Study And Practice Of Noninteractive Literature Based Knowledge Discovery In Medicine

With rapid development and further specification of science and technology, Connections between speciaties are overwhelmed by a far greater amount of information in various fields and cross-specialty communication becomes more difficult, which is called Fragmentation of Knowledge.The idea of noninteractive literature based knowledge discovery was advanced by professor Swanson, with a purpose of mining undiscovered public knowledge among linkages between complementary literature sets and formulating rational hypothesis to give some enlightenment to scientific practice. Swanson's idea gradually aroused interest of information researchers. They carried out explorations on the knowledge discovery process with the help of statistics and computer software. Main methods included frequency statistic analysis, semantic filtration as well as co-word clustering. Units of analysis involved words or phrases in titles, abstracts and subject headings.This dissertation consists of two main parts, the first one is a review on the idea of noninteractive literature-based knowledge discovery, which includes the emergence, development and current exploration of this information method, as well as its significance for scientific research. The second part is practice of the open knowledge discovery method in medicine. With the help of software WORDSTAT and Foxpro programming, we carried out our work by borrowing ideas from overseas research, taking MEDLINE records as data source and subject headings as analytic unit. We followed 2 main steps in the practice: first we simulated professor Swanson's first two examples using two complementary but disjoint literature pairs raynaud's disease/fish oils and migraine/magnesium deficiency through which we tried to test the feasibility and find out rules of our method; next we practiced our method with "diabetes mellitus, type 2" as source term and tried to find out valuable target terms which may be undiscovered new knowledge. The results shows that it is feasible to conducted knowledge discovery practice with our method of subject heading analysis, nevertheless the method still has some incompleteness compared with overseas related work and needs to be completed in future study.The idea of noninteractive literature based knowledge discovery has enriched information methods and brought new vigor into information science with its powerful information functions and practical significance. It opens new area for information research and complete information science. It is suggested that information researchers in our country carry out further study in this field, face difficulties and challenges bravely and positively, communicate and cooperate with scientific workers to extend the use of this method, as to give enlightment to scientific research.