Research And System Implement Of Noninteractive Literature-based Knowledge Discovery In Chinese Literature

With the fast development of science and technology, different professional fields lead to further specialization. Supposing that implicit relations may exist in a large amount of seemingly non-relative science literature, so discovering these valuable implied connections will promote the development of science.During the latest 20 years, scientists of information science have flung themselves into this new field. A typical example is Don R. Swanson, professor of information science from the University of Chicago, who firstly demonstrate his idea of noninteractive literature-based knowledge discovery theory based on a huge amount of science literature. Later, scientists from each other country all simulate and test Swanson's ideas in succession. They all utilize the data from Medline database as their corpus.Succeeding in the Swanson's research work, this paper probes into the arithmetics of the open and closed noninteractive literature-based knowledge discovery and designs a new open and closed non-interactive literature-based knowledge discovery system based on Chinese literature, which originates from the domains of biology and medicine of CNKI(China National Knowledge Infrastructure) from the year of 1979 to 2006. Swanson's arithmetic only deals with "Title", while this paper expends Swanson's arithmetic, dealing with both "Title" and "Abstract". Also, this paper need to solve many pratical problems of dealing with Chinese literature, such as Chinese Segmentation, polysemy and so on.As to the open noninteractive literature-based KDD system, whose aim is to test whether do exist a logical connection between "migraine headache" and "magnesium" so as to estimate the feasibility of the open noninteractive literature-based KDD arithmetic. This system processes "Title" and "Abstract" respectively according to the Swanson's research approach. But the arithmetic applied in dealing with "Title" has a little difference from dealing with "Abstract". The conclusion is, in the process of dealing with both "Title" and "Abstract", we all can get the target word "magnesium" from the beginning retrieval word "migraine headache". The advantage of the arithmetic of dealing with "Title" are high-speed calculating and more efficiency, whose disadvantage is the fewer number of "B" word retrieved and the target word "magnesium", whose order lags behind compared with the arithmetic of dealing with "Abstract".Concerning about the closed noninteractive literature-based KDD system, whose aim is to validate whether do exist logical relation between "Raynaud's syndrome /fish oil" and "migraine headache/magnesium" to test the feasibility of the arithmetic of the closed noninteractive literature-based KDD. If it is feasible, the closed noninteractive literature-based KDD process can be used to prove the legitimacy of the open discovery process. This system also deals with "Title" and "Abstract" respectively in the light of Swanson's method and retrieves the correct "B" word between the start literature"A" and the target literature"C". The conclusion indicates that the result is better deal with "Abstract" than deal with "Title".In the end, this paper validates and analyses both the open and closed discovery processes, and compared the results with Swanson's results. The conclusion is, due to the minor differences between languages, concrete research approaches and subjective factors etc, the results of this paper is not completely the same as Swanson's. But Swanson's method is proved to be feasible in Chinese literature similarly.