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.
出处
《计算机应用与软件》
CSCD
北大核心
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)