基于知识库的多层级中文文本查错推理模型  

Reasoning Model of Multi-level Chinese Text Error-detecting Based on Knowledge Bases

作　　者：吴林[1] 张仰森[1] 

机构地区：[1]北京信息科技大学智能信息处理研究所,北京100192

出　　处：《计算机工程》2012年第20期21-25,共5页Computer Engineering

基　　金：国家自然科学基金资助项目(60873013;61070119);北京大学计算语言学教育部重点实验室开放课题基金资助项目(KLCL-1005);北京市属市管高等学校人才强教计划基金资助项目(PHR201007131)

摘　　要：以往的中文文本查错研究主要针对字词错误,对句法、语义的错误推理研究不够。为此,利用统计模型和大规模人民日报语料库构建并扩充查错知识库,针对文本中字词、语法以及语义3个层次的错误,提出相应的多层级查错推理模型。设计并实现3个层级的文本查错算法,构建自动查错系统进行综合查错。实验结果表明,该系统查错性能较优,召回率达到85.62%。The previous search on text error-detecting focused on words-level,and the syntax and semantic errors reasoning are not paid enough attention.Words,syntax and semantic knowledge bases are enlarged and constructed by taking advantage of statistic model and massive People＇s Daily corpus to resolve the relevant level error reasoning,and the relevant multi-level reasoning model is put forward.The comprehensive three-level text proofreading algorithm is implemented.The constructed system can check various text errors.Experimental results show that the system has a better performance,and the recall rate is 85.62%.

关 键 词：中文文本 知识库 多层级查错 查错推理 语义查错 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]