Research On Data Mining Technology Based On Ant Colony Algorithm

The phenomenon about "the data explode but knowledge is poor" caused by rapid development of information technology and network technology is serious day by day. In order to solve this problem effectively, data mining was put forward as a kind of new technology, and grew vigorously under the special environment, which demonstrated its strong vitality more and more. In some domestic primary website, data mining technology was chosen one of the most popular technologies in the future. Nowadays, some super companies such as IBM, Microsoft have attended the research of data mining technology. Data mining is a research field of crossing multi-disciplines, and its development will influence the process of the global informationization greatly. So it is the desirability of informationization development to have a research on data mining technology in an all-round and systematical way. This dissertation deeply studies and analyses the data mining technology, especially the one for clustering analysis. And then some ideas and improvements have been proposed. The main contents are listed as follows:Description of the data mining technology in brief. The basic concepts of data mining technology are introduced first. Based on the concepts, the dissertation classifies the objects of data mining in detail. In succession, the popular methods of data mining are summarized and the tasks of data mining are summed up. All of the above become the basis for this dissertation.Description of the clustering analysis in brief. As one of the most important domain of data mining, clustering analysis is mainly used to discover the valuable data distribution and data mode in the potential datum. The definition, methods, data types and standards of measuring the results are briefly introduced.Description of the ant colony algorithm(ACA) in brief. ACA is a new-type bionic algorithm which comes from the research on the biological world of nature. It is a typical representative algorithm of Swarm Intelligence and has been widely applied. The dissertation reviews the appearance and development of ACA, and explained the principle and realization of the algorithm in detail.Research on the algorithm based on ACA and clustering algorithm combination. On the basis of studies on basic clustering model, the theory of information entropy and two classical clustering analysis algorithms(LF and K-means), an algorithm of K-means based on the pheromone is presented firstly. The algorithm works with the transformation probability to realize the clustering. It has reduced the number of the parameters and improved the speed of clustering. Then, an improved LF algorithm based on entropy is introduced detailedly. At last, an algorithm based on ACA and clustering algorithm combination is proposed, which includes three courses. The first is to accomplish the task of clustering by using the improved LF algorithm based on entropy, in the second course, the datum are collected and identified with the method of SACA. At last, the result of clustering will be optimized by the algorithm of K-means based on the pheromone.