¾÷¾¹Â½Ä¶ªº³Ì·sµo®iÁÍ¶Õ ±i´º·s ¡E ³¯²Q®S ¤Þ¨¥ ¦b°ê»Ú¥æ³q¤é¯qµo¹Fªº¤µ¤é¡A¦a²z¤Wªº¹j»Ò¤w¤£¦Aºc¦¨¤£¦P°ê®a ©Î¥Á±Ú¤§¶¡ªºÂE·¾¡C¨ú¦Ó¥N¤§ªº¡A«h¬O¤HÌ©¼¦¹·¾³q©Ò¥Îªº»y¨¥ ¡C¦b¸ê°T§Ö³t¿±µÈ¡B°ê»Ú¶T©ö¿³²±ªº®É¥N¡A¯à§Ö³t´x´¤¤j¶q¥~¤å ¸ê°T¡A¤]´N¬Û¹ï¦a¼W¥[¤@¥÷¤O¶q¡C¦]¦¹¡A§Q¥Î¾÷¾¹Â½Ä¶¨t²Î¨Ó¨ó §U¤ḨֳtÀò¨ú¸ê°T¡A¤w¦¨¬°¥²µMªºÁͶաCµM¦Ó¡A¶Ç²Îªº¾÷¾¹Â½ Ķ¨t²Î¡A¥Dn¬O¨Ì¾a¤H̫ؤJ¤j¶q½ÆÂø¦Óº¾¸Hªº»y¨¥ª¾ÃÑ¡A¨Ó«Ø ¥ß¨t²Îªºª¾ÃÑ®w¡C³oºØ»Ýn¤j¶q¤H¤u¤¶¤Jªº¤èªk¡A¹ï©ó¶}µo¹ê»Ú ªº¤j«¬¾÷¾¹Â½Ä¶¨t²Î¦Ó¨¥¡A¤w±Á{ÄY«ªº¦ÒÅç¡C¥»¤å±N²³æ¦a¬° ±z¤¶²Ð³Ì·sªº¾÷¾¹Â½Ä¶¨t²Îµo®iÁͶդάã¨s¤è¦V¡A¨Ã¦V±z¤¶²Ð¥Ñ °ê¤H¦Û¦æ¶}µoªº¤@®M¾÷¾¹Â½Ä¶¨t²Î¡C ²¤¶ ©Ò¿×¡y¾÷¾¹Â½Ä¶¡z (Machine Translation, MT)¬O«ü§Q¥Î¾÷¾¹ ( ¹q¸£) ªº§Ö³t³B²z¯à¤O¡A¨Ó¨ó§U¤H̱q¨Æ¤å¥ó½Ķªº¤u§@¡C¾÷¾¹ ½Ķ¨t²Î¯à§â¤HÃþ»y¨¥Â½Ä¶ªºªk«h¡AÂàÅܦ¨¹q¸£ªº¹Bºâªk«h¡A¨Ï ±o¹q¸£¯à®Ú¾Ú¹Bºâªk«h¡A±N¿é¤Jªºì©l»y¨¥(source language) ½Ķ¦¨©Ò»Ýªº¥Ø¼Ð»y¨¥ (target language)¡C ¥Ñ©ó»y¨¥ªºl¥Í¯à¤O«D±`±j¡A¦]¦¹¡A¥i¯à²£¥Íªº¥y¤l¤dÅܸU¤Æ¡C ¤@ÓºÙ¾ªº¾÷¾¹Â½Ä¶¨t²Î¡A¥²»Ý¨ã¦³¤@©wªº§PÂ_¯à¤O¡A¤~¯à°w¹ï ¦UºØ¿é¤J¤å¥y¥[¥H½Ķ¡C§{¶¡¦³³\¦hºÙ¬°Â½Ä¶¾÷ªº¤p«¬pºâ¾¹¡A °ò¥»¤W¥u¯àºÙ¬°¹q¤lµü¨å©Î±`¥Î·|¸Ü®w¡A¦pªG±z§â³o¨Ç¾÷¾¹¤]ºÙ ¬°¾÷¾¹Â½Ä¶¨t²Îªº¸Ü¡A¨º¥u¯à©Ç³o¨Ç°Ó¤H¼s§i¥´±o¤Ó¼F®`¤F¡C ¦b¸ê°T§Ö³t¿±µÈªº¤µ¤Ñ¡A°ê»Ú¶¡ªº·¾³q¤éÁÍÀWÁc¡C¦]¦¹¡A§Ö³t³B ²z¤j¶qªº¥~¤å¸ê®Æ¡A¤w¬O¤@ºØ´¶¹Mªº»Ý¨D¡C¦b³oºØ»Ý¨D±a°Ê¤§¤U ¡A¥Î¾÷¾¹Â½Ä¶¨t²Î¨Ó¨ó§U¤Ḩֳt½Ķ¡B«ØÀÉ¡A¤]´N¦¨¬°µLªkÁ× §KªºÁͶաC¦Ó¾÷¾¹Â½Ä¶¨t²Î©Òl¥Í¥X¨Óªº»y¨¥³B²z§Þ³N¡A¤]¥i¥Î ¦b¤@¨Ç¥ý¶iªº¤å¦r³B²zÀ³¥Î¤W¡C¦]¦¹¡A³oºØ§Þ³N¹ï±N¨Ó¦ÛµM»y¨¥ ¤H¾÷¤¶±¡A¤D¦Ü©ó¤f»y³B²z (spoken language processing)ªº µo®i¡A³£¦³¨M©w©Êªº¼vÅT¤O¡C ¾÷¾¹Â½Ä¶ªº¾ú¥v ¾÷¾¹Â½Ä¶¨t²Îªº¬ã¨s¡A¨ä¹ê¤w¦³¼Æ¤Q¦~ªº¾ú¥v[Huct 86]¡C¦b¹q ¸£µo©ú¤§ªì¡A´N¦³¤H·Q¨ì§Q¥Î¹q¸£¨Ó¯}¸Ñ±K½X¡A·íµM¤]¦³¤H·Q¨ì §Q¥Î¹q¸£¨Ó¯}¸Ñ¤H»P¤H¤§¶¡·¾³qªº±K½X¡Ð¡Ð»y¨¥¡C¦´Á¬ã¨s¤Hû ¶}©l¬ã¨s¾÷¾¹Â½Ä¶¨t²Î®É¡A¤H̹ï©ó¾÷¾¹Â½Ä¶¨t²Î´¶¹M¦s¦b¬Û·í °ªªº´Á±æ¡A§Æ±æ¯àÂǵ۹q¸£ªº³B²z¯à¤O¡A¹F¨ì¥þ¦Û°Ê (Fully Automatic)¡B°ª«~½è (High Quality)½Ķªº¹Ò¬É¡F³oºØ¥þ¦Û°Ê ¡B°ª«~½èªº¾÷¾¹Â½Ä¶¹Ò¬É¡A²ºÙFAHQMT (Fully Automatic High Quality Machine Translation),¦b·í®É¦¨¬°¬ã¨s¤Hû¤@Pªº¹Ú ·Q¡CµM¦Ó¡A¾÷¾¹Â½Ä¶©Ò¯A¤Îªºª¾Ãѽd³ò¬Û·í¼sªx¡A°£¤F¦UºØ¹q¸£ ºtºâªkªºµo®i¤§¥~¡AÁÙ²o¯A¨ì»y¨¥¾Ç¡B¤ß²z¾Çµ¥»â°ì¡C ´N¨t²Î¾ã¦X¦Ó¨¥¡A§ó²o¯A¨ì¤j«¬ª¾ÃÑ®w¤§ª¾ÃѪº©â¨ú¡A¤@P©Êªº ºûÅ@µ¥¹ê»Úªº¤uµ{°ÝÃD¡C¦p¦ó¾A·í¦a¿ï¾Üµü·J¡B¥Î¦X¾Aªºªí¹F¤è ¦¡¥¿½T´yz»y¨¥ªº±±¨î¾÷¨îµ¥¡A³£¤£¬O¤HÌ©Ò·Q¹³ªº¨º¼Ë®e©ö¡C ¨Ò¦p¡A"go"³oÓ¦r¦bµü¨åùؤj¬ù¦³¤»¤Q´XºØ¤£¦PªºÂ½ªk¡A¦b¤°»ò ±¡ªp¤UÀ³¦p¦ó½¡H¬°¤°»ò·|³o»ò½¡H³o¼Ëªº³W«h¬O§_¥i¥H©µ¦ù¨ì ¨ä¥¦¦rµüªºÂ½Ä¶¡H³o¼Ëªº³W«h·|¤£·|¸ò¨ä¥¦ªº³W«h¬Û½Ä¬ð¡H®£©È ±z°Ý¤QÓ¬ì¯Z¥X¨ªº±M®a¡A³£¥i¯à±o¨ì¤QºØ¤£¦Pªº¦^µª¡C ¤@¯ë¦Ó¨¥¡A»y¨¥¥»¨¨Ã¤£¹³¤HÌ©Ò·Q¹³ªº¨º»ò³W«h¡A¶Ç²Î»y¨¥¾Ç ©Ò«µøªº»y¨¥²{¶H¤]µLªk²[»\³\¦h²Óº¾ªºª¾ÃÑ¡A¾ÉP¥Î¨Ó³B²z¹ê »Ú°ÝÃD®É¡A±`±`¥ª¤ä¥k²Ò¡F¦Ó³oºØ²Óº¾ªºª¾ÃÑ·¥Ãø¥Î¦³®Äªº¤è¦¡ ©âÂ÷¥X¨Ó¡A¨Ï±o»y¨¥ª¾ÃѪº©â¨ú(knowledge acquisition)¦¨¤F µo®i¾÷¾¹Â½Ä¶ªº¤@¤j²~ÀV¡C¦]¦¹¡A¦´Áªº¬ã¨s¦¨ªG¨Ã¤£²Å¦X¤HÌ ¹ï¾÷¾¹Â½Ä¶ªº´Á±æ¡C³oºØ´Á±æªº¸¨®t¡A¨Ï±o·í®É¾÷¾¹Â½Ä¶µo®iªº ¤j¥»Àç¡Ð¡Ð¬ü°ê¡Ð¡Ðªº¤@Ó½Õ¬d©eû·|¡Aµo¥X¤@¥÷·¥¤£§Q©ó¾÷¾¹ ½Ķµo®iªº³ø§i¡A¥HP¤ä´©¾÷¾¹Â½Ä¶µo®iªº¸g¶O´X¥G§¹¥þ³Q¨ú®ø ¡C¾÷¾¹Â½Ä¶ªºµo®i¹E¶i¤J¾ú¥v¤Wªº¶Â·t´Á¡C ªñ¦~¨Ó¡A¥Ñ©ó¹q¸£³B²z¸ê®Æªº¯à¤O¤j¼W¡A¨Ï±opºâ»y¨¥¾Ç®a¯à³v º¥¸õ¥X¥H©¹¥u±q»y¨¥¾Ç¨¤«×¤Î¤HÃþ¾Ç²ß»y¨¥¼Ò¦¡ªº®Ø®Ø¤¤¸õ¥X¡A «i´±¦a¹Á¸Õ¤@¨Ç·sªº§@ªk¡F¦Ó¬ÛÃö»â°ìªº¬ã¨s§Þ³N¤éÁͦ¨¼ô¡A¤] «P¨Ï¾÷¾¹Â½Ä¶¨t²Îªº¬ã¨s¡A«·s¦b¥@¬É¦U¦a½´«k®i¶}¡C¾¨ºÞ»y¨¥ ªº³W«ß¤´¬O¨º»ò½ÆÂø¡A¤j³¡¥÷¼ôª¾¾÷¾¹Â½Ä¶¥H©¹¥¢±Ñ¸gÅ窺¬ã¨s ¤Hû¡A¤w¯à¥Î¤ñ¸û¦¨¼ôªºÆ[ÂI¨Ó¬Ý«Ý¾÷¾¹Â½Ä¶¡A¹ï¥¦ªº´Á±æ¤]§ï Åܬ°¸û¤Á¹ê»Úªº¾÷¾¹»²§U½Ķ (Machine-Aided Translation, MAT)¤è¦¡¡A¦Ó«DFAHQMT¡C¤H̶}©l¤F¸Ñ¨ì¡A¾÷¾¹Â½Ä¶§Y¨Ï¤£¯à§@ ¨ìFAHQªº¹Ò¦a¡A¥un¯àÀ³¥Î±o·í¡A¤´¥i§êºt±j¦Ó¦³¤Oªº»²§U©Ê¨¤ ¦â¡A¨ó§U¤H̦b¬Ù®É¬Ù¤Oªº±¡ªp¤U¡AÀò±o»P¤H¤u½Ķ¬Û·íªº«~½è ¡A¨Ã¤£»Ýn±j¨D©Ò¿×ªºFAHQMT¡C¦Ü©ó¾÷¾¹Â½Ä¶»P¤H¤u½Ķ¤§¶¡ªº ®t¶Z¡A¥i³z¹L±M·~ªºÄ¶«á×µ¤Hû (post-editor)¡A±N¨ä®t¶ZÀ± ¸É¨ì»P¤H¤u½Ķ¬Û¦Pªº«~½è¡C¦p¦¹¤@¨Ó¡A¦b¾÷¾¹Â½Ä¶¨t²Îªº»²§U ¤§¤U¡A¶Ç²ÎªºÂ½Ä¶¤u§@¥i´îÁY¨ì¥u³Ñ¼f®ÕĶ½Zªº¤u§@¡A½Ķªº³t «×¦ÛµM¤j´T´£°ª¡A¦Ó½Ķªº¦¨¥»¤]¥i±æ°§C¡C¨Ï«È¤á¯à¥H¸ûµuªº ®É¶¡¡AÀò±o»P¤H¤u½Ķ«~½è¬Û·íªºÄ¶½Z¡A¦b»P®É¶¡Ävª§ªº¸ê°TªÀ ·|¤¤¡Aűo¥ý¾÷¡C µM¦Ó¡A¦b³\¦h¹ê»ÚÀ³¥Î¤¤¡A¨Ï¥ÎªÌ©Î«È¤á¤´ÂÂn¨D°ª«~½èªºÂ½Ä¶ µ²ªG¡A¬ã¨s¤Hûªº°ò¥»°²³]¤Î¨Ï¥ÎªÌ¶¡ªº»Ý¨D¡A¤´¦³¬Û·íªº®t¶Z ¡F¦p¦ó¹ï»Ý¨D¸û°ªªº¨Ï¥ÎªÌ¡A¦b¨t²Î¤¤´£¨Ñ¦^õXªººÞ¹D¡A¥H´î¤Ö Ķ«á×µ¦¨¥»¡A³£¬O¬ã¨sµo®i³æ¦ì©Ò¥²»Ý±¹ïªº°ÝÃD¡C °ê¤º¦³Ãö¾÷¾¹Â½Ä¶ªº¬ã¨s»F©l©ó1985¦~¥Ñ²M¤j¹q¾÷¬ã¨s©ÒĬ§J¼Ý ±Ð±Â©Ò»â¾ÉªºArchTran¬ã¨s¤p²Õ¡]ArchTran²{¤w§ïºÙ BehaviorTran [Su 87, 90, Chen 91]¡^¡A¨ä«á¦³¤u¬ã°|¹q³q©Ò ¡B¥x¤j¡B¥æ¤j¡BªQ¤Uµ¥³æ¦ì¬ÛÄ~§ë¤J¾÷¾¹Â½Ä¶ªº¬ã¨s¡C¨Ã¥Ñ©ó³o ªÑ¬ã¨s·¼éªº±a°Ê¡A¦b¤¤¬ã°|¡B¥x¤j¡B¥æ¤j¡B²M¤j¼Æ¦ì±Ð±Â¤ÎĬ ±Ð±Âªº¦X¤O±À°Ê¤§¤U¡A©ó1989¦~¶}©l¡A¥ý«á¦¨¥ß¤F¤¤µØ¥Á°êpºâ »y¨¥¾Ç·|¡A¤ÎSIGMTµ¥ÁõÄÝ©ópºâ»y¨¥¾Ç·|¤Uªº¯S®í¬ã¨s¸s (Special Interest Groups, SIG's)¡A¥þ¤O±À¼spºâ»y¨¥¾Ç¬ã¨s ¡A¨Ã±µ¨ü¬F©²³æ¦ì¡A¦p¤¤¥¡¼Ð·Ç§½©e°U¡A°Ñ»P°ê¥~¬ÛÃö·|ij¡C¥Ø «epºâ»y¨¥¾Ç·|¨C¦~©w´ÁÁ|¿ì¤@¦¸¾Ç³N·|ij¡A·|û°Ñ»P°ê¥~¬ÛÃö ¾Ç³N·|ijªº±¡ªp¤]¬Û·í¿ãÅD¡C¥H¥h¦~ (1992)¸Ó¾Ç·|·|ûµoªí¦b ¥þ¥@¬É³Ì¤jªºpºâ»y¨¥¾Ç·|ijCOLING-92ªº½×¤å½g¼Æ¦Ó½×¡A¥Ø«e ¥xÆW¬ÛÃö¤§¬ã¨s¡A¦b¨È¬w¦a°Ï¤w¦¨¤F¶È¦¸©ó¤é¥»ªº¬ã¨s«Âí¡C³o ¨Ç¦¨´N¸ò°ê¤º¾÷¾¹Â½Ä¶¬ã¨sªºµo®i¦³µÛ¬Û·í±K¤ÁªºÃö«Y¡C BehaviorTranªº¬ãµo¬¡°Ê¨´¤µ¤w¦³8¦~ªº®É¶¡¡C¨Ã¤w©ó1989¶}©l ¦bP»·¬ì§ÞªºÂ½Ä¶¤¤¤ß¹BÂà¡A´À³\¦h¦b¥xªºµÛ¦W°ê»Ú¯Å³nµwÅé¹q ¸£¤½¥q´£¨Ñ½ĶªA°È¡A¦P®É³z¹L¹ê»Úªº¹B§@¸gÅç¡A¤£Â_¼W±j¨t²Î ¥\¯à¤Î³Ð·s¨t²Î¬[ºc¡C¦U¦ìŪªÌ¤]³\¤w¸gŪ¹L¤@¨Ç¥Ñ¾÷¾¹Â½Ä¶¨t ²Î©Ò½Ķªº¨Ï¥Î¤â¥U¦Ó¤£¦Ûª¾©O¡I ¥Ñ©óBehaviorTran¬ã¨s¤p²Õªº¾Ç³NI´º¤Î²z©À¡A°£¤F¾÷¾¹Â½Ä¶¨t ²Îªº¹ê»Ú¹B§@¤§¥~¡A¦b¾Ç³N¬ã¨s¤Î¦ÛµM»y¨¥³B²z§Þ³N¤Wªº±À°Ê¤] »á¬°¥Î¤ß¡A¨´¤µ¤w¦b°ê¤º¥~¾÷¾¹Â½Ä¶©Î¦ÛµM»y¨¥³B²zªº±M·~½×¤å ¶°¤Î´Á¥Z¤W¡Aµoªí40¦h½gªº½×¤å¡A¥§¡¨C¦~Á`¦³4¡B5½g¬ÛÃö½×¤å ²£¥Í¡C¦]¦¹¡A¦b°ê»Úªº¾÷¾¹Â½Ä¶¬ãµo³æ¦ì¤¤¡A¤w¨ã¦³¬Û·íªºª¾¦W «×¡C¥»¤å©Òn¤¶²Ðªº¡A´N¬OBehaviorTranªº¨t²Î¬[ºc¡A¤Î¤@¨Çªñ ¦~¤~µo®i¥X¨Óªº¥ý¶i³]p·§©À¡C ¾÷¾¹Â½Ä¶»P¤H¤u½Ķ ¾÷¾¹Â½Ä¶¬Û¹ï©ó¤H¤u½ĶªºÀu¶Õ¡A¥Ñ¨âªÌ¬yµ{ªº¤ñ¸û¡A§Y¥i²¤¿s ¤@¤G¡C¡]½Ð°Ñ¾\¹Ï¤@¡^¡C <<¹Ï¤@¡G¾÷¾¹Â½Ä¶»P¤H¤u½Ķ¬yµ{¤ñ¸û>> ¶Ç²ÎªºÂ½Ä¶¤è¦¡¡A¤jÅé¤W¬O¤@ºØ¥Ñ½Ķ¤Hû¦b½Z¯È¤W¤£Â_¦a«½Æ ®Ñ¼g¡B¶î§ï¡B¬d¾\µü¨å¡BÁýZµ¥°Ê§@ªº¬yµ{¡F¦¹ºØ¬yµ{·¥¬°¯Ó®É ¶O¤O¡A§Y¨Ï¬O¬Û·í©T©wªº¥y«¬¡A³oºØ¯Ó®É¶O¤Oªº°Ê§@¤´¥²»Ý¤@¦A ¦a«½Æ¡C¼f½Z¤§«á¤S»Ýn¦³±Mªù¤Hût³d¤¤¤å¥´¦r¡A¸g¹L¼Æ¦¸®Õ ¹ï¡A³Ì«á¤~¯à¶i¤J¦L¨ê¼t±Æª©¡B¦L¨ê¡A§¹¦¨¤@¥÷¥Xª©«~¡C ¤ÏÆ[¾÷¾¹Â½Ä¶¨t²Î¡A¦Û¤å¥ó±½´y¥H¨´¹q¸£±Æª©¡A³£¬O¦b¹q¸£ªºÂ½ Ķ¤Î½s¿èÀô¹Ò¤¤§¹¦¨¡C¦³¨}¦nªº¤å¥ó±½´y³nÅé¡B¤¤^¤å¤å½Z®Õ¥¿ µ{¦¡¡B¾÷¾¹Â½Ä¶¨t²Î¡B¹q¸£±Æª©³nÅéµ¥¾ã¦XÀô¹Òªº¤ä´©¡C°£¤Ö¶q ªº¤H¤u¤¶¤J¥~¡A¨ä¥¦³¡¥÷³£¥i¥H¥Ñ¹q¸£¦Û°Ê°õ¦æ¡C§Y¨Ï¤H¤u×¹¢ ³¡¥÷¤´»Ý¤H¤O¤¶¤J¡A¤]¬O¦b¤@Ó¾ã¦Xªº½sĶÀô¹Ò¤U§@·~¡A¤H¤Oªº ªá¶O¥i»¡¬Û·í¤Ö¡C¦]¦¹¡A¦b¾÷¾¹Â½Ä¶¨t²Îªº»²§U¤§¤U¡A¾ãÓ½Ķ ¬yµ{¥u³Ñ¤UĶ«á×µ¤Î¼f½Zªº¤u§@¡A¨Ï±o¾ãÅé³t«×¤ñ¤H¤u½Ķ§Ö ¡A¦¨¥»¤]¸û§C¡C ¾÷¾¹Â½Ä¶ªºÀuÂI »P¤H¤u½Ķ¬Û¤ñ¸û¡A¾÷¾¹Â½Ä¶¥DnªºÀuÂI¥i¥HÂkµ²¨ì¥H¤U´XÂI¡G¡@ ¡´ ½Ķ³t«×§Ö ¾÷¾¹Â½Ä¶¤£¶È³t«×§Ö¡A¦Ó¥B¥i¥H¤@¤Ñ24¤p®É¤£°±¦a¤u§@¡A«O«ù¤@ ©wªº®Ä²v»P«~½è¦Ó²@µL«è¨¥¡C¦]¦¹¡A¥i¥H¦bµu®É¶¡¤ºÀò±o¤j¶qªì ½Z¡A¾ãÅé³t«×¶È¨ü©óĶ«á×µ³t«×¡C¦Ó½Ķªº®Éµ{¤]¸û¤H¤u½ Ķ®e©ö¹w¦ô¡A¹ï©ó»Ýn§Ö³t¡B¤j¶q½Ķªº¾÷ºc¦Ó¨¥¡A¥i¥H´£¨Ñ³Ì ¨Îªº¿ï¾Ü¡C ¡´ ¦¨¥»§C·G ¥Ñ©ó¤H¤u¦¨¥»·U¨Ó·U°ª¡A½Ķ¤H¤~Ãø¨D¡A¨ã¦³¯S©w±M·~ª¾ÃѪºÂ½ Ķ¤Hû§ó¥[µ}¤Ö¡A¦]¦¹¡A¤H¤u½Ķ¦¨¥»³vº¥´£°ª¬O¥²µMªºµ²ªG¡A ¦Ó¨ä²£¯à«o¥¼¥²µ¥¶q¼W¥[¡A¹ï§Ö³t¿±µÈªº¸ê°T¦Ó¨¥¡A¥²µMÃø¥HÀ³ ¥I¡F¤Ï¤§¡A¹q¸£»ù®æ¤é¯q¤U°¡A¹Bºâ¯à¤O·U¨Ó·U±j¡A©Ò¯àÀx¦sªº ½Ķª¾ÃѶV¨Ó¶V¦h¡C¦]¦¹¹ï¤j¶q½Ķ¦Ó¨¥¡A¾÷¾¹Â½Ä¶¥i¥H¤j¶q´î ¤Ö«ÂЩʪº¾÷±ñ°Ê§@¡A¨Ï½Ķ¤Hû±Mª`©ó«~½è¤è±ªº´£ª@¡C¬G´N ¬Û¦P½Ķ«~½èªº¨â¥÷¤å¥ó¦Ó¨¥¡A¨ä¦¨¥»¥²µM¤ñ¤H¤u½Ķ¨Ó±o§C¡C ¦Ó¥B¡A¾÷¾¹Â½Ä¶¨t²Î¥i¥Hª½±µ»P¹q¸£±Æª©¨t²Î³s¦¨¤@Ó§¹¾ãªº¡u ½Ķ¡Ð¥Xª©¨t²Î¡v¡A¨Ã»P¨ä¥¦¿ì¤½«Ç¦Û°Ê¤Æ³]¬I¤Î¤å®Ñ³B²z¨t²Î ª½±µ³sµ²¡A¦b¤H¤u¥´¦r¡B±Æª©¡B®Õ½Zµ¥»ù®æ¤éÁÍ°ª©ùªº¤µ¤Ñ¡A¨ä ¸gÀٮįq§ó¥[©úÅã¡C ¡´ ±M·~³N»y«e«á¤@P ¦b¾÷¾¹Â½Ä¶¨t²Î¤¤¡A¥Ñ©ó¨t²Î¥»¨¤º«Øªº±M·~µü¨å¦³¤@©wªºÂ½Ä¶ ¤è¦¡¡A¦]¦¹¡A¥i¥H¨Ì¾Ú«È¤áªºn¨D¨Ó½Ķ¡A¤£·|¦³«e«á¤£¤@Pªº ±¡ªp¡C¤Ï¤§¡A¦b¤H¤u½Ķ¹Lµ{¤¤¡A§Y¨Ï¬O¦P¤@Ó½Ķ¤Hû¡A¹ï¦P ¤@³N»y¡A¦b¦P¤@¤Ñ¤§¤º¡A¤]¥i¯à¥X²{«e«á¤£¤@Pªº±¡ªp¡CY¤@¥÷ ¤å¥ó¥Ñ³\¦h¤H¦XĶ¡A§ó®e©ö²£¥Í°ÝÃD¡C±¡ªpÄY«®É¡A¥i¯à¦]¥Î»y ¤£¤@P¦Ó¾ÉPĶ½Z¤£³ô¨Ï¥Î¡A¥²»Ý¥þ±§ï¼g¡A³y¦¨®É¶¡»Pª÷¿ú¤W ªºÂù«·l¥¢¡C¦]¦¹¡A¹ï©ó±M·~³N»yn¨D¸ûÄY®æªº¤å¥ó¡A¦p§Þ³N©R ¥O¡B¨Ï¥Î¤â¥Uµ¥¡A¾÷¾¹Â½Ä¶¨t²Î¥i¥H´£¨Ñ¸û¦nªº¸Ñµª¡C ¸Õ·Q¡G¤@¬[¸¾÷ªººû×°V½m¤â¥U¥i¯à¼Æ¥H¾·p¡A¦pªG¥]µ¹¤£¦P±M ·~I´º¤Îª¾ÃѤô¥ªºÂ½Ä¶¤Hû¡A¥H¶Ç²Îªº¤H¤u¤è¦¡¨Ó½Ķ¡A¦p¦ó ¯à½T«O¨C¤@Ó¹s¥ó¦WºÙ¤Î¨C¤@Ó¯S©w§Þ³N©R¥Oªº¤@P©Ê¡C¨ä¼vÅT ¦ÛµM¤£¨¥¦Ó³ë¡C¦P®É¡A·í«È¤á»Ýn§ó°Ê¬Y¤@¥Î»yªºÂ½Ä¶¦WºÙ®É¡A ¾÷Ķ½Z¥u»Ý§Q¥Î²³æªº½s¿è«ü¥O§Y¥i¥þ¼Æ¥N´«¦¨·sªºÂ½ªk¡F¦Ó¤H ¤u½Ķ¦]¬°Â½Ä¶¤£¤@P¡A«h¥i¯à¥²»Ý§Q¥Î¤H¤u¥þ¼ÆÀ˵ø¹L¤~¯à½T ©w¡A©Ò¯Óªº¤H¤O°]¤O¥i¯à¼Æ¿©ó¾÷¾¹Â½Ä¶ªº¦¨¥»¡C³oºØ¹ï©ó¤@P ©Êªº»Ý¨D¡A¨ä¹ê¬O«Ü¦hª`«¥Xª©«~½èªº¤½¥q¡A±N§Þ³N¤â¥UÂà¥Ñ¾÷ ¾¹Â½Ä¶ªº¤@¤jì¦]¡A¤]¬O¤j®a¹ï¾÷¾¹Â½Ä¶¦@¦P»{©wªºÀuÂI¡C ¡´ ¸ê®Æ«O±K©Ê¨Î ¤£½×¬O¬F©²¾÷Ãö¡Bx¨Æ³æ¦ì©ÎªÌ¤½¥ÁÀç³æ¦ì¡A¤@¯ëÁ`§Æ±æ¸ê®Æ¯à °÷«O±K¡A·U¤Ö¤Hª¾¹D·U¦n¡C¥Ñ©ó¤H¤u½Ķ¸g¤âªº¤H¸û¦h¡A¬G¸ê®Æ ¨ì³B¶Ç»¼©Ò³y¦¨ªº¬ª±K·ÀI¤]¤j¬°´£°ª¡C§Q¥Î¾÷¾¹Â½Ä¶¨t²Î¡A«h ¥i¥H°§C¾÷±K¥~¬ªªº¥i¯à©Ê¡A½T«O¸ê®Æ¦w¥þ¡C ¡´ ¤@³eªº§@·~¬yµ{ ¾÷¾¹Â½Ä¶°£´£¨Ñ½ĶªA°È¤§¥~¡A¶¡±µ²£¥Íªº¤¤¤å¤Î^¤åÀɮסA¤] ¥i¥Hª½±µ§Q¥Î¨Ó§@¹q¸£±Æª©¡A¬Ù¥h¤F¤H¤u¥´¦r¡B±Æª©µ¥¤âÄò¡A¤] ¸`¬Ù¤F®Õ¹ï¡Bק諸®É¶¡¡C¦¹¥~¡A©ÒÀò±oªº¤¤¡B^¤åÀÉ®×µ¥°ÆÄÝ ²£«~¡A¹ï©óÀɮ׫إߡB¦Aª©³B²z©Î«·s½Ķ¦¨¨ä¥¦ªº»y¨¥¡A³£·¥ ¬°¤è«K¡C³oºØ¥H¤@³e§@·~ªº¤è¦¡¡A§¹¦¨Â½Ä¶¤Î¥Xª©ªº°Ê§@¡A´£¨Ñ ¤F«È¤á§¹¾ãªº¸Ñ¨M¤è®×¡A¹ï©ó»Ýn½Ķ¥Xª©¤j¶qĶ½Zªº«È¤á¦Ó¨¥ ¡A¥i¸`¬Ù³\¦h¤£¥²nªº®É¶¡¤Îª÷¿ú¤Wªº®ö¶O¡C ¾÷¾¹Â½Ä¶ªº¨î ·íµM¡A¾÷¾¹Â½Ä¶¤]¦³¤@¨Ç¥ý¤Ñ¤Wªº¨î¡A¥u¦³¥R¥÷¤F¸Ñ¨ä¨î¡A ¤~¯à¹ï¾÷¾¹Â½Ä¶§@³Ì¾A·íªºÀ³¥Î¡G ¡´ ½Ķªì½Z«~½è¥¼¿²§¹¬ü ¾÷¾¹Ä¶¤åªº«~½èµø¨t²ÎªºÂ½Ä¶ªk«h¡B±M¥Î³N»yªº¦h¹è¡B¤å½Zªº¤å ªkµ²ºc¦Ó©w¡C¦]¦¹¡A¾÷¾¹Â½Ä¶©Ò²£¥Íªºªì½Z¥¼¥²§¹¥þ²Å¦X¥Ø¼Ð»y ¨¥ªº¥Îªk¡CµM¦Ó¡A¾÷¾¹µLªk³B²z±o«Ü¦nªº²Ó¸`¡A¥i¥HÂǵÛ×µ¤H ûªº×¹¢¡A±o¨ì»P¤H¤u½Ķ¬Û¦Pªº«~½è¡C¦]¦¹¡A±µ¨ü½ĶªA°Èªº «È¤á¨Ã¤£·|·P¨ü¨ì³oºØ¨î¡C ¡´ ½Ķ½d³ò¨ü©ó±M·~µü·J »P¤H¤u½ĶÃþ¦ü¡A¾÷¾¹Â½Ä¶¨t²Î¹ï¯S©w±M·~»â°ìªºÂ½Ä¶¯à¤O¡A¨ü ©ó¨t²Î¤¤±M·~µü¨å³N»yªº¦h¹è¡A¥H¤Î×µ¤Hûªº±M·~ª¾ÃÑ¡CY µL¨Ï¥ÎªÌ©Ò»Ýªº±M·~µü·J¡A´N¥²»Ý³z¹L¦r¨åºÞ²z¨t²Î«Ø¥ß±M·~µü ¨å¡C ¡´ ¸û¾A¦X¦³©T©wÅéµôªº¤å½Z ¥Ñ©ó¾÷¾¹Â½Ä¶¬O¨Ì¾Ú¤@©wªºªk«h¨Ó¶i¦æ½Ķ¤u§@¡A¦]¦¹¤ñ¸û¾A¦X ½Ķ´yz¨ãÅé¨Æª«¥B¤åÅé¸û¬°Â²³æ©T©wªºªº¤å½Z¡A¦p¬ì§Þ¤å³¹¡B §Þ³N¤â¥U¥H¤Î¨Ï¥Î«ü«nµ¥µ¥¡C¤@¯ë¤p»¡¡B´²¤å©M¸Öºqµ¥¤å¾Ç§@«~ ¡A¸g±`¨Ï¥Î«Z»y¡B¤Þ³ë¡B·t³ë¤Î¯S®í¤åªkµ²ºc¡A¦]¦¹¡A½Ķ®É¥² »Ý¥[¤J±¡·P¡A¦Ó¥B½ĶªÌn¦³¤@©wªº¤å¾Ç³y¸Ú¡A¹ï©ó¤å¾Ç¨å¬G¦³ ©Ò¤F¸Ñ¡C³o¨Ç±ø¥ó§Y¨Ï¬O¤@¯ë½Ķ¤Hû¤]Ãø¥H¦P®ÉÝÑ¡A§ó¦óªp ¬O¨Ì·Ó¤@©w³W«h°õ¦æªº¾÷¾¹¡I¦]¦¹³oÃþ¤å³¹¨Ã¤£¾A¦X¥Î¾÷¾¹Â½Ä¶ ¨t²Î³B²z¡C¦¹¥~¡A¥Ñ©ó¤@¯ë¾÷¾¹Â½Ä¶¨t²Î³£°²³]¿é¤J¥y¤l¬O¦Xªk ªº¥y¤l (well-formed sentences)¡A¦]¦¹¡A©¹©¹¥u°w¹ï¦Xªk¥y¤l §@³B²z¡F¹ï©ó¤£²Å¦X¨t²Î»yªkªº¥y¤l (ill-formed sentences) ªº³B²z¯à¤O¸û®t¡C©Ò¥H¡A¦³¨Ç¤å¥ó¡A¦perror messages, system prompts,¤Îmenu items µ¥¡A°£«D¨t²Î¦³¯S®íªº¹ïµ¦¨Ó³B²z¡A°ò ¥»¤W¬O¤£¾A¦Xª½±µ¥Î¤@¯ëªº¾÷¾¹Â½Ä¶¨t²Î¥h³B²zªº¡C °ò©ó«ez¦U¶µ¾÷¾¹Â½Ä¶ªº¯S©Ê©MÀu¯ÊÂI¡A¤@¯ë¦Ó¨¥¡A¦b¤U¦C±¡ªp ¤U¡A¸û¾A¦X§Q¥Î¾÷¾¹Â½Ä¶¨t²Î©Ò´£¨ÑªºÂ½Ä¶ªA°È¡G ¡D¤å½Z¼Æ¶qÃe¤j¡A¥²»Ý¾a¤j¶qªº¤H¤â¤~¯à§¹¦¨¡C ¡D¤å½Z®É®Ä©Ê±j¡A¥²»Ý¦bµu®É¶¡¤º§¹¦¨¡A©Î»Ý¥¿½T¹w¦ô½Ķ®Éµ{¡C ¡D¤å½Z¥Î»y¤§¤@P©Ên¨DÄY®æ¡C ¡D¤å½Z¾÷±K©Ê°ª¡A¸g¤â¤Hû¶V¤Ö¶V¦n¡C ¾÷¾¹Â½Ä¶ªº¹B§@¤è¦¡ ¥H¤µ¤éªº¾÷¾¹Â½Ä¶§Þ³N¦Ó¨¥¡A¥¿½T¨Ï¥Î¾÷¾¹Â½Ä¶ªº½T¥i¥H¹F¨ì«e ±©Ò»¡ªº¹w´Á¸gÀٮįq¡CµM¦Ó¡A¥Ñ©óFAHQMTªº¥Ø¼ÐÃø¥H¹F¦¨¡A¦] ¦¹¡A¨Ï¥ÎªÌ¦p¦ó¥¿½T¬Ý«Ý¾÷¾¹Â½Ä¶¬O«D±`«nªº¡C¤£·íªº»{ª¾©¹ ©¹©è®ø¾÷¾¹Â½Ä¶¹ï¤@Ó½Ķ±M®×©Ò±a¨Óªº¦n³B¡C¦]¦¹¡A¨Ï¥ÎªÌº ¥ý¥²»Ýµû¦ô¦Û¤v©Î¥ø·~¥»¨ªºÂ½Ä¶³W¼Ò¡B¤å½Z©Ê½è¡B¦³µL±Mªùªº ½Ķ³¡ªùµ¥¦]¯À¡A¨Ó¨M©w¦p¦ó¦³®Ä¦aµo´§¾÷¾¹Â½Ä¶ªº¸gÀٮįq¡C ¤@¯ë¦Ó¨¥¡A½Ķªº¹Lµ{¤¤¡A¤H¤Oªº¤¶¤J¶V¤Ö¡A¸g¤â¤Hû¶¡©Ò»Ýªº ·¾³q®É¶¡¶V¤Ö¡A¹ï¾ãÅé½Ķ¦¨¥»ªº°§C¤Î®Ä¯qªº´£ª@¶VÅãµÛ¡C¦] ¦¹¡A¾÷¾¹Â½Ä¶³Ì¦³®Äªº§Q¥Î¤è¦¡¡AÀ³¸Ó¬O©Ò¿×ªºin-house translation,§Y±N¾÷¾¹Â½Ä¶ªº¤u§@¡A¥æµ¹¥Ñ¤@¸s±M·~ªº¾÷¾¹Â½Ä¶ ¨t²ÎºûÅ@¤Hû¡B±M·~ª¾ÃѿԸߤHû¤Î±M·~×µ¤Hû©Ò²Õ¦¨ªºÂ½Ä¶ ¤¤¤ß¨Ó°õ¦æ¡C¥Ø«e¥@¬É¤W¸û¦¨¥\ªº°Ó·~¹BÂà¡A¦pµÛ¦WªºSYSTRAN ¨t²Î, ¤j©è¤W³£¬O¥H³oºØ¤è¦¡§@·~¡C°£¤F¨t²ÎºûÅ@¥~¡A¤@Ó§¹µ½ ªºÂ½Ä¶¤¤¤ß¥i¥H´£¨Ñ¨Ï¥ÎªÌ©Ò»Ýªº±M·~ª¾ÃÑ¡A¦Ó³oºØ½Ķ¤¤¤ß¬° ¤£¦P«È¤á½Ķ©Ò«Øºc©Î²Ö¿nªºª¾ÃÑ®w¡A¤]¥i¥H¬°¦U¨Ï¥ÎªÌ¦@¥Î¡A ¸`¬Ù¨Ï¥ÎªÌ¦Û¦æ«Øºc©Î±ÄÁʪº¦¨¥»¡C¦pªG¨Ï¥ÎªÌ¦³¯S®í«O±K»Ý¨D ¡A©Î¦Û¤v¦³Ãe¤jªºÂ½Ä¶³¡ªù¡A«h¥i¦Ò¼{¯²¥Î¾÷¾¹Â½Ä¶¨t²Î¡A¨Ã¥Ñ ¾÷¾¹Â½Ä¶¨t²Îªº¶}µoªÌ´£¨Ñ°V½m¤Î§Þ³N¤ä´©¡A«Ø¥ß¦Û¤vªºÂ½Ä¶¤¤ ¤ß¡C ¤@¯ë¨Ï¥ÎªÌ³q±`¶È¯à§@¤@¨Ç²³æªº¦r¨å§ó·s°Ê§@¡A¹ï©óºûÅ@¨t²Î Ãe¤jªºª¾ÃÑ®w«hµL¯à¬°¤O¡C¦]¦¹¡A¤@¯ë¥HPC¬Æ¦Ü©ó¤u§@¯¸©Ò«Øºc ªº©Ò¿×Ó¤H½Ķ¤u§@¯¸ (personal MT), ¨ì¥Ø«e¬°¤î¡AÁÙ¬O¤£¤Á ¹ê»Úªº¸Ñ¨M¤è®×¡C©Ò¥H¡A¾¨ºÞ°ê¤º¥~¤w¦³¤@¤j°ï°t°â¦bPC©Î¤u§@ ¯¸ªº©Ò¿×¾÷¾¹Â½Ä¶¨t²Î¡A¦Ó¥B°Ê»³«ÅºÙ¦³¼Æ¤Q¸Uµü·J¡A¥i¥H¹ïĶ ¼ÆºØ»y¨¥¡AµM¦Ó¡A¯u¥¿¹ï¾÷¾¹Â½Ä¶¦³ÁA¸Ñªº¤H³£ª¾¹D¡A³o¨Ç¨t²Î ¦h¼Æ¥u¬O¹L¥÷§j¼Nªºtoy systems¦Ó¤w¡C¸Õ·Q¡A·í¤@¨Ç¸û²³æªº «÷¦rÀˬd¾¹ (spelling checkers)¡B«÷¦r®Õ¥¿¾¹(spelling correctors)¤Î¤åªkÀˬd¾¹ (grammar checkers)³£ÁÙ¤£¬O§@±o«Ü ¦¨¼ôªº®ÉÔ¡A¦p¦ó¬Û«H§ó¥[½ÆÂøªº¾÷¾¹Â½Ä¶¡A¯à¥H³oºØ²³æªº«¬ ºA¡Aº¡¨¬¤@¯ë¥ø·~©Ò»ÝªºÂ½Ä¶»Ý¨D¡H ¥H¥Ø«e¥X²£³Ì¦h¾÷¾¹Â½Ä¶¨t²Îªº¤é¥»¨Ó»¡¡A¥H³oºØ«¬¦¡½æµ¹¨Ï¥Î ªÌªº¾÷¾¹Â½Ä¶¨t²Î¡A¦h¼Æ¦b¥Î¹L¤@¨â¦¸¤§«á¡A§Y³Q§ô½Ñ°ª»Õ¡A¹ï ©ó¨Ï¥ÎªÌ¯u¥¿ªºÂ½Ä¶»Ý¨D¡A¨ÃµLªk´£¨Ñ¾A·íªº¸Ñ¨M¤è®×[JEIDA 89]¡C¨ä¥Dnì¦]¡A¦b©ó¤@¯ë³c½æ½Ķ³nÅ骺³nÅé¼t°Ó¡A©¹©¹¤£ ·|¨è·N´£¿ô¨Ï¥ÎªÌ¦³Ãö¾÷¾¹Â½Ä¶ªº¥ý¤Ñ©Ê¨î¡A¦³·NµL·N¶¡»~¾É ¤F¨Ï¥ÎªÌ¹ï¾÷¾¹Â½Ä¶ªº»{ª¾¡F¥t¤@¤è±¡A¤]¦]¬°¥»¨¨Ã¨S¦³¹ê»Ú ¹B§@¾÷¾¹Â½Ä¶¨t²Î¤Î½Ķªº¸gÅç¡A¨Æ¹ê¤W¨Ã¨S¦³¯à¤O§i¶D¨Ï¥ÎªÌ ¦p¦ó¤~¯à¹F¦¨¹w´Áªº¸gÀٮįq¡Cµ²ªG¡A¨Ï¥ÎªÌ¨S¦³¯à¤OºûÅ@¾÷¾¹ ½Ķ©Ò»ÝªºÃe¤jª¾ÃÑ®w¤Î¾÷¾¹Â½Ä¶©Ò±M¥Îªºµü¨å¡A¥u¦n±N½Ķ³n Åé§ô½Ñ°ª»Õ¡C¦]¦¹¡A¨Ï¥ÎªÌ±¹ï°Ó·~¤Æªº«Å¶Ç®É¡A¦p¦ó§@¥X¥¿½T §PÂ_¬O¤@Ó«nªº½ÒÃD¡F¦Ót³d¥ôªº¾÷¾¹Â½Ä¶±À¼sªÌ¤£¦ýn¯à´x ´¤¾÷¾¹Â½Ä¶ªºÀu¯ÊÂI¡A¤]À³µ¹¨Ï¥ÎªÌ¾A·íªº«Øij»P±Ð¨|¡A¤~¯à¨ó §U¨Ï¥ÎªÌ§¹¦¨¤@ÓÓ¦¨¥\ªºÂ½Ä¶±M®×¡C ¾÷¾¹Â½Ä¶°ò¥»ì²z ¥Ø«eªº¾÷¾¹Â½Ä¶¨t²Î¦³¤TºØ°ò¥»ªºÂ½Ä¶µ¦²¤¡A¤À§O¬°ª½±µ¦¡ (direct)¡BÂà´«¦¡(transfer)¤Î¤¤¤¶¦¡ (interlingual)½Ķ¡C ª½±µ¦¡Â½Ä¶¥Dn¬O°w¹ï¤@²Õ¯S©wªºì©l»y¨¥¤Î¥Ø¼Ð»y¨¥¨Ó¼¶¼g½ Ķ³W«h¡F¤¤¤¶¦¡Â½Ä¶ªk«h¥ø¹Ï±Nì©l»y¨¥Âà´«¦¨¤@ºØ¦U»y¨¥§¡³q ¥Îªº¤¤¶¡ªí¥Ü¦¡ (intermediate form)¡A¦A¥Ñ¸Óªí¥Ü¦¡²£¥Í¥Ø¼Ð »y¡CÂà´«¦¡Â½Ä¶«h¤¶©ó¨âªÌ¤§¶¡¡A¥ý±Nì©l»y¨¥¤ÀªR¦¨¤@ºØì©l »yªº¤¤¶¡ªí¥Ü¦¡¡A¦A³z¹L¤@ÓÂà´«ªº¹Lµ{±N¸Óªí¥Ü¦¡Âà´«¦¨¥Ø¼Ð »yªº¤¤¶¡ªí¥Ü¦¡¡A¦Ó«á§Q¥Î¥Ø¼Ð»yªº¤¤¶¡ªí¥Ü¦¡¡A¦X¦¨©Î²£¥Í¥Ø ¼Ð»y©Ò¯S¦³ªº»yªkµ²ºc¤Î¥y¤l¡C ¤¤¤¶¦¡ªº¾÷¾¹Â½Ä¶¨t²Î¡A²z½×¤W¥un¬°¨C¤@ºØ»y¨¥¼¶¼g¤@Ó¤ÀªR ¼Ò²Õ¤Î¤@Ó¦X¦¨¼Ò²Õ¡A§Y¥i»P¨t²Î¤¤ªº¨ä¥¦»y¨¥¤¬¬Û¹ïĶ¡A¬G¾A ¦X¥Î¨Ó§@¦h°ê»y¨¥¶¡ªº¹ïĶ¡CµM¦Ó¡An§ä¨ì¤@Ó¾A¥Î©ó©Ò¦³»y¨¥ ªº¤¤¶¡ªí¥Ü¦¡¨Ã¤£®e©ö¡A¦]¦¹¡A¥u¾A¥Î©ó¦P¤@»y¨t©Òl¥Í¥X¤§¦U »y¨¥¶¡ªº¹ïĶ¡C¹ê»Ú¹B§@¤W¡A«h©|¥¼¬Ý¨ì¤ñ¸û¦¨¥\ªº¨Ò¤l¡Cª½±µ ¦¡Â½Ä¶¨t²Î¥i°w¹ï¤@²Õ»y¨¥¡A§¹¦¨«Ü²Ó·LªºÂ½Ä¶°Ê§@¡A¦ýn¥[¤J ·sªº»y¨¥®É¡A¬Û·í§xÃø¡C©Ò¥H¡A¥Ø«e¥HÂà´«¦¡ªº¾÷¾¹Â½Ä¶¨t²Î³Ì ¬°´¶¹M¡C ¥Ñ¤Wz»¡©ú¥iª¾¡AÂà´«¦¡ªºÂ½Ä¶¨t²Î¡A¤jP¥i¥H¤À¦¨¤ÀªR (analysis)¡BÂà´« (transfer)¡B¤Î¦X¦¨ (synthesis)©Î¥Í¦¨ (generation)µ¥¤TÓ¨BÆJ¡C¥H¤U¥H¤@Ó²³æªº^¤å¥y¤l¡A¨Ó»¡©ú ½Ķ¦¨¤¤¤å®É¨CÓ¨BÆJªº³B²z¹Lµ{¡G ^¤åì¤å¡GMiss Smith put two books on this dining table. ¤¤¤åĶ¤å¡G¥v±K´µ ¤p©j §â ¨â ¥» ®Ñ ©ñ ¦b ³o ±i À\®à ¤W±¡C <<¹Ï¤G¡G¨Ò¥y¤§¥yªkµ²ºc>> ¡i ¤ÀªR ¡j 1.ºcµü¤Îµü·J¤ÀªR¡G±q¨t²Îµü¨å¤¤¬d¥X^¤å¥y¤l¸Ì¨C¤@Óµü¶µªº ¸ê®Æ¡A¨Ã¨Ì^¤åºcµüì«h¡A°õ¦æºcµü¤ÀªR¡A§ä¥Xµüªºµü®Ú¡Bµüº ¤Îµü§Àµ¥µü¯À¡F¨Ì¾ÚµüÃþªº¦@¦s²{¶H¡A¨M©w¸ÓµüªºµüÃþ¡Cµü¨å¤¤ °O¸üµÛ¨C¤@ÓµüªºµüÃþ¡B¤HºÙ¡B³æ½Æ¼Æ¤Î¤¤¤å½Ķµ¥°T®§¡A¥i§@ ¬°¤ÀªRªº°ò¥»¸ê®Æ¡C¨Ò¦p¡A¨Ò¥y¤¤ªº¡uMiss¡v¬°¦Wµü¡B³æ¼Æ¡B¤¤ ¤å½Ķ¬°¡u¤p©j¡v¡C 2.¥yªk¤ÀªR¡G®Ú¾Ú^¤å¥yªk¡A¶i¦æ¥y¤lµ²ºcªº¤ÀªR¡C¨Ò¦p¥»¨Ò¥y ¥i¤ÀªR¥X¦Wµü¤ù»y(Noun Phrases, NP: Miss Smith¡Btwo books ¡Bthis dining table)¡B°Êµü (Verb, V: put)¤Î¤¶¨tµü¤ù»y (Prepositional Phrase, PP: on this dining table)µ¥¡C¸Ô²Ó µ²ºc¦p¹Ï¤G©Ò¥Ü¡G 3.»y·N¤ÀªR¡G¨M©w¦Uµü©Îµü²Õªº»y·N¨¤¦â¡C¨Ò¦p¡AMiss Smith¦b ³o¸Ì¬Oput³oÓz»y (predicate)ªº¥D¨ÆªÌ¡Athis dining table «h¬O¨Æ¥óµo¥Íªº¦a¤è¡C ¡i Âà´« ¡j 1.µü·JÂà´«¡G±N^¤å¥y¤l¸Ì¨C¤@µü¶µÂà´«¬°¬Û¹ïªº¤¤¤å¡A¦pMiss Âà´«¬°¡u¤p©j¡v¡BtableÂà´«¬°¡u®à¤l¡v (¦Ó¤£¬O¡uªí®æ¡v)¡C 2. µ²ºcÂà´«¡G±N^¤åªºµ²ºc¤Îµü§ÇÂà´«¦¨¤¤¤åªºµ²ºc¤Îµü§Ç¡C¥H¥»¨Ò¥y¦Ó¨¥¡A ^¤åµü§Ç¬°¡G¡u¤p©j ¥v±K´µ ©ñ ¨â ®Ñ ¦b...¤W± ³o À\®à¡v¡A Âà´««áªº¤¤¤åµü§Ç¬°¡G¡u¥v±K´µ ¤p©j ¨â ®Ñ ©ñ ¦b ³o À\®à ¤W±¡v¡C ¡i ¦X¦¨ ¡j «Y°w¹ï¤¤¤å¯S¦³ªº»yªkµ²ºc§@¥²nªº³B²z¡A¥H²£¥Í¾A·íªº¤¤¤å¥y ¤l¡C¨Ò¦p¡A¦b¥y¤¤¥[¤W¡u§â¡v¦r¤Î¶qµü¡u¥»¡v¡B¡u±i¡vµ¥¡C¥H¥» ¨Ò¥y¦Ó¨¥¡A¸g¹L¦X¦¨ªº¨BÆJ«á¡A¤¤¤åÅܦ¨¤F¡u¥v±K´µ¤p©j§â¨â¥» ®Ñ©ñ¦b³o±iÀ\®à¤W±¡v¡C ¤W±¦UÓ¨BÆJ¶¡ªº¬É½u¥¼¥²³£«Ü²M·¡¡A¦U¨t²Î¦U¦³¨ä¤£¦Pªº§@ªk ¡C¨Ò¦p¡Aµ²ºcÂà´«¤]¥iµø¬°µ²ºc¥Í¦¨¡C ²{¦³Â½Ä¶¨t²Îªº§Þ³N²~ÀV¤Îµo®iÁÍ¶Õ ±q°Ó·~Æ[ÂI¨Ó¬Ý¡A¥Ø«eªº¾÷¾¹Â½Ä¶½T¤w¥i¥H¹F¦¨¤@¨Ç¸gÀٮįq¡A ¨ó§U»Ýn¤j¶q½Ķªº¾÷ºc¡A¨³³tÀò±o¥~¤å¸ê®Æ¡C¦ý±q¬ã¨sªÌªº¨¤ «×¨Ó¬Ý¡A³\¦h¾÷¾¹Â½Ä¶ªº¬ã¨sijÃD¤´¥¼¦³¾A·íªº¸Ñ¨M¤§¹D¡C¥[¥H ¤j³¡¥÷ªº¾÷¾¹Â½Ä¶¬ã¨s¤Hû¡A¨Ã¨S¦³¹ê»Ú¹BÂà¾÷¾¹Â½Ä¶¨t²Îªº¸g Åç¡A¦]¦¹¡A©Ò¬ã¨sªºÄ³ÃD©Î¤è¦VÃø§K°¾¦V¾Ç°|¦¡ªº§@ªk¡C¦Ó¶Ç²Î ªº¾÷¾¹Â½Ä¶µo®iµ¦²¤¡A¤j©è¤W¬O¶Ç²Î»y¨¥¾Ç¡B§Î¦¡»y¨¥¾Ç¤Î¤Ö¼Æ ¤H¤u´¼¼z§Þ³Nªººî¦XÅé¡C§Q¥Î³o¨Ç§Þ³N©Òµo®iªº¾÷¾¹Â½Ä¶¡A¦b¾ú ¸g¼Æ¤Q¦~ªº¬ã¨s«á¡AÁÙ¬O¥¼¯à¤j¶q¨Ï¥Î¡AÅã¥Ü¥H©¹ªº§@ªk¦³¨ä¥ý ¤Ñ¤Wªº¨î¡C¥H§Ú̦h¦~µo®i¤Î¹BÂà¾÷¾¹Â½Ä¶¨t²Îªº¸gÅç¨Ó¬Ý¡A ¶Ç²Îªº¾÷¾¹Â½Ä¶¬[ºc¡Bµo®i¼Ò¦¡¤Î¤èªk¡A¥Dnªº¯Ê³´¥i¥HÂkµ²¬° ´XÂI¡G 1. ¯Ê¥F¤@P©Êªºª[¸q¸Ñ¨M¤è®× ¥Ñ«e±ªº¬[ºc¤Îì²z¡A¤j®a¥i¥Hµo²{¾÷¾¹Â½Ä¶¨t²Î©Òn³B²zªº²Ó ¸`¬Û·íÁcº¾¡C´N§Þ³N¼h±¨Ó¬Ý¡A¾÷¾¹Â½Ä¶³Ì¤jªº°ÝÃD¡A§Y¦b©ó¨C ¤@Ó³B²z¨BÆJ³£¥Rº¡¤F¤£½T©w©Ê (uncertainty)¤Îª[¸q©Ê (ambiguity)¡C¨Ò¦p¡A"bank"¥i¯à·í°Êµü©Î¦Wµü¨Ï¥Î¡A³o¬OµüÃþ ªºª[¸q¡F·í¥¦°µ¬°¦Wµü®É¡A¥i¯à¬O"»È¦æ"¡A¤]¥i¯à¬O"ªe©¤"¡A³o ¬O¦r¸q (word sense)¤Wªºª[¸q¡F"I saw a girl with a telescope."¥i¯à¬O"¬Ý¨ì¤@Ó±aµÛ±æ»·Ã誺¤k«Ä"©ÎªÌ"¥Î±æ»·Ãè ¬Ý¨ì¤@Ó¤k«Ä"¡A³o¬O×¹¢»y×¹¢¤£¦Pµü²Õ©Ò³y¦¨ªº»yªkª[¸q¡F "The police were ordered to stop drinking by midnight." ¥y¤¤ªº"police"¥i¯à¬Onªý¤î§O¤H³Ü°s (©Î¨ä¥¦¶¼®Æ) ªº¤H¡A¤] ¥i¯à¬O³Qªý¤îÄ~Äò³Ü°sªº¹ï¶H¡A³o¬O»y·N¨¤¦â¤£¦P©Ò³y¦¨ªºª[¸q ¡F"She was elected the president of the United States by the public."