Indexing conceptual graph for abstracts of books

Indexing a book is a prerequisite and basis for retrieval. The books or their abstracts are indexed conceptual graphs, which will be advantageous for the requirements for book retrieval based on the natural language. However, the major problem and crux for book retrieval and the application of conceptual graphs is how to construct conceptual graphs of books or their abstracts automatically. This paper begins with indexing conceptual graphs of the abstracts of technological books by handwork, which results in some frequently-used conceptual relations that are taken as a relation set for indexing conceptual graphs. Then, a conceptual terms library, which is used to help recognize the terms of text, is constructed through thesaurus, and MBL is adopted to construct the conceptual relation of a sentence. And then, an algorithm of automatically generating conceptual graphs is proposed, and a prototype system for semi-automatically indexing conceptual graphs is developed. The experiment of the prototype system shows that the proposed method of semi-automatically indexing conceptual graphs of abstract-texts has a high accuracy.