基于K近邻分类算法的涉恐信息过滤模型研究

Research on the Risk Information Filtering Model Based on K-nearest Neighbor Classification Algorithm

[目的/意义]针对现有的涉恐信息过滤在语义过滤和主题无关性方面还存在一定的不足,建立了基于K近邻分类算法的涉恐信息过滤模型。[方法/过程]通过采用K近邻分类算法、TF-IDF分词系统和语义相似度计算建立了一种新的涉恐信息过滤模型,并从爬取主题的一次过滤到关键词匹配的二次过滤,最终进行语义分析的三次过滤,构建了一套完整的涉恐信息过滤体系。[结果/结论]本模型能够快速高效地获取涉恐信息,克服了语义和主题无关性等因素的干扰,较于传统的涉恐信息过滤在查全率和查准率上面有了很大的提升。

[Purpose/Significance]For the existing fear information filtering in the semantic filtering and topic-independence,there are some deficiencies,this paper established a k-nearest neighbor classification algorithm for the risk information filtering model.[Method/Process]By using the k-nearest neighbor classification algorithm,TF-IDF word segmentation system and semantic similarity calculation,a new fear of information filtering model is established,and from one filter to the second filter of the keyword matching,and finally the three filtering of semantic analysis,a set of complete information filtering system is constructed.[Result/Conclusion]This model can quickly and efficiently obtain the information about terrorism,and overcome the interference of semantic and topic independence.Compared with the traditional terrorist information filtering in recall and precision,there is a great improvement.