用户查询意图的层次化识别方法  

A Hierarchical Framework for User Intention Recognition

作　　者：唐静笑 吕学强[1] 柳成洋[2] 李涵[2] 

机构地区：[1]北京信息科技大学网络文化与数字传播北京市重点实验室,北京100101 [2]中国标准化研究院,北京100191

出　　处：《现代图书情报技术》2014年第1期36-42,共7页New Technology of Library and Information Service

基　　金：国家自然科学基金项目"基于本体的专利自动标引研究"(项目编号:61271304);北京市教委科技发展计划重点项目暨北京市自然科学基金B类重点项目"面向领域的互联网多模态信息精准搜索方法研究"(项目编号:KZ201311232037)的研究成果之一

摘　　要：【目的】向搜索引擎提交的查询均有其潜在的查询意图,准确识别查询意图可以提高查询的效率。【方法】针对有明显意图的查询,采用滑动窗口寻找最大公共子串的策略抽取用户的意图模板,然后用模板匹配的方法识别用户查询意图。对无明显意图的查询,采用多特征融合的分类方法进行识别。【结果】实验结果表明,采用层次化识别方法和单独使用分类器方法相比,识别查询意图的实验结果正确率得到19.04%的提升。【局限】可获得的意图模板是有限的,因此显式意图查询的识别存在局限性。大规模数据情况下,模式匹配及机器学习算法的运算量很大,需要进一步优化算法。【结论】实验证明该方法在Web意图识别中是有效的,对意图识别率的提高有积极意义。[Objective] Any query search engine has its potential query intention, and accurate intention identification can improve the efficiency. [Methods] For the explicit intent queries, the authors employ sliding window strategy to find the maximum common substring for extracting user intent templates and then use the templates to identify the user intention. For implicit intent queries, the authors use a multi-feature integration method to build classifier for the final query intention recognition. [Results] Experimental results show that the hierarchical intention recognition framework can achieve better precision comparing with methods based on classifier, and the accuracy enhances 19.04%. [Limitations] Intention template obtaining is limited, so explicit intention recognition has limitation. For large-scale data, complexity of the pattern match and machine learning algorithm is very high, the algorithm need further optimization. [Conclusions] Experiment shows that this method is valid in Web intention recognition, which has a positive significance for improvement of intention recognition rate.

关 键 词：信息检索查询意图 层次化识别 意图模板 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]