搜索引擎日志中“N+V”型短语分析  

PHRASE PARSING OF “N+V” STRUCTURE IN SEARCH ENGINE LOGS

作　　者：赵红改[1] 吕学强[1] 肖诗斌[1] 

机构地区：[1]北京信息科技大学中文信息处理研究中心,北京100101

出　　处：《计算机应用与软件》2012年第11期126-129,共4页Computer Applications and Software

基　　金：国家社会科学基金项目(09CYY021)

摘　　要：句法分析是自然语言处理中的一个难点和重点。基于搜狗日志语料,提出一种用层叠条件随机场模型实现搜索引擎日志中"N+V"型短语分析的方法。将短语分析分为两个阶段:"N+V"型短语内部结构分析和外部功能分析。这为"N+N+V"型短语和"N+V+V"型短语等外显型歧义结构的消歧提供了解决方法,从而为搜索引擎用短语词典构建提供基础研究服务。Parsing is a difficulty point and focus in natural language processing.Based on Sogou log corpus,in this paper we present an approach for realising the phrase parsing of ＂N＋V＂ structure in search engine logs adopting cascaded conditional random fields model.The phrase parsing is divided into two phases： the internal structure parsing and the external function analysis,both are of the phrase of ＂N＋V＂ structure.The method offers a solution to the disambiguation of the phrase structure with explicit ambiguities including ＂N＋N＋V＂ type and ＂N＋V＋V＂ type,therefore it provides a basic service to the study on constructing the phrase dictionary used by search engine.

关 键 词：“N+V”型短语内部结构 “N+V”型短语外部功能 层叠条件随机场模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]