基于最大熵方法的中英文基本名词短语识别_周雅倩.pdfVIP

40 3 Vol140, No13 2003 3 JOURNAL OF COMPUTER RESEARCH AND DEVELOPMENT Mar1 2003 周雅倩 郭以昆 黄萱菁 吴立德 ( 200433) ( archzhou@ yahoo1 com) 1 Chinese T reeBank , , 7143%/ 109% / 1 , , , 1 TREEBANK Ò, 93131%/ 931 04% / , 1 , 1 ; ; TP391 Chinese and English BaseNP Recognition Based on a Maximum Entropy Model ZHOU Ya-Qian, GUO Y-i Kun, HUANG Xuan-Jing, and WU L-i De ( Dep ar tment of Co mp uter Science and Engineer ing, Fudan University , Shanghai 200433) Abstract A maximum entropy model in Chinese BaseNP recognition is used in this paper1 The open test on Chinese T reeBank, the public corpus, indicates the average recall and precision of 7143% and 109% respectively with limited know ledge ( text itself and its POS tag) 1 Because of the incomparability of Chinese BaseNP recognition results, the same algorithm is applied in English BaseNP recognition1 The test on TREEBANK Òshows that the recall and precision are 93131% and 93104% , w hich are close to the state of the art1 This not only proves the availability of the algorithm, but also indicates its language indepen- dence1 Key words maximum entropy; BaseNP; natural language processing , 1 1 (TREC) , , 1 TREC , ( ) 1 , , , , SMART 1 , ( ) 1 , 1 SMART , / 0 1 , / 0, 1 1 , , , , , 1 , , , 1 : 2001-11-09; : 2002- 11-2 : ( 69 73011) ; Intel 3 : 441 , ¹ [ 1] , 2 ( system combination) , 7 211