改进细菌觅食算法的聚类在舆情分析中的应用

发布时间：2018-04-03 03:33

本文选题：细菌觅食优化算法　切入点：聚类　出处：《广西民族大学》2017年硕士论文

【摘要】：随着网络交易的盛行和B2C的出现,网络上出现的纷繁复杂的信息有时会让人难以理解和运用,而具有不同要求和目的的用户的行为模式也不尽相同。分析网络用户行为模式最有效的途径之一就是聚类分析。通过舆情分析中的聚类可以探寻用户的行为习惯、需求和喜好,更好的帮助网站开发者有针对性的规划网站,进而改善用户的上网浏览体验。比如在一些大型的购物网站,存在着许多不同种类的用户行为,包括无目标性的随机浏览商品的用户;具有特定网购目标的浏览用户;待购商品在购物车中的用户等。上面的举例是为了说明网站可通过分析用户的行为模式了解不同用户的需求和心理,更有针对性的改善网站的布局和内容,从而促进产品的推广和销售。K-means是一种应用颇为广泛的数据聚类算法。K-means算法以及改进的K-means算法,在对海量数据集聚类时总会面临容易陷入局部最优的问题。群智能优化技术应运而生,它是借鉴了不同动物或昆虫的各种生物本能行为,从而建立一个数学模型来解决实际问题,克服了传统经典算法无法搜索到全局最优解的缺陷。本文提出改进的细菌觅食算法,将对原有的细菌觅食算法进行改进,改进原有算法的不足,有效提高了聚类的收敛速度和准确度,本文主要的工作及特色如下:(1)本文提出了一种改进的细菌觅食聚类算法,对原算法中的趋向性操作、复制操作和迁徙操作进行改进,改善聚类精度和收敛速度。实验中将引入多种不同数据集对改进后的算法有效性进行测试,细菌觅食算法的参数选取问题也将得到改进。聚类后将实验结果与其他常用的算法对比,验证本文改进算法的有效性。(2)将改进后的细菌觅食算法应用于舆情分析,建立热度评价模型并用改进的算法对网页页面进行聚类,最后设置实验从时间和正确率等方面对改进算法的有效性进行分析和验证。
[Abstract]:With the popularity of network transactions and the emergence of B2C, the complicated information on the network sometimes makes people difficult to understand and use, and the behavior patterns of users with different requirements and purposes are also different.One of the most effective ways to analyze the behavior patterns of network users is clustering analysis.Through the clustering in the analysis of public opinion, we can explore the behavior habits, needs and preferences of users, and better help website developers to plan their websites, and then improve the browsing experience of users.For example, in some large shopping websites, there are many different kinds of user behaviors, including random browsing users, users with specific online shopping targets, users in shopping cart and so on.The above example is to show that the website can understand the needs and psychology of different users by analyzing the user's behavior patterns, and improve the layout and content of the website more pertinently.Thus to promote the promotion and sale of products. K-means is a widely used data clustering algorithm. K-means algorithm and improved K-means algorithm.The swarm intelligence optimization technique arises as the times require. It draws lessons from various biological instinctive behaviors of different animals or insects, and thus establishes a mathematical model to solve practical problems, and overcomes the defects of traditional classical algorithms that cannot find the global optimal solution.In this paper, an improved bacterial foraging algorithm is proposed, which will improve the original bacterial foraging algorithm, improve the shortcomings of the original algorithm, and effectively improve the convergence speed and accuracy of clustering.The main work and features of this paper are as follows: (1) in this paper, an improved bacterial foraging clustering algorithm is proposed, which improves the trend operation, replication operation and migration operation in the original algorithm, and improves the clustering accuracy and convergence speed.In the experiment, a variety of different data sets will be introduced to test the effectiveness of the improved algorithm, and the parameter selection of the bacterial foraging algorithm will also be improved.After clustering, the experimental results are compared with other commonly used algorithms to verify the effectiveness of the improved algorithm. The improved bacterial foraging algorithm is applied to the analysis of public opinion, and the heat evaluation model is established and the improved algorithm is used to cluster the web pages.Finally, the effectiveness of the improved algorithm is analyzed and verified in terms of time and accuracy.
【学位授予单位】：广西民族大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】