Application Research Of Granular Computing In Data Mining

Granular computing was first put forward in 1997 by T.Y.Lin, a professor in the university of San Jose State, USA, which has marked the emergence of an applied research area involving multi-subjects. Granular computing is an emerging research method to deal with information and knowledge processing which is an umbrella term to cover any theories, methodologies, techniques and tools that make use of granules in problem solving. It plays an important role in the processing for fuzzy, uncertain, partial, and huge information.Data Mining serves as an important tool for mining numerous data, and some methods have been well established. In recent years, many researchers abroad have explored and investigated the application of granular computing to data mining. They have put forward some models and methods, enabling granular computing to be widely used in data mining.The theory of granular computing, which adopts the principle of granularity to deal with and solve matters, is an innovation of some traditional ways of data mining. It designs and constructs granule or granularity according to one or more matters or theories, and it untilizes the constructed granule or granularity to compute and analyze the matters, which in the end leads to a solution.The present thesis first systematically clarifies the text clustering method and the analysis of drawing association rules. Secondly, the thesis gives a detailed introduction of the theories related to granular computing and the status of its application to data mining. Based on the study above, the thesis further puts forward a method of web document clustering based on granular computing (WDCGrc) and a method based on granular computing to efficiently obtain association rules in decision-making relation dataset.In this paper, a method of web document clustering based on the granular computing is presented. The method computes the weight value of the words in documents by adopting the TF-IDF principle. Meanwhile, combining ways defining documents threshold and average weight value are adopted to reduce dimensions and extract the keywords in each document. The method establishes the transformation between the keywords in documents and the binary granules, and adopts the algorithm of association rules based on granular computing to obtain frequent itemsets between documents. A method of the clustering is presented in the paper to obtain the clustering result. The experiment shows that the method is practical and feasible, with good quality of clustering.The paper has presented a method based on granular computing to effectively obtain association rules in decision-making relation dataset. Based on granular computing, the method constructs models for the relation dataset, utilizes equivalence class to classify the entity based on attribute so as to construct granules, and brings forward the algorithm based on granular computing to obtain association rules in decision-making relation dataset, the ways remedy part inadequate extract association rules of the current relational database.