Interdiscipline Knowledge Discovery And Its Visualization Research

The development of science and technology not only brings new opportunities, but also rigorous challenges to each discipline. The most common approach to disciplinary innovation is to introduce, absorb and integrate theories, methods and techniques of other disciplines. Currently, the interdisciplinary research has become the focus of attention from the academia, and even the whole society. However, from analysis on status quo of domestic and foreign researches, interdiscipline research still stays at the superficial or empirical level and lacks of deep mining research based on scientific literatures clustering analysis.The problem to be solved in the paper is how to effectively discover and exhibit the inter-relationship of disciplines by means of clustering analysis. The research is aimed to deeply mine the potential interdisciplinary knowledge so as to provide a scientific, objective and advanced method for interdisciplinary research through cluster analysis techniques based on massive scientific literature data. The paper takes interdisciplinary knowledge as the research object and starts with the analysis of text mining theory and key technique to carry out the research in such 6 aspects as the study and improvement of document clustering algorithms as well as interdisciplinary knowledge discovery and visualization, i.e. (1) Summarize the progress of domestic and foreign interdisciplinary researches, point out deficiencies of the research and bring forward the feasibility of interdiscipline research baded on document clustering; (2) Sum up the research actuality of text mining, intelligence optimization algorithm and information visualization in a systematic way; (3) Make a research into key techniques of document clustering, analyze document clustering problems such as similarity precision, high-dimensional reduction and fuzziness of clustering numbers and propose corresponding solutions; (4) Make a research into document clustering algorithms. Start with the fundamental algorithm FCM; apply the optimization algorithms such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), etc. to document clustering; make improvement of PSO mainly and put forward a document clustering algorithm based on immune selection particle swarm (IS-PSO) integrated with FCM; (5) Make a research into interdisciplinary knowledge discovery and visualization model. Propose the interdisciplinary literature discovery model, interdisciplinary knowledge mining model and interdisciplinary knowledge visualization model and design the view template for visualizing interdisciplinary knowledge; (6) Demonstration research:Make use of Chinese Core Journal Literature Data of the recent decade (From 2000 to 2009) of informatics and computer science to study the interdisciplinary knowledge; make use of improved document clustering algorithm to analyze documents; make use of VIK, a developed independently interdisciplinary knowledge-based visualization software as well as other visualization software CiteSpace and UCINET to conduct visualization research and analysis. The detailed research chiefly adopts literature research method, survey research and expert consultancy method, statistical analysis method, experimental simulation method and Meta analysis method, etc.The main results (conclusions) from this paper are:(1) The inter-relationship among disciplines can be effectively identified through massive literature clustering analysis; (2) The document similarity model based on weighted keywords and abstract and keywords matching model based on partially similar strings can effectively improve the document similarity; (3) The document clustering algorithm, I-PSO integrated with FCM can be applied to mass data clustering effectively; (4) The co-word-based interdisciplinary literature discovery model can accurately discover interdisciplinary literatures; (5) The interdisciplinary knowledge discovery model based on co-word clustering analysis can mine more knowledge about interdiscipline, such as extent of disciplinary crossing, extent of integration, interdisciplinary intersection and new growing point; (6) The interdisciplinary knowledge-based visualization model makes the visualization of interdisciplinary knowledge possible; (7) From demonstration research into interdisciplinary knowledge of informatics and computer science, it can be found that over the decade's development, two disciplines can be studied in such factors as data mining, information security, search engine, ontological technique, information retrieval as well as software engineering and image processing, etc. In the future, image retrieval, domain ontology and personalization will become their new growing points; (8) The developed independently interdisciplinary knowledge visualization software VIK(Visualization of Interdisciplinary Knowledge) can show the interdisciplinary knowledge directly.In conclusion, the major innovations involve:(1) The application of clustering analysis technique to interdisciplinary research and it provides a feasible method for interdisciplinary research; (2) The integrated document clustering algorithm based on IS-PSO and FCA is brought forward; (3) The Interdisciplinary knowledge discovery and visualization model; (4) VIK, a visualization platform of interdisciplinary knowledge.