基于搜索的短文本分类算法研究

Search-based short-text classification

针对传统分类算法在处理短文本时的不足,提出了一种基于搜索的NaveBayes文本分类方法。该分类方法对文本数据集规模、文档长度、类别数量、分布等情况综合考虑,对朴素贝叶斯算法进行改进,将搜索技术应用到了文本分类领域。该分类算法能够更好地适用于微博、微信、短信、短语评论等短文本分类领域。并且在分类算法、分类器构造和评估3方面进行了详细的介绍。实验证明,基于搜索的文本分类器对于短文本有更好的分类效果。

For short-text classification in case the traditional classification algorithm does not work well,this paper proposes a search-based method employing NaiveBayes.The classification method is considered in the text data set scale,document length,the number of categories,distribution and so on.The NaiveBayes algorithm is improved,and the search technology is applied to the domain of text classification.This classification algorithm can be applied to the short text categorization fields such as twitter,WeChat,short message,phrase comment and so on.This paper describes the whole process,including the classification algorithms,training and the evaluation.The results indicates that the classifier has better performance comparing with other methods.