. Experiment result shows that our method is efficient.