基于改进K-Means算法的教学反思文本聚类研究  

Research on Text Clustering of Teaching Reflection Based on Improved K-Means Algorithm

作　　者：何聚厚[1,2] 范文静[1] 

机构地区：[1]陕西师范大学计算机科学学院,陕西西安710062 [2]陕西师范大学现代教学技术教育部重点实验室,陕西西安710062

出　　处：《计算机技术与发展》2013年第11期99-102,107,共5页Computer Technology and Development

基　　金：中央高校基本科研业务费专项资金(GK201002028);国家985优势学科"教师教育创新平台"项目(GJ9850104)

摘　　要：对教学反思内容的准确评估是教师基于教学反思过程提升其专业能力的重要保障。基于改进的K-Means算法对相同主题的教学反思文本进行聚类,通过给定初始聚类中心K的取值范围使其可以在给定范围内自动增加,在聚类过程中加入相似度阈值以限定文本间相似度的取值范围,实现对教学反思文本的分类和对自我反思文本的定位。实验结果表明改进的K-Means算法在反思文本聚类的准确率和稳定性方面比传统算法有所提高,且能根据教学反思内容准确地进行自动分类。An accurate assessment of teaching reflection content is an important guarantee based on teachers＇ teaching reflection process to enhance their professional capabilities. Clustering the same theme of the teaching reflection text based on an improved K-Means algo- rithm, through given the initial cluster center K a value ranges, so that it can be automatically increased within the given range, during the clustering process, similarity threshold is introduced to limit the reflection texts＇ similarity ranges, realizing the teaching reflection text classification and the self-reflection text classification. The experiment result indicates the improved algorithm has a higher accuracy,better stability, and can accurately automatically classify according to the teaching reflection content.

关 键 词：K—Means算法 文本聚类 教学反思 相似度 均值 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]