Improved Bacterial Foraging Optimization Applied On People Opinion Analysis

As the emergence of B2 C and the popularity of online trading,there are plenty of information coming out of the Internet,which sometimes make it hard to understand and use.Furthermore,the behavior patterns of users who have different demands and purposes are not the same.Clustering analysis is one of the most effective methods to analyze behavior patterns of network users.Clustering in public opinion analysis is applied to explore the behavioral habit,requirements and likes of users,which can better help developer plan websites accordingly and further improve the experience of surfing the Internet.For instance,there are different kinds of user's behavior on some huge shopping websites,including those who randomly browse products without any clear purpose,those who browse products with certain goals and those who add products to shopping cart.Above examples are presented to indicate that website developers can understand different users' needs and mind activities,accordingly improve the structure and content of websites to encourage product promotion and sale by analyzing their behavior patterns.K-means is a kind of clustering algorithm which is used worldwide.K-means and improved K-means will always easily run into partial optimization when coping with huge date set.To deal with the problem,Swarm intelligence optimization techniques are introduced.It use instinct of all kinds of animals or insects for reference to establish a mathematical model to solve practical problem,which overcomes shortcomings of traditional classically algorithm that unable to find global optimums.An improved BFO algorithm is put forward in this paper.The original BFO algorithm will be improved to cover the shortage,which effectively improves the rate of convergence and accuracy.Some main contributions and features of the paper include:(1)An enhanced BFO algorithm is presented in the paper.The operations of chemotaxis,duplication and migration of the original BFO are to be improved.Accuracy and convergence speed are ameliorated.Different kinds of data sets are introduced to test the validity of the improved algorithm in the experiment.The selection of parameters of BFO is also improved.The experimental result of clustering of improved BFO is compared to that of other algorithms,which verifies the effectiveness of the improved algorithm.(2)The improved algorithm is applied to public opinion analysis.A model of hot topic is established and the improved algorithm is used to cluster web page.Experiments are set up to analyze and validate the effectiveness of the improved algorithm from the aspect of time and accuracy.