摘要： 通过计算数据集与样本集在主题特征向量上的相似度对数据集进行信息筛选与分类处理，以便有效地组织和分析交通管理领域内的数据资源，使查询结果分布在最相关的数据集中。利用Hadoop分布式应用程序框架使各工作节点协同完成索引的构建。根据用户查询的类别，只在最相关主题索引库中进行查找，以提高检索效率。

关键词: 主题特征向量, 分类索引, 交通管理

Abstract: To effectively organize and analyze data within traffic management, this paper makes use of computing similarity between data set and sample set topic feature vector to filter and decide which categories the data sets belong to. All work stations work together to build index with the help of Hadoop distributed application framework. According to the category of query, the algorithm only consults correlative topic index for results to achieve a better retrieval performance.

Key words: topic feature vector, classified index, traffic management

中图分类号: 

• TP391