Research Of Clustering Analysis Based On Swarm Intelligence Algorithm

In order to effectively solve the clustering problem in accuracy and stability, the paper focuses on two swarm intelligence algorithms: Particle Swarm Optimization(PSO) and Artificial Bee Colony(ABC) algorithm. We conduct relevant research and improvement.These works mainly include the following two aspects:1. A text clustering algorithm of particle swarm optimization with environmental factors constraints is proposed. Environment has two important characteristics in biology. It can strengthen the characteristics of the species and it also can limit the range of the activity of species. Characteristics of species can well distinguish species categories. Based on this idea,the environmental factors will be introduced into the clustering algorithm of particle swarm optimization, which can be used to enhance the characteristics of the particles and can effectively classify text data objects. In addition, since the global best information can possible bring the side effects, namely the global best particle itself locates at the local optimum, so the whole particle swarm algorithm is divided into two stages. The first stage will be not used the global information and take advantage of local search ability of particles within a limited space in their meticulous search. The second stage will re-introduce the global best information, accelerating algorithm optimization process and the convergence rate.2. A strengthen search strategy ABC clustering algorithm is proposed. In standard ABC clustering algorithm, employed bee and onlooker bees use the same update equation of the food source position. In this paper, for onlooker bees, a new search strategy can be proposed.Such intermediate results that the average of data objects belong to the cluster center will be used to feed back to the clustering algorithm and strengthen the characteristics of the food source, improving its search intensity to tap the diversity of food sources. Meanwhile,inspired by PSO, the global information will be introduced into the improved ABC algorithm and improve the process of the improved ABC algorithm.