自然语言的的处理在信息检索中的应用综述.pdfVIP

21 2 V ol . 2 1, No . 2 2007 3 J OU RNA L OF CH INESE IN FO RM AT ION PR OCESSIN G M ar., 2007 : ( 2007 0200 3511 , , ( , 1000 84 : 在信息检索 发展的过程中, 研究者们不断尝试着将自然语言处理应用到检索里, 望能够为检索效果提 高带来帮助然而这些尝试的结果大多和研究者们最初的设想相反, 自然语言处理在大多数情况下没有改进信息 检索效果, 甚至反而起了负面作用即便有一些帮助, 也往往是微小的, 远远不如自然语言处理所需要的计算消耗 那么大研究者们对这些现象进行了分析, 认为: 自然语言处理更适合于应用在需要精确结果的任务中, 例如问答 系统信息抽取等; 自然语言处理需要针对信息检索进行优化才可能发挥积极作用最新的一些进展( 例如在语言 模型中加入自然语言处理 在一定程度上印证了这一结论 : 人工智能; 自然语言处理; 综述; 信息检索 : T P39 1 : A A Survey of Natural Language Processing in Information Retrieval WA NG Canhui, ZH AN G M in, M A Shaoping ( Depar tment o f Computer Science and T echno lo gy , T sing hua U niv ersity , Beijing 1000 84 , China Abstract: N atural language pr ocessing ( N L P has been u sed in informat ion retr ieval ( IR by researchers, in the ho pe o f improv ing retr ieval eff ect . But most of t he results are in the o pposite w ay hy po thesized . In most cases, N LP didn t y ield increases in IR precisio n but took a negative eff ect . Even if N L P helped I R under some cir cum stances, the impr ov ements w ere much smaller than the pro cessing co st needed by N L P. R esear cher s perfo rm analy sis on these pheno mena and come to t he co nclusio n that : IRr elated tasks that acquir e accur ate results, such as question answ ering ( Q A and info rmatio n ex tr act ion ( I E , ar e more suited for the use o f N LP . N L P needs t o be optimized for I R in or der to be eff ectiv e. Recent r esear ch, e. g . adding NL P f acto rs to lang uage model, has mor e or less con f irmed the concl