多维度特征的用户查询意图自动分类  

Automatic Identification of Query Intent by Using Multiple Features

作　　者：蒋宗礼[1] 张恒庆[1] 

机构地区：[1]北京工业大学计算机学院,北京100124

出　　处：《电脑与信息技术》2015年第1期1-4,共4页Computer and Information Technology

基　　金：国家级教学团队建设项目(项目编号:00700054J1901)

摘　　要：为了提高搜索引擎查询结果的质量,越来越关注于对用户提交的网络查询意图的识别。基于查询session对用户提交的查询进行多维度特征提取,尽量能全面系统地描述查询分类特征,并使用SVM进行分类。实验结果表明通过结合查询的多个特征有助于识别查询意图,在人工标注的测试集中对查询意图分类的正确率达到80%。In order to improve the quality of ranking result from search engine, increasing attentions are paid to identify the underlying user intents. In this paper, we identify user＇s query intent with the multidimensional feature extracted from the queries based on sessions. Then classify by using SVM. Experiment result show that by combing multiple features help identify user intent, classification accuracy is up to 80% in our test data set manually annotated.

关 键 词：搜索引擎 查询意图 查询分类 分类特征 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]