Research Of Some Problems On Chinese Automatic Abstract Technology

In this thesis, the author first introduces the latest development of Automatic Abstract System in domestic and abroad, which shows the lack of the automatic abstract system research. Then the author introduces some basic concepts about automatic abstract system.Secondly, some basic concepts about Abstract and automatic abstract system are introduced, and the main formal models and methods of system are compared and analyzed, such as statistics based, meaning based, concept based, knowledge based etc. We induce their characteristics and put forward a kind of comprehensive automatic abstract system.Thirdly, the concepts and category of reference of noun phrase are discussed, and noun phrase reference algorithm is introduced. The author also gives the analysis results of the noun phrase reference algorithm.Finally, based on noun phrase reference algorithm and from the aspects of text macro-constitution and micro-constitution, the text topic sentence abstracting algorithm is given and the author does some trial research on the automatic abstract system. At first, we refer the text by the use of the noun phrase reference algorithm. Then, we abstract the key word collections of phases, sections and the whole text Finally, we analyze the macro-constitution and micro-constitution of the text and abstract topic sentences of the text by the use of the text topic sentence abstracting algorithm.