Research On Improved Ant Colony Clustering Algorithm

Data clustering is an important data mining technology,is a effective means that people understand and explore the intrinsic link between things.It can be used as an independent data mining tools and found the data-depth information of database distribution.It also can be used as other data mining preprocessing algorithm steps.In the field of engineering and technology,it applies wildly.In recent decades,researchers at home and abroad,put forward a number of clustering algorithms,trying to find the best program.With the rise of ant colony algorithm,it was found that in certain aspects of the use of ant colony clustering model of the cluster closer to the actual problem.This paper has analyzed the domestic and foreign research present situation of the clustering analysis and ant colony algorithm.Cluster Analysis is a very active area of Data Mining,mainly found the meaningful data in the data distribution and data model that used in the implied data.In this paper,the definition of cluster analysis,clustering methods,data types,as well as the results of clustering metrics are briefly introduced. This paper has researched on the ant colony algorithm.It simulated the swarm intelligence and played a very good role of settling optimization of treatment and studied the basic model of ant colony clustering analysis and ant colony clustering analysis of the basic model of the LF algorithm and analyzed the advantages and disadvantages of its algorithm.The thesis mainly focuses on studying clustering analysis based on ant colony algorithm and its application.The main work includes:First,Improved LF algorithm based on pheromone was present.LF algorithm demanded to set up a lot of parameters,which are more sensitive.At the same time as a result of the definition of the ants in the two-dimensional grid is arbitrary movement and any movement in certain areas is no significant,for example,those fundamental data object is not regional,so the effect of clustering algorithms is poor and efficiency is not high.By modifying the algorithm to improve the LF groups,the similarity function by adding parameters to adjust the adaptive strategy,the use of short-term memory and grid of the local pheromone distribution of the random movement control ants,combined with the speed of the dynamic changes ants,the radius increase,such as mandatory down characteristics of a pheromone-based algorithm to improve the LF.Second,This paper designed and implemented a simple ant colony algorithm platform based on cluster analysis.Also via compare experiment data testing and vary algorithm analysis,the algorithms shows preferable performance.Third,This paper proposed the ant colony clustering structure of the document mining system.In a typical process the document based on the excavation,we analyzed and designed of the ant colony clustering mining the overall structure of documents and document segmentation subsystem,the document feature vector computing subsystem and subsystem structure of cluster analysis of ant colony.