¥i¥H½¦¨³Q°Ê¦¡ªº"¦o³Q¤j²³±À¿ï¬°¬ü°êÁ`²Î"©Î¥D °Ê¦¡ªº"¤j²³±À¿ï¦o¬°¬ü°êÁ`²Î"¡A³o¬OÂà´«¤Î¥Í¦¨¹Lµ{²£¥Í¤£¦P ¥yªkµ²ºc©Ò³y¦¨ªºª[¸q¡C ³o¨Çª[¸q²{¶H²[»\ªº¼h¦¸¡A¥]¬A¤ÀªR¡]µü·J¤ÀªR¡B¥yªk¤ÀªR¡B»y ·N¤ÀªR¡^¡BÂà´«¡]µü·JÂà´«¡Bµ²ºcÂà´«¡^¡B¦X¦¨¡]µ²ºc¦X¦¨¡Bµü ¯À¦X¦¨¡^µ¥¦UÓ¶¥¼h¡C¨C¤@ºØª[¸q²{¶H³£»Ýn¨ä¥¦ªº°T®§¨ÓÀ°¦£ ¸Ñ¨M¡C¶Ç²Îªº¾÷¾¹Â½Ä¶¨t²Î¡A¥Dn¬O¨Ì¾a»y¨¥¾Ç®a©Î¹q¸£±M®a¨Ì ²{¦³ªº»y¨¥ª¾ÃÑ¡Aºc«Ø¤@±ø±øªº³W«h¡A¨Ó¨M©w¦b¦óºØ±¡ªp¤U¸Ó±Ä ¨ú¤°»ò¦æ°Ê¡C³o´N¬O©Ò¿×ªº³W«h¦¡¾÷¾¹Â½Ä¶ (rule-based MT)¡C ¹J¦³¬Y¤@Ó¼h¦¸¤£¯à¸Ñ¨Mªº°ÝÃD¡A´N§Q¥Î§ó¦h²Óº¾ªºª¾ÃÑ¡A§@§ó ²`¤@¼hªº¤ÀªR¡CµL¥i§_»{ªº¡A³oºØ§@ªk¥i¥Hª½±µ§Q¥Î¤@¨Ç²{¦³ªº »y¨¥¾Ç²z½×¤Î¤H¤u´¼¼zªº²z½×¡A¨Óºc«Ø°ò¥»ªº¾÷¾¹Â½Ä¶¨t²Î¡C¦ý ¬O¡A·í¨t²Î¥Ñ¹êÅç«ÇªºÂú§Î¨t²ÎÂà¤Æ¨ì¤j«¬ªº¹ê»Ú¨t²Î¡A±¹ï¦U ¦¡¦U¼Ëªº¤å³¹®É¡A°ÝÃD´N¨Ó¤F¡C º¥ý¹J¨ìªº°ÝÃD¬O¡A¤£ºÞ¥[¤J¦h¤Öª¾ÃÑ¡A¥un¬O¨S¦³¿ìªk§â©Ò»Ý ª¾ÃÑ¥[¥þ¡A¤£½T©w©Ê¤Îª[¸q²{¶H´N·|¦s¦b¡C¥Ñ©ó¨t²Îªº¾ãÅéÁZ®Ä ª½±µ»P¨t²Î»~§Pªº¾÷²v¦¨¥¿¤ñ¡A¦]¦¹¡A°§C¨t²Î»~§Pªº¾÷²v¡A´N ¦¨¤Fºnªº¤u§@¡C¦b¦¹±¡ªp¤U¡A¨t²Î¥²»Ý§Q¥Î¤@Óµû¤Àªº¼Ð·Ç¡A ¨Ó§@³Ì¤Ö·ÀIªº§PÂ_¡C¥Ø«e¦U¨t²Î¥Dn¬O¥H³W«h¦¡¨t²Î¬°¥D¡A¬G ÂA¤Ö¹ï¤£½T©w¤§±¡ªp¤©¥Hµû¤À¡F³\¦h¨t²Î¦b¹êÅç«Çµo®i®É¡A¤@¹J ¨ì°ÝÃD´N§Æ±æ³z¹L§ó²`¼hªº¤ÀªR¡A©Î¥[¤J¤@¨Ç¸gÅçªk«h¶i¥h¡A¨Ó ¸Ñ¨M²´«e©Ò¹J¨ìªº°ÝÃD¡A¨Ï±o³o¼Ëªº³W«h¦¡¨t²Î²V¶Ã²§±`¡C¨Æ¹ê ¤W¡An·Q§Q¥Î¤@¨Ç¤@¯ë©Êªº³W«h¡A¨Ó¸Ñ¨M½ÆÂøªº»y¨¥°ÝÃD¨Ã¤£®e ©ö¡C¦Ó¥B¡A¨C¥[²`¤@¼h¤ÀªR®É¡A¤S·|¤Þ¤J¥t¤@Ó¼h¦¸ªºª[¸q²{¶H ¡C¦]¦¹¡A¨ÃµLªk¯u¥¿¸Ñ¨Mª[¸q¤Î¤£½T©w©Êªº°ÝÃD¡C ÁöµM¦³¨Ç³W«h¦¡¨t²Î¤]±Ä¥Îµû¤Àªº¤è¦¡¨Ó³B²z¤£½T©wªº±¡ªp¡A¦ý ©Ò¥Îªºµû¤À¼Ð·Ç¦h¼Æ¬O¯Ê¥F¾Ç²z®Ú¾Úªº¸gÅçªk«h¡C³oºØ¸gÅ禡¤] ³\¦b¹êÅç«Ç¸Ì¯à°w¹ï¤Ö¼Æ¨Ò¤l¤©¥H½Õ¾ã¦Ü³Ì¨Îª¬ªp¡A¦ý¾ãÅé¦Ó¨¥ ¡A¨Ã¤£¯à±À¼s¨ì¨ä¥¦¤å³¹¤Î»â°ì¡A¹F¨ì³Ì¨Î¤Æªºµ²ªG¡A¤]µLªk¥Î ¤@ºØ¦³¨t²Î¡B¥i±±¨îªº¤è¦¡«ùÄò¦a§ï¶i¨t²ÎÁZ®Ä¡F·í¨t²Î×¥¿¬Y ¨Çªk«h©Î°Ñ¼Æ¨Ó³B²z¬Y¨Ç°ÝÃD®É¡Aªí±¤W¸Ñ¨M¤F³o¨Ç°ÝÃD¡A¹ê»Ú ¤W¹ï¨ä¥¦°ÝÃD¤Î¾ãÓ¨t²Îªº¼vÅT«oµLªkµû¦ô¡A¬Æ¦Ü©ó³y¦¨¤£¨}¼v ÅT¡C¦]¦¹¡A¨t²Î·|¤@ª½¦b¤ÏÂÐ×¥¿³o¨Ç¸gÅçªk«hªº¹Lµ{¤¤¥´ºu¡A «ÜÃøÂX¥R¨ì¤j«¬¨t²Î¡C n¬ð¯}³oºØ§Þ³N²~ÀV¡A°ß¦³µo®i¥X¤@®M«ÈÆ[ªºµû¤À¨ç¼Æ¡A§Q¥Î¦³ Ų§O¤Oªºµü·J¡B»yªk¤Î»y·N¯S¼x§@§PÂ_¡A¤~¯à¥Î³Ì¤Öªºª¾ÃÑ¡A§@ ³Ì¤£¨ã·ÀIªº§PÂ_¡A¦P®ÉÀH®É«O«ù¨t²Î¾ãÅ骺ÁZ®Ä¡C¥Ø«e BehaviorTran©Ò¨Ï¥Îªºµû¤À¤èªk¡A¬Oµ²¦Xµü·J¡B»yªk¤Î»y·N¼Ò¦¡ »P¨©¦¡¨Mµ¦ªk (Baysian Decision Rule)©Òµo®i¥X¨Óªº¤@®Mµû¤À ¨ç¼Æ[Su 88, 91, Liu 90, Chan 92a, Lin 92]¡C¥Ñ©ó¬O§Q¥Î³Ì ¨Î¤Æªº¨Mµ¦ªk«h¦b»y¨¥¼Ò¦¡¤§¤W¡A¦]¦¹¡A²z½×¤W¥i¥H¹F¨ì³Ì¤Ö¨M µ¦¦¨¥»ªº³Ì¨Î¤Æµ²ªG¡C 2. ª¾ÃÑ©â¨ú¤£©ö ¥t¤@Ó°ÝÃD¬O¡A¦p¦ó©â¨ú³o¨ÇÁcº¾ªºª¾ÃÑ (knowledge acquisition)¦Ó¹F¨ì¬Û·íªº²[»\±¡A¥B¦b¤£Â_§ó·sªº¹Lµ{¤¤¡A¤´ ¯àºû«ù³o¨Çª¾ÃѪº¤@P©Ê¡C¤@¯ëªº¾÷¾¹Â½Ä¶¨t²Î³Ì°_½Xn¦³¤@¡B ¤G¤Q¸UÓµü·J¡B¤@¨â¤d±ø¤ÀªR¤åªk¡A¤Î¤@¨â¤d²ÕÂà´«³W«h¤~¨¬¼Å ¨Ï¥Î¡C¦p¦ó¥Î¤H¤u§â³o¨Çµü¨å¸ê®Æ¤@¤@¶ñ»ô¡A¦Ó¨Ï±o³o¸Ì±ªº³W «h«O«ù¤@P¡A¤£·|¤¬¬Û½Ä¬ð¡A¬OÓ¬Û·í¯Ó®É¶O¤O¡B°ª¦¨¥»¡B°ªÃø «×ªº¤uµ{°ÝÃD¡C¤£n»¡³oºØ¤u§@»Ýn³z¹L¤@¸s¤H¨Ó§@¡A´Nºâ¤@Ó ¤H¥]¿ì¾ãÓ¤u§@¡A¤]«ÜÃø«O«ù¸ê®Æ¤Î³W«hªº¤@P©Ê¡C¨Æ¹ê¤W¡A§Y «K¬O±N¦P¤@Óµüµ¹¤©¾A·íªº»y·N¤ÀÃþ¡A¦¤W¸ò¤U¤Èªº·Qªk³£¥i¯à ¤£¤@¼Ë¡C§ó¦óªp¤@®M¾÷¾¹Â½Ä¶¨t²Î³q±`n¾ú¸g¼Æ¦~ªº®É¶¡¡A¥Ñ¤@ ¸s¤£¦Pª¾ÃÑI´ºªº¤H¸g¤â©O¡C³oºØ¤H¤u©â¨úª¾ÃѪº¤è¦¡¡A¹ï¤@Ó ¹êÅ禡ªº¨t²Î¤]³\¤w¸g¨¬°÷¡A¦ý¹ï©ó¹ê»Ú¹B§@ªº¤j«¬¨t²Î¦Ó¨¥¡A ¨ä²[»\±©M¤@P©Ê´NȱoÃhºÃ¡C¦ÓnºûÅ@¤@Ó³o¼Ëªº¨t²Î¤]·¥¬° §xÃø¡C ¥Ø«eªºÁͶլO§Q¥Î¤j¶qªº»y®Æ¡A³z¹L¤@¨Ç¦Û°Ê©Î¥b¦Û°Êªº²Îp¤è ¦¡¨Ó©â¨ú¥²nªºª¾ÃÑ¡C¨Ò¦p¡A§Q¥Î¬Û¾FµüÃþ¶¡¦@¦sªºÀW²v©Î¾÷²v ¡A¥i¥H«Ü·Ç½T¦a¥Ñ¼Æ¦Ê¥H¦Ü©ó¼Æ¤dºØ¥i¯àªºµüÃþ²Õ¦X¤¤¡A¨M©w¦U Ó¦rªº¥¿½TµüÃþ¡C¦Ó¥H©¹¨Ï¥Î³W«h¨Ó¨î³oºØµüÃþ¦@¥Í²{¶H®É¡A ¨ÃµLªk§@«Ü§¹µ½ªº³B²z¡C µM¦Ó¡A¯Âºé¨Ï¥Î²Îp¤è¦¡¦Ó¤£ÅU¤Î»y¨¥¥»¨ªº³W«ß¡A¤]®e©ö³y¦¨ °Ñ¼Æ¼Æ¥Ø¹L¤j¡A¦Ó¥²»Ý¥Î·¥¬°Ãe¤jªº»y®Æ¡A¤~¯à¾Ç²ß¨ì»y¨¥¾Ç¤W ¤w¼ôª¾ªº²{¶H¡C¨Ò¦p¡A³Ìªñ¦³¨Ç¾ÇªÌ¥ø¹Ï§Q¥Î·¥¬°Â²³æªº»y¨¥¼Ò ¦¡¨Óªí¹F½Ķªº¤º¦b¦æ¬°¡A§Q¥Î¼Æ¦Ê¸U¥yªºÂù»y»y®Æ(¥[®³¤j°ê ·|·|ij°O¿ý)¡A¨Ó¨D¨ú³oºØ½Ķ¼Ò¦¡ªº°Ñ¼Æ¡C¨äµ²ªGÁöµMÅã¥Ü³¡ ¥÷ªº¥i¦æ©Ê¡A¦ý¥Ñ©ó©Ò»Ýªº»y®Æ®w·¥¬°Ãe¤j¡A¬G¨ü¨ì¾Ç¬É¬Û·í¦h ªº½èºÃ»P§åµû¡C ¦]¦¹¡A¤ñ¸û¦nªº¤è¦¡À³¸Ó¬O±N¤@¨Ç¤wª¾ªº°ª¼h¦¸»y¨¥¾Ç²{¶H¥[¥H ¼Ò¦¡¤Æ¡A¦A§Q¥Î²Îpªº¤è¦¡¥Ñ»y®Æ¤¤¨D±o¾A·íªº¼Ò¦¡°Ñ¼Æ¡C¦p¦¹ ¡A¬J¥iµo´§»y¨¥¾Ç®aÂk¯Ç¤@¯ë©Ê»y¨¥³W«h¤Î«Ø¥ß»y¨¥¾Ç¼Ò¦¡ªºªø ¤~¡A¤S¥i±NÁcº¾ªºª¾ÃÑ©â¨ú¤u§@¥æµ¹¾Õªø³B²z¤j¶q¸ê®Æªº¹q¸£¨Ó ³B²z¡AÀò¨ú¤@P©Êªº¼Ò¦¡°Ñ¼Æ¡AÁקK¤H¤u©â¨úªº°ÝÃD¡A¦P®É´î»´ ¤H¤u¦¨¥»¡A¥i¿×¤@Á|¼Æ±o¡C³oºØ¥H»y®Æ¬°°ò¦¡A¬[ºc¦b°ª¼h»y¨¥ ª¾ÃѤWªº²Îp¤è¦¡¡A²ºÙCBSO (Corpus-Based Statistics-Oriented)¤è¦¡[Su 92a]¡A¦b³Ìªñ¤@¦¸¥l¶}ªº¾÷¾¹Â½ Ķ²z½×»P§Þ³N·|ij(TMI-92, 1992)¤¤¡A¤w³Qµû¬°ºî¦X¶Ç²Î»y¨¥¾Ç »P²Îp§Þ³Nªº¤@ºØ³Ì¥i¯àªº¥¼¨ÓÁͶաC 3. ³æ¦V¦¡ªº¨t²Î³]p ¾÷¾¹Â½Ä¶ªº¥t¤@Ó«n°ÝÃD¡A¦b©ó¶Ç²Îªº¾÷¾¹Â½Ä¶¨t²Î¬O¤@ºØ³æ ¦V¦¡ªº³]p¡C¥Ñ©ó¶Ç²ÎªºÂà´«¦¡¾÷¾¹Â½Ä¶¬O¿í´`µÛ¡y¤ÀªR¡z¡B¡y Âà´«¡z¤Î¡y¥Í¦¨¡z¤TÓ¨BÆJ¤@¸ô§@¤U¨Ó¡A¨Ï±o¡yÂà´«¡z¤Î¡y¥Í¦¨ ¡z±a¦³¿@«pªºì©l»y¨¥¦â±m¡C¦]¦¹¡A^¤¤Â½Ä¶¨t²Îªº¿é¥X³q±`¦¨ ¤F¡y^¤å¦¡ªº¤¤¤å¡z¡A¤é¤¤Â½Ä¶¨t²Î³q±`²£¥Íªº¬O¡y¤é¤å¦¡ªº¤¤ ¤å¡z¡A¦Ó«D¹D¦aªº¡y¤¤¤å¦¡¤¤¤å¡z¡C´«¨¥¤§¡A¥Î¨Ó²£¥Í¤¤¤åªº¥Í ¦¨¤åªk»P¤@¯ë¥Îªº¤¤¤å (¤ÀªR) ¤åªk¨Ã¤£¬Û¦P¡A¦Ó¬O¥Ñ^¤å©Î¤é ¤å¤åªk®M¹L¨Óªº¤åªk¡C³oºØ¿é¥X¡A¦ÛµM»Ýn¬Û·íªº×µ°Ê§@¡A¤~ ¯à¹F¨ì¨Ï¥ÎªÌªº´Á±æ¡C¬°¤FÁקK¨Ï¥Î³oºØ¨ã¦³ì©l»y¨¥¶É¦Vªº¥Ø ¼Ð»y¥Í¦¨¤åªk¡A°y©³©âÁ~ªº¤è¦¡¡A¤D¬O§Q¥Îì©l»y¨¥¤Î¥Ø¼Ð»y¨¥ ¨â®M¤åªk¡A¨Ó¾á¥ô¤ÀªR¤Î¥Í¦¨»yªk¡A¨Ã§Q¥Îì©l»y¨¥¤Î¥Ø¼Ð»y¨¥ ©Òºc¦¨ªºÂù»y»y®Æ®w(bilingual corpus)¦Û°Ê¾Ç²ßÂà´«°Ê§@¤Î¥Í ¦¨°Ê§@[Chan 92b]¡C 4. ©¿²¤»y¨¥¯S¼xªºÅ²§O¤O¤Î±j°·©Ê ¥Ñ©ó³\¦h¾÷¾¹Â½Ä¶¬ã¨s¤Hû¨ã¦³pºâ¾÷¬ì¾Çªº¾ÇÃÑI´º¡A¦Ó¤H¤u ´¼¼z¤S±j½Õª¾ÃѪí¹F¤è¦¡ (knowledge representation)ªº«n ¡A¦]¦¹¡A¤£¤Ö¾÷¾¹Â½Ä¶ªº¬ã¨s¤Hû³£¬Û«H¡A¨}¦nªºª¾ÃѪí¹F¤è¦¡ ¦³§U©ó¾÷¾¹Â½Ä¶«~½èªº´£ª@¡C¦P¼Ë¦a¡A³\¦h»y¨¥¾Ç®a¤]¤£Â_´M¨D ¸û¾A·íªº¤åªk«¬¦¡ (grammar formalism)¡A¥ø¹Ï¥Î¤ñ¸û¾A·íªº»y ªk²z½×¨Óªí¹F©Î¸ÑÄÀ¦P¤@ºØ»y¨¥²{¶H¡C³oºØ§@ªk¨Ã«D¨S¦³¦n³B¡C ¦ý¬O¡AÄY®æ¦a»¡¡Aª¾ÃѪí¹F¤è¦¡¥u¸ò¨t²Îªºªí¹F¯à¤O (expressive power)¦³Ãö¡A¦Ó»P³Ì«nªºª[¸q¸Ñ¨M¯à¤OµLÃö¡C¤@ Ӭݦü²³æªºªí¹F¤è¦¡¡A´N¾Ç²z¨Ó¬Ý¡A¥i¯à»P¤@Ó½ÆÂøªº»yªk²z ½×¨ã¦³¬Û¦Pªºªí¹F¯à¤O (¨Ò¦p¡A¨âªÌ¥i¯à³£¥i¥H³B²z context-free«¬ºAªº»y¨¥)¡F¨ä¥Dn®t§O³q±`¥u¦b©óªí¹Fªº¾A¤Á ©Ê¤ÎåªRªº®Ä²v (linguistic felicity, expressiveness, computational effectiveness)¡A¦Ó«Dª[¸q¸Ñ¨M¯à¤O¡C ¤@¯ë¦Ó¨¥¡Aª[¸q¸Ñ¨M¯à¤O¥Dnµø¨t²Î©Ò±Ä¥Îªº»y¨¥¯S¼x (feature set)¡B³o²Õ»y¨¥¯S¼x¦b¹ê»Ú»y®Æ¤¤ªº¦æ¬°©Î¾÷²v¤À§G (probabilistic distribution)¡A¤Î¨t²Î©Ò±Ä¥Îªº¨Mµ¦¼Ò¦¡ (decision rules)¦Ó©w¡C¦]¦¹¡A§ïµ½¨t²ÎÁZ®Ä¥i¥H¥Ñ´X¤è±¨Ó¶i ¦æ¡C´N»y¨¥¯S¼x¤§¿ï¨ú¦Ó¨¥¡A¥i¥H§Q¥Î¸û¦³Å²§O¤OªºÅ²§O©Ê¯S¼x (discriminative features)¡A¨Ó§@¨Mµ¦¨Ì¾Ú¡A¦Ó¤£¤@©w¬O¥Î«Ü ®e©öÆ[¹î¨ìªº¯S©Ê¦¡¯S¼x(characteristic features)¡C±Ä¥Î¨} ¦nªº»y¨¥¯S¼x¤Î»y¨¥¼Ò¦¡¡A¦³§U©óÅý¨C¤@ºØ¤£¦Pªº»y¨¥¦æ¬°¡A¦b Æ[¹îªÅ¶¡¤¤¡A³£¦³Ó¤ñ¸û¬ð¥Xªº¾÷²v¤À§G¡C¦p¦¹¡A¸û®e©ö§P©w¬Y Ãþ¦æ¬°À³¸Ó±Ä¨ú¦óºØ¯S©w¦æ°Ê¡C¦Ó¨}¦nªº¨Mµ¦¼Ò¦¡«h¥i°§C»~§P ©Ò³y¦¨ªº·ÀI¡C³q±`Ų§O©Êªº¯S¼x¥i¥H¥Ñ¦UºØ¯S©Ê¦¡¯S¼x¥[Åv¤§ «áÀò±o¡A¦]¦¹¡A¨Ï¥Î¨ã¦³¼W±jŲ§O¤O¤Î±j°·©Êªº¾Ç²ßµ¦²¤¡A¥i¦b ¨Ï¥Î¬Û¦P¸ê®Æªº±¡ªp¤U¡A§ïµ½¨t²ÎªºÁZ®Ä¡C µM¦Ó¡A¤@¯ëªº¾÷¾¹Â½Ä¶¨t²Î¦h¼Æ±Ä¥Î¶¥¼h¦¡ªº¤ÀªR¤èªk¡Aµü·J¤À ªRµLªk¸Ñ¨Mªº°ÝÃD§Y±Àµ¹»yªk¤ÀªR¡A»yªk¤ÀªRµLªk¸Ñ¨Mªº§Y±Àµ¹ »y·N¤ÀªR¡A¦AµLªk¸Ñ¨Mªº¡A«K±H±æ©ó©Ò¿×ªº¤@¯ë©Êª¾ÃÑ (real world knowledge, common sense knowledge)¡A¸û¤Ö±j½Õ¦b¨Ï¥Î ¦P¼Ëªº¸ê°T¤U¡A¦p¦ó±j¤Æ¨t²ÎªºÅ²§O¤O (discrimination power)¤Î±j°·©Ê (robustness)¡C ¹ê»Ú¤W¡A¤@¯ë¨t²ÎÁöµM¥i¨Ìµü·J¡B»yªk¡B»y·Nµ¥¨BÆJ³v¤@¤ÀªR¡C µM¦Ó¡A¨t²Î¤ÀªRªº²`«×¨Ã«D¤@¦¨¤£«K¡F¦³¨Ç¸ê®Æ¥¼¥²¦³¥Î¡A¦³¨Ç ¸ê®Æ«h»P¨ä¥¦¸ê®Æ±a¦³¬Û¦Pªº°T®§¡A¦Ó§Î¦¨¤¾Âتº¸ê®Æ¡C³oºØ¥[ ²`¤ÀªR¼h¦¸ªº§@ªk¤jP¬Û·í©ó¼W¥[¨t²Î©Ò¥Îªº»y¨¥¯S¼xªººû«× (dimensionality)¡C¨Ì¾Ú»yµ³B²z¤Î¼Ë«¬ÃѧOµ¥Ãþ¦ü§Þ³Nªº¸gÅç ¡A¤@¨ý¥[²`¤Î¥[²Ó¨t²Î¤ÀªR²`«×¡A©Î¼W¥[¨t²Î¾Ú¥H¨î©w¨Mµ¦ªº¯S ¼x¡A®e©ö¾ÉP¨t²Îªºª¾ÃÑ®w¹L¥÷°¾¦V¬Y¨Ç¯S©wª¾ÃÑ¡A¨Ï¨t²Î¹ï´ú ¸Õ¥Îªº¨Ò¥yÅã±o¬Û·í¦³®Ä¡A¦ý«o°§C¤F¨t²Î¹ï¤@¯ë¿é¤Jªº¾ãÅéÁZ ®Ä¡C ¦]¦¹¡A¥Ø«e·sªº§@ªk¬O±j½Õ¨Ï¥Î¦³Å²§O¤Oªº»y¨¥¯S¼x¤Î»y¨¥¼Ò¦¡ ¡A§Q¥Î±j°·©Êªº¦Û°Ê¾Ç²ßµ¦²¤¨Ó½Õ¾ã¨t²Î°Ñ¼Æ¡F¯à¦b§C¼h¦¸¤ÀªR ¤¤¸Ñ¨Mªº°ÝÃD¡A¾¨¶q§Q¥Î´£°ª¨t²ÎŲ§O¤O¤Î±j°·©Êªº¤è¦¡¡A¦b§C ¼h¦¸¤ÀªR¤¤¥[¥H¸Ñ¨M¡A¦Ó¤£¦A¤@¨ý¦a¥[²`¤ÀªRªº¼h¦¸¡C³oºØ§@ªk ¡A¤£¶È¨Ï±o¨t²Î¯à¹ï¥i¥Îªº»y¨¥¯S¼x§@³Ì¦³®Äªº¹B¥Î¡A¤]¦]¤£¹L ¥÷¨Ï¥Î²`¼hªº»y¨¥¯S¼x¡A¦Ó´î»´¤Fª¾ÃÑ©â¨úªºt¾á[Chia 92, Lin 92]¡C 5. ³]pªÌ»P¨Ï¥ÎªÌªº´Á±æ¤£¦P¡A¨t²Î§ïµ½ªº¤è¦VµLªkÀH«È¤á»Ý¨DÅÜ°Ê «e±¤w¸g»¡¹L¡A¥Ø«e¬ã¨s¤Hû¤w¸g¤ñ¸û¯à»{²M¾÷¾¹Â½Ä¶Ãø¥H¹F¦¨ FAHQMTªº¨Æ¹ê¡A¬G¦b³]p¨t²Î®É¡A¬O¥H¸ê°T©â¨ú (information retrieval)µ¥¤£»Ý¤Ó°ªÂ½Ä¶«~½èªºÀ³¥Î¬°¥Ø¼Ð¡A¨t²Î§ïµ½«h¥Dn ¬O¥H¼W¥[¿é¥Xªº¥iŪ©Ê¬°¥D¡A¥H´Á´î¤ÖĶ«á×µªº¤u§@¡C¤£©¯ªº ¬O¡G(1)n¨D°ª«~½è½Ķªº¨Ï¥ÎªÌÁÙ¬O¦ûµ´¤j¦h¼Æ¡A³o¨Ç¨Ï¥ÎªÌ ªº°ò¥»n¨D¬O½Ķµ²ªGn¯à¥H¥Xª©«~ªº«¬ºA¥X²{ (publishable) ¡A¦Ó¤£¥u¬O¥i¥HŪ±o³q (readable, understandable)¡F(2)§ïµ½ ¿é¥Xªº¥iŪ©Ê¡A¤@¯ë¬O¥i¥H´î¤ÖĶ«á×µ¤u§@¡C¦ý¬O¡A¹ê»Úªº¹ê ÅçÅã¥Ü¡A¨âªÌ¨Ã«Dª½±µ¬ÛÃö¡F¦³®É¬Æ¦Ü©óÁÙ·|µy·L¼W¥[×µªº¦¨ ¥»¡C¦]¬°¡A×µ¦¨¥»¥¿¤ñ©ó×µ¨ì«È¤á©Ò¯à±µ¨üªº¦¨«~©Ò»Ýªº½s ¿è«ü¥O (editing stroke)ӼơA¦Ó¤£¬OµLªk¨ãÅé¿Å¶qªº¡y¥iŪ ©Ê¡z¡C n¸Ñ¨M³o¼Ëªº°ÝÃD¡Aº¥ýn¦³«ÈÆ[ªº¤èªk¡A¨Ó¿Å¶q«È¤á¥i±µ¨üªº ¼Ð·Ç»P¨t²Î¿é¥Xªº®t¶Z¡A¨Ãª½±µ§Q¥Î³o¼Ëªº®t¶Z·í§@¦^õX¡A¨Ó½Õ ¾ã¨t²Îªº°Ñ¼Æ¡A¥H¹F¨ì¯S©w«È¤á©Ò«ü©wªºµü·J¤Î·®æ¡C©Ò¿×«È¤á ¥i±µ¨üªº¼Ð·Ç¡A¨ãÅé¦a»¡¡A´N¬O«È¤á©Ò´£¨ÑªºÂ½Ä¶½d¨Ò¡A©ÎªÌ¥H «e¬°¸Ó¯S©w«È¤á©Ò½Ķªº§¹½Z½Z¥ó¡C®Ú¾Ú³o¨Ç«È¤á¤w±µ¨üªº½Z¥ó ¡A¥H¤Î¨t²Î¹ï³o¨Ç½Z¥ó©Ò²£¥Íªºì©lªì½Z¡A§ÚÌ¥i¥H«Ü«ÈÆ[¦ap ºâ¥X¡A¨âªÌ¤§¶¡ªº®t¶Z¨ì©³n¥Î¦h¤ÖÓ½s¿è«ü¥O¤~¯à×¥¿¹L¨Ó¡A ¥H¤Î»Ý×¥¿³¡¥÷¨ì©³¬O¦]¦óì¦]¦Ó¾ÉP¿ù»~¡C¦p¦¹¡A§Y¥i±N¡y«È ¤áªº´Á±æ¡z»P¡y¨t²Î¿é¥X¡zªº®t¶Z¡A¥Î¨ãÅ骺¡y×µ¦¨¥»¡zªí¥Ü ¥X¨Ó¡A¨Ã¥H¦¹×¥¿¨t²Îªº¦æ¬°¡C n¯à¹ê²{³oºØ¬[ºc¡A¨t²Î¥»¨¥²»Ý¨ã¦³½Õ¾A©Ê (adaptable)¤Î¥i ±±¨î©Ê (controllable)¡A¦Ó¥B¥i¥H±N¨Ï¥ÎªÌªº»Ý¨D§Q¥Î¤@±ø¦^ õXªº¸ô®|¨Ö¤J¨t²Î¡AÀH¨Ï¥ÎªÌªº»Ý¨D½Õ¾ã¨t²Îªº°õ¦æ¦æ¬°¡A¥H¹F ¨ì¾ãÅ骺³Ì¨Î¤Æ¡C¦]¦¹¡A¾÷¾¹Â½Ä¶¨t²Î¥»¨¥²»Ý¬O¤@Ó¥i°Ñ¼Æ¤Æ ªº¦^õX¨t²Î (parameterized feedback control system)¡F¨t²Î ®Ö¤ß«nªº¨Mµ¦¼Ò²Õ¡A¦pÃã·J¤ÀªR¡B¥yªk¤ÀªR¡B»y·N¤ÀªR¡BÂà´« ¼Ò²Õ¤Î¥Í¦¨¼Ò²Õ¡A³£n¯à°÷°Ñ¼Æ¤Æ¡A¤~¦³¥i¯à±N«È¤áªº¦^õX¯Ç¤J ¨t²Î¡Aª½±µ°§C¥i¯àªº×µ¦¨¥»¡C ¹Ï¤T¥Nªí¤@Ó³oºØ¨ã¦³°Ñ¼Æ¤Æ¤Î¦^õX±±¨î¥\¯àªº¾÷¾¹Â½Ä¶¨t²Î¬[ ºc¡C¨t²Î¤¤§@¬°µû¤À¤§¥Îªºµû¤À°Ñ¼Æ¡A¥i¥H³z¹L§¹½Z»Pªì½Z¶¡ªº ®t²§¡A§@»~®t×¥¿¡A¥H½Õ¾ã¨ä°Ñ¼ÆÈ¡C¥Ñ¦¹©ÒÀò±oªº×¥¿°Ñ¼ÆÈ ¡A±N¥i¤ÏÀ³¯S©w¨Ï¥ÎªÌ©Î«È¤á¦b½Ķ¹Lµ{¤¤¡A¹ï¬Y¨Çµü·J¤Î»yªk µ²ºcªº°¾¦n¡A±q¦Ó°§CĶ«á×µªº¦¨¥»¡C <<¹Ï¤T¡G§Q¥Î°Ñ¼Æ¤Æ¨t²Îªº¦^õX±±¨î¥\¯à×¥¿¨t²Î°Ñ¼Æ>> ³oºØ°Ñ¼Æ¤Æªº¨t²Î¡A¨ä¯S©Ê¬O®e©ö§Q¥Î¨t²Îªº°Ñ¼Æ¡A¥H¥i¹w´Áªº ¤è¦¡±±¨î¨t²Îªº°õ¦æÁZ®Ä¡C¦Ó¤@¯ë³W«h¦¡¨t²Î¡A°ò¥»¤W«ÜÃø´x´¤ ¤£¦P³W«h¤§¶¡ªº¤¬°ÊÃö«Y¤Î¬Û¤¬¼vÅT¡AµLªk¹w´ú¼W§R³W«h¹ï¾ãÅé ÁZ®Ä¥i¯à³y¦¨ªº¼vÅT¡C¦]¦¹¡ABehaviorTran ¤j¬ù¦b¶}©l¬ã¨s«á ªº²Ä¤T¦~¡A§Y¶}©lÂà¦V³o¤è±ªº¬ã¨s¡C¥h (1992)¦~¡A¦b¥þ¥@¬É ³Ì¤jªºpºâ»y¨¥¾Ç·|ijCOLING-92¤W¡A´£¥X°Ñ¼Æ¤Æªº¦^õX¨t²Î¬[ ºc®É¡A¤]¤Þ°_¤£¤Ö¾ÇªÌªº¿³½ì»P°Q½×[Su 92b]¡C³oºØ¿Ä¦X¨Ï¥ÎªÌ ´Á±æªº¨t²Î¬[ºc¡A¶Õ±N¦b¥¼¨Óªº¾÷¾¹Â½Ä¶µo®i¤¤¡A§êºt¤@Ó«n ªº¨¤¦â¡C ·s¤@¥Nªº¾÷¾¹Â½Ä¶¨t²Î ºî¦X¤WzÆ[¹î¡A¤@Ө㦳¦^õX¾÷¯àªº°Ñ¼Æ¤Æ¨t²Î¡A³Ì¯à¤ÏÀ³·s¤@¥N¾÷¾¹Â½Ä¶ ªº»Ý¨D»P¯S¦â¡C³z¹L°Ñ¼Æ¤Æªº¨t²Î¼Ò¦¡¡A¥i¥H«Ü®e©ö´x´¤¨t²ÎÅÜ°Ê©Ò³y¦¨ªº ¼vÅT¡A¨Ã¥B«Ü²M·¡¦a©w¸q¥X¨t²Îªº³Ì¨Î¤Æ¥Ø¼Ð¡A¦Ó¥Î¤@®M¦³¨t²Îªº¤è¦¡¡A³v ¨B¹F¦¨§ïµ½¨t²ÎÁZ®Äªº¥Ø¼Ð¡C¦p¦¹¡A±N¥i¦³®ÄÁקK³W«h¦¡¨t²Î¼W§R³W«h®É¡A Ãø¥H¹wª¾¹ï¾ãÅéÁZ®Ä¤§¼vÅTªºµ~¹Ò¡C¥t¤@¤è±¡A¾÷¾¹Â½Ä¶µo®iªÌ¥i§Q¥Î°Ñ¼Æ ¤Æªº¨t²Î¬[ºc¡A¦³¨t²Î¦a±±¨î¤Î¹w´ú¾÷¾¹Â½Ä¶¨t²Îªº¦æ¬°¡F¨t²Î¥i§Q¥Î«ÈÆ[ ªºµû¤À¨ç¼Æ§@³Ì§C¦¨¥»ªº¨Mµ¦§PÂ_¡F©Ò»Ýªºª¾ÃÑ®w¥i§Q¥Î»y®Æ¦Û°Ê©Î¥b¦Û°Ê ¦a©â¨ú¡AÁקK¤H¤u©â¨ú©Ò¾D¹Jªº°ª¦¨¥»¡B§C²[»\±¡B¤£¤@P¤Î¤£©öºûÅ@§ó·s µ¥°ÝÃD¡C¦Ó¨Ï¥ÎªÌªº»Ý¨D«h¥i³z¹L¨t²Îªº¦^õX¥\¯àª½±µ¿Ä¤J¨t²Î¡A¯u¥¿°§C Ķ«á×µªº¦¨¥»¡C BehaviorTran ²¤¶ BehaviorTran (¥ý«eºÙ¬° ArchTran) ¬OP»·¬ì§Þ©Òµo®iªº¤@®M^º~¾÷¾¹Â½ Ķ¨t²Î¡A¤]¬O°ê¤º²Ä¤@®M¦Û¦æ¬ãµoªº¾÷¾¹Â½Ä¶¨t²Î¡C BehaviorTran ¦´Áªº¬ã¨s¬O¦b²M¤j¹q¾÷¬ã¨s©Ò¶i¦æªº¡C 1985 ¦~ 5 ¤ë¥ÑĬ §J¼Ý±Ð±Â©Ò«ü¾Éªº^º~¾÷¾¹Â½Ä¶¤p²Õ¡A¦b¥u¦³¨â¦W¦¨ûªº±¡ªp¤U¶}©l³oÓp ¹º¡C¥Ø«e³oÓp¹º¦b^¸s¥ø·~ªº¤ä«ù¤U¡A¤wÂಾ¨ì¬ì¾Ç¶é°ÏªºP»·¬ì§Þ¤½¥q Ä~Äò¬ã¨s¡A°Ñ»P¬ã¨s»P¹B§@ªº¤Hû¤]ÂX¥R¨ì¤F¤@¡B¤G¤Q¤H¡C¨Ã¥B¦Û 1989 ¦~ °_¡A¥¿¦¡¦¨¥ß½ĶªA°È¤¤¤ß¡A±µ¨ü«È¤á©e°U¡A±q¨Æ¹ê»ÚªºÂ½Ä¶·~°È¡C ¦P®É¡A ³z¹L½Ķ¤¤¤ßªº¦^õX¡A ¤£Â_§ïµ½¨t²ÎªºÁZ®Ä¤Î¨Ï¥ÎªÌ¤¶±¡C ¦´Á BehaviorTran ½Ķªº¥Dn¹ï¶H¡A ¬O»P¹q¸£¦³Ãöªº¤â¥U¡B¨Ï¥Î»¡©ú¥H¤Î®ÑÄy µ¥¡CÀHµÛ±M·~Ãã·Jªº¼W¥[¡A½Ķªº¤u§@¤]³vº¥ÂX¥R¨ì¹q¾÷¡B¾÷±ñ¡B¯èªÅ¡B¤ß ²z¾Çµ¥»â°ì¡C BehaviorTran ªº¨t²Î¼Ò²Õ BehaviorTran ½Ķ¨t²Îªº°ò¥»¬[ºc¡AÄÝ©óÂà´«¦¡ªº¾÷¾¹Â½Ä¶¨t²Î¡C ¦ý¦b¶Ç ²ÎªºÂà´«¦¡¬[ºc¤¤¡A¥[¤J¤F³\¦h¥ý¶iªº§Þ³N¡A¨Ï¸Ó¨t²Î§ó¯à¾AÀ³·s¤@¥N¾÷¾¹ ½Ķªº»Ý¨D¡C¨t²Îªº¤ÀªR¡BÂà´«¡B¦X¦¨µ¥³¡¥÷¡A¤jP¥Ñ¥H¤Uªº³nÅé¼Ò²Õ¨Ó§¹ ¦¨¡G ¡´µü·J¤ÀªR¾¹ (Lexical Analyzer) µü·J¤ÀªR¾¹ªº¥\¯à¡A¥Dn¦b©ó¬É©w¿é¤J¤§ì©l»y¨¥ªº³æµü¤Î¥y¤l½d³ò¡A¬d¥X ¥y¤l¸Ì¨CÓµü¶µªº°ò¥»¸ê®Æ¡A¥H¨ÑåªR (parse) ¥yªkµ²ºc¤§¥Î¡C ¥Dn³B²z ªº¶µ¥Ø¦³¡G 1. ®æ¦¡³B²z¡G°w¹ï¤£¦P¿é¤J®æ¦¡¡A°Ï¤À¥X±Æª©«ü¥O¤Î¥»¤å³¡¥÷¡C 2. ¤Á¥y³B²z¡G¬É©w¨C¤@¥y^¤åì¤åªº½d³ò¡A §ä¥XåªR³æ¦ì (§Y¨t²Î©w¸qªº ¥y¤l )¡A¥H¬°åªR¤§¥Î¡C 3. µü¨å¬d¾\¡G±qµü¨å¤¤¨ú¥X¥y¤l¸Ì¨C¤@Óµü¶µªº¸ê®Æ¡A¥H¨ÑåªR¤§¥Î¡C Y ¥y¤l¸ÌªºµüµLªkª½±µ¥Ñµü¨å¤¤§ä¨ì¡A«h³z¹Lºcµü¤ÀªR¡A¸Õ¹Ï¤ÀªR¥X¸Óµüªºµü ®Ú¡Bµüº¡Bµü§Àµ¥¡A¦A¨Ì¾Úµü®Ú¡A©â¨úµü¶µ¸ê®Æ¡C¦pªGºcµü¤ÀªR¥¢±Ñ¡A«hµ¹ ¤©³oÓµü¤º©wµüÃþ¡C 4. ¯S®íµü§ÎÁÙì¡G¦pÁY¼gµü I'm ÁÙ즨¥¼ÁY¼gªº§Î¦¡ I am µ¥¡C ¡´«e³B²z¾¹ (Preprocessor) ¡u«e³B²z¡vªº§@¥Î¥Dn¦b©óåªR¥y¤l«e¡A±N¤@¨ÇµüÃþ©Î¥yªk¤Wªº°T®§¾¨¥i¯à ´£¨Ñµ¹åªR¾¹ (parser) ¥H²¤ÆåªRªº¹Lµ{¡C«e³B²zªº¤u§@¬Û·íº¾¸H¡A¥H¤U ¸Õ¦CÁ|¤@¤G»¡©ú¡G 1. °Ï¤À²Å¸¹»Pµü¶µ¡G^¤å¦r¥À¸g±`³Q¥Î¨Ó§@¬°²Å¸¹¡A ¨Ò¦p¡ua¡v¡B¡ui¡vµ¥ µ¥¡C ¥Ñ©ó¡ua¡v¥»¨¤S¬O¤@Ó«nªº^¤å«aµü¡A©Ò¥H¦b¶i¤JåªR¤§«e¡A¥²»Ý ¥ý§PÂ_¦b¸Ó¤å½Z¤¤¡A¡ua¡v¨s³º¬O²Å¸¹ÁÙ¬O«aµü¡C 2. ¤½¦¡¿ëÃѤγB²z¡GÀË´ú¤å³¹¤¤ªº¤½¦¡¡B¹Ïªíµ¥¡A¦Ó«á¥[¥H¼Ð°O¡A ¤£¥[¤J åªR¡C 3. §R°£µüÃþª[¸q¡GÀˬd¤å¥y¤¤¦UµüªºµüÃþ¡A ®Ú¾Ú²Îp¼Ò¦¡¡B¨t²ÎªºµüÃþ¨Ã ¦s¨î³W«h¤Î³æµüµüÃþ¨Ã¦s¨î¡A¹ï¦UµüÃþ²Õ¦X¤©¥Hµû¤À¡A§R°£¸û¤£¥i¯àªº µüÃþ²Õ¦X¡A¥H´î¤ÖåªR®Éªºª[¸q¡C 4. ³B²z¦¨¹ï¦¡³s±µµü¡GÀË´ú¥y¤l¤¤©Ò§tªº¦¨¹ï¦¡³s±µµü¡A¦p¡uboth... and ¡v¡B¡uneither... nor¡vµ¥¡A±N¤§µø¬°¹ïµ¥³s±µµü³B²z¡C 5. ³B²z¤ÀÂ÷¦¡¦¨»y¡G®Ú¾Úµü¨å°O¸ü¡AÀˬd¬O§_¦³¤ÀÂ÷¦¡¦¨»y (¦p¡uturn... on¡v¡A¡ulook... at¡vµ¥°Êµü©M¤¶µü¤¤¶¡¥i¥H´¡¤J¦Wµü©Î°Æµüªº½Æ¦X°Êµü ) ¡A¦pªG¦³³o¨Ç¤ÀÂ÷¦¡¦¨»y¡A«hµø¬°¤@Óµü¶µ¡A¨Ã±N¸Óµü¶µªº¸ê®Æ¨ú¥X¡C¦¹Ãþ ¦¨»y¤¤¶¡©Ò´¡¤Jªº¦¨¥÷¥i¯à¨Ï°Êµü¸ò¤¶µüÂ÷±o«Ü»·¡]¦p¡uturn the radio on¡v¡^¡A¥B¤¤¶¡¥i¯à¥Î¨ì¬Û¦Pªº¤¶µü¡C¦]¦¹¡A¥²»Ý§@¬Y¨Ç³B²z¤~¯à½T©w¦¨»y ªº½d³ò¡C ¸g¹L¤Wz¤Î¨ä¥L«e³B²z¨BÆJ¤§«á¡A¥y¤l±N³Q¤À¸Ñ¦¨¤@ÓÓ¿W¥ßªºµü¶µ¡A³o®É ¤~¥¿¦¡¶i¤JåªR¶¥¬q¡C ¡´åªR¾¹ (Parser) åªR¾¹ªº¥\¯à¬O¤ÀªR¥y¤lªºµ²ºc¡A®Ú¾Ú¨CÓ¦rªºµüÃþ¡A§ä¥X¥y¤¤¦U¦¨¥÷ªº¶¥ ¼hÃö«Y¡A «Øºc¦¨åªR¾ð (parse tree) ¡A ¨ÃÀˬd¨ä»y·N¬O§_¾A·í¡C BehaviorTran ªºåªR¾¹¦³¤U¦C´X¶µ¯S¦â [Su 90]¡J 1¡D±j¤Æªº LR åªR¾¹¡G¥H LR åªR¾¹µo®i¥X¨Ó¡A¨Ã®Ú¾Ú¨t²Îªº»Ýnקï¡C 2¡D¡uTop-down ¹LÂo¡v¥[¡uBottom-up åªR¡v¡G BehaviorTran °ò¥»¤W¬O±Ä bottom-up¡]¥Ñ¤U¦Ó¤W¡^ªºåªR¤è¦¡¡A ¦ý¦P®É¤]§Q¥Î¤åªk¤Á¤Àªº¤è¦¡¡A±Nå ªR¥Îªº¤åªk¤Á¤À¦¨¼ÆÓ¦¸¤åªk (subgrammars)¡A¨Ï¨ä¨ã¦³ top-down ¤è¦V¡] ¥Ñ¤W¦Ó¤U¡^ªº¹w´ú¥\¯à¡A«K©ó»yªk²{¶Hªº´yz¡A¤Î¹LÂo¤@¨Ç¤£¾A·íªº¤ÀªR¡C 3¡D±j¤Æªºµü²Õ«ß¡G¦bµü²Õ«ß¤¤¥[¤J¡u¨î±ø¥ó¡v(condition) ªºÀˬd¡A §Q ¥Î»y·N¤ÀªR¡A±j¤Æ¤ÀªRªº¥\¯à¡A´£°ªåªRªº¥¿½T©Ê¡C 4¡D°ÊºAµü²Õªí¡GÀx¦såªR¹Lªºµü²Õ¸ê®Æ¡A§@¬°¦^·¹ (backtrack) ®Éªº°Ñ¦Ò °T®§¡AÁקK«½ÆåªR¡C 5¡D³¡¥÷åªR¡G®Ú¾Ú¨t²Îªº¯S®ín¨D¡A ¥H¤@Ó¥y¤l¤¤ªº³¡¥÷µü¦ê (¦Ó«D§¹¾ã ¥y¤l ) ¬°åªR³æ¦ì¡A¶i¦æåªR¡C ¨Ò¦p¡AåªR¼ÐÃD©M¤ÀÂ÷¦¡¦¨»yªº¨üµü®É¡A Àˬd¨ä¤¤¬O§_§t¦³¦Wµü²Õµ¥µ¥¡C 6¡D¿ù»~×Å@ (fail-soft) ¥\¯à¡G¥y¤l¨Ì¾Ú¨t²Î¤åªkµLªkåªR¥X§¹¾ãªºµ²ºc ®É¡A¥Ñ¤wåªRªº³¡¥÷µ²ªG¡A¹Á¸Õ§ä¥X¥i¯àªºµ²ºc¡A¥H¨Ï¨t²Î¯àÄ~Äò³B²z¡A´î ¤ÖĶ«á×µªºt¾á¡C 7¡D®Ú¾Úµû¤À¨ç¼Æ¡A §R°£¥i¯à©Ê¸û§CªºåªR¸ô®|¡G°£¤F§Q¥Î»y¨¥¾Çª¾ÃѨӧR °£ª[¸q¤§¥~¡A¨Ã¥H¾÷²v¤è¦¡¡ApºâµüÃþ¡B»yªk©M»y·Nªº¤À¼Æ¡A±N¥¼¹F³]©w¼Ð ·ÇªºåªR¸ô®|§R°£¡A¥H´î¤Öª[¸qªº¼Æ¥Ø»PåªRªº®É¶¡¡C³Ì«á¨Ã¿ï¥X³Ì¥i¯àªº åªRµ²ªG¡AÄ~Äò°õ¦æÂà´«¤Î¥Í¦¨ªº°Ê§@¡C ¡´Âà´«¾¹ (Transfer Module) åªR«á¡A¨Ì¾Ú²£¥Íªº^¤ååªR¾ð¡A^¤¤Âù»yµü¨å¤Î¨t²ÎªºÂà´««ß¡A°õ¦æ^¤å µ²ºc¨ì¤¤¤åµ²ºcªºÂà´«¡C³z¹L^¤¤µü·JÂà´«¡A¿ï¾Ü¾A·íªº¤¤¤åµü·J¡A¨Ã³z¹L µü²ÕÂà´«¡A¨Ï¨ä²Å¦X¥Ø¼Ð»y (¤¤¤å ) ªºµ²ºc¤Îµü§Ç¡C ¡´¦X¦¨¾¹ (Synthesis Module) ¦X¦¨¾¹t³d¦X¦¨¤¤¤å¯S¦³ªºµ²ºc¡A¥[¤J¥²nªº¤¤¤å×¹¢µü¡A¥H²£¥Í¬yºZªº¤¤ ¤å¿é¥X¡C¨Ò¦p¡G²£¥Í¤¤¤å¯S¦³ªº¡u§â¡v¡B¡u³Q¡vµ¥¯S®í¥y«¬¡A¥[¤J¡uªº¡v¡B ¡u¤F¡vµ¥¤¤¤å¯S¦³µü·J¡C ¡´µü¨å¡]Lexicons¡^ µü¨å¸ÌÀx¦s¤¤^¤åµü·J¸ê®Æ¡A§@¬°Â½Ä¶®É^¤¤Âà´«¤§¥Î¡C BehaviorTran ªº µü¨å¨ÌÀ³¥Îªº¼h¦¸¤À¦¨´X³¡µü¨å¡J 1¡D¤@¯ëµü¨å¡G¦s¤@¯ë^º~µü¨å¸Ì¬d±o¨ìªº³æµü¡A¦p¡Glanguage¡C 2¡D¤@¯ë¦¨»yµü¨å¡G ¦s¤@¯ë^º~µü¨å¸Ì¬d±o¨ìªº¦¨»y©M½Æ¦Xµü¡A ¨Ò¦p "in order to"¡C 3¡D¤ÀÂ÷¦¡¦¨»yµü¨å¡G ¦s¤ÀÂ÷¦¡¦¨»y¡C ¤ÀÂ÷¦¡¦¨»y¬°°Êµü©M¤¶°Æµü (particle) ²Õ¦¨ªº°Êµü²Õ¡A ¨ä»«»y¥X²{¦b°Êµü©M¤¶°Æµü¤¤¶¡¡A¤]¥i¯à¥X²{ ¦b¤¶°Æµü«á±¡A¨Ò¦p¡uturn on¡v¡F °Êµü©M¤¶µü¤¤¶¡¡A¤]¥i´¡¤J°Æµü¡A¦p¡u look (carefully) at¡v¡C 4¡D±M·~µü¨å¡G¦s¯S©w»â°ìªº±M·~µü·J¡A¨Ò¦p¡G"disk drive" (¹q¸£Ãþ )¡C 5¡D«È¤áµü¨å¡G¦s¨Ï¥ÎªÌ¯S¦³©Î«ü©w¤¤Ä¶ªº³æµü¡C 6¡D±M®×µü¨å¡G¦s«È¤á±M®×¯S¦³©Î«ü©w¤¤Ä¶ªº³æµü¡C ¥Ñ©ó½Ķªº¤å½Z¤¤¡A¤@Óµü¦b¤£¦Pµü¨åªº¥Îªk¤Î·N¸q¥i¯à¦P®É¥X²{¡A¦]¦¹¥² ¶·±N¦U³¡µü¨å¸Ìªº¸ê®Æ¥[¥HÁp¨Ö (unification)¡A¥H±o¨ì©Ònªº¤¤¤å¡Cµü¨å Áp¨Ö®É¡A¸ê®Æ¨ú¥ÎªºÀu¥ý¶¶§Ç¬°¡G±M®×µü¨å³Ì¥ý¡B«È¤áµü¨å¨ä¦¸¡B±M·~µü¨å ¦A¨ä¦¸¡A¤@¯ëµü¨å³Ì«á¡C ¡´¨Ï¥ÎªÌ¤¶± 1¡DBehaviorTran ¨Ï¥ÎªÌ¤u§@Àô¹Ò¡G BehaviorTran ¨t²Î¬°¨Ï¥ÎªÌ´£¨Ñ¥æ½Í ¦¡ªº¨Ï¥ÎÀô¹Ò¡A¨ó§U¨Ï¥ÎªÌ¥H²«K¦Ó¦³¨t²Îªº¤è¦¡±±¨î BehaviorTran ©Ò°õ ¦æªºÂ½Ä¶¤u§@¡C¨Ò¦p§@«÷¦rÀˬd¡A¥HÀˬd¿é¤Jªº^¤å¬O§_¦³«÷¦r¿ù»~¡FÀˬd ¤å½Z¤¤¬O§_¦³µü¨å¥¼«Øªºµü¡A¥H«K±N¥Í¦r¸É¤Jµü¨å¡F½s¿èÀɮסA¥H²Å¦X½Ķ ¨t²Îªº¿é¤J®æ¦¡¡F¥H¤Î¶i¦æ½Ķ¤u§@µ¥¡C 2. Ķ«á×µ¤u¨ã¡G BehaviorTran Ķ«á×µ¤u¨ã¬O®Ú¾Ú¾÷¾¹Â½Ä¶¨t²Îªº¯S ¦â¦Ó³]p¡A§@¬°×µ¤Hûªº»²§U¤u¨ã¡A¥H´£°ª×µªº³t«×¡C³oÓ½s¿è³nÅé¥H ¤¤¤åµü²Õ¬°³æ¦ì¡A´£¨Ñ¨Ï¥ÎªÌ¦hºØ¥\¯à¡A¦p¡Gµü²Õªº½Õ´«¡B±`¥Îµüªº¿ï¾Ü¥H ¤Î¦C¥Xµü¨å¤¤^¤åµü·Jªº¨ä¥L¤¤¤å½Ķµ¥¡F ¦P®ÉÁÙ¥i¥H¶i¤@¨B°t¦X BehaviorTran ¦^õX¸ê®Æµ¹¬ã¨s¤Hû¡A§@¬°§ï¶iªº°Ñ¦Ò¡C ¡´¥ý¶iªº§Þ³N ¦b¥ý¶iªº§Þ³N¤è±¡A BehaviorTran ±Ä¥Î¥H»y®Æ¬°°ò¦¡A¬[ºc¦b°ª¼h»y¨¥ª¾ ÃѤWªº²Îp¤è¦¡¡Aµo®i¥X¤@¨t¦Cªº§Þ³N¨Ó¨ó§Uª¾ÃÑ©â¨ú¡B¸Ñ¨Mª[¸q¤Î±±¨î¨t ²Îªº¦æ¬°¡C³o¨Ç§Þ³N¥]¬A¡G 1. §Q¥Î¾ã¦X¦¡ªºµû¤À¨ç¼Æ¡A ¸Ñ¨Mµü·J¡B »yªk¤Î»y·Nµ¥¦U¼h¦¸¤§ª[¸q²{¶H [Su 88,91, Liu 90, Chan 92a, Lin 92]¡C 2. ¨Ï¥Î¥HŲ§O¤O¡B±j°·©Ê¬°¾É¦Vªº¦Û°Ê¾Ç²ßµ¦²¤¡A½Õ¾ã¨t²Î°Ñ¼Æ¡A ¥H´î¤Ö ¤£¥²nªº²`¼h¤ÀªR¡A¨Ã´£°ª¨t²Îªº°õ¦æÁZ®Ä [Chia 92, Lin 92]¡C 3. §Q¥Î¾÷²v¦¡Âà´«¤Î¥Í¦¨¼Ò¦¡¡A ¦Û°Ê§ä¥XÂà´«³æ¦ì (transfer units) ¤Î Âà´«³W«h¡C¨Ã§Q¥Î¥Ø¼Ð»yªº¤åªk¡A°f¦V§ä¥X¨t²Îªº¥Í¦¨³W«h¡A¥H²£¥Í²Å¦X¥Ø ¼Ð»y»yªkªºÂ½Ä¶¡A ÁקK¤@¯ëÂà´«¦¡Â½Ä¶¨t²Î©Ò²£¥Íªº¥Íµw½Ķ [Chan 92b] ¡C 4. ¦b¨t²Î¤¤¨Ö¤J¦^õX¥\¯à¡A ¨Ï¨t²Îªº¦æ¬°¯à¨Ì¨Ï¥ÎªÌªº»Ý¨D½Õ¾ã¡A´î¤Ö©Ò »ÝªºÄ¶«á×µ [Su 92b]¡C ¦b³o¨Ç¥ý¶i§Þ³Nªº¤ä´©¤U¡A BehaviorTran ±N¥i³v¨B¾ã¦X»y¨¥¾Ç©M²Îp¾Çªº ª¾ÃÑ¡A«Øºc¤@Ө㦳¤@P©Êªº¤j«¬¦ÛµM»y¨¥³B²z¨t²Î¡C¨Ã¥H¦Û°Ê¤Æ¡B¨t²Î¤Æ ªº¤è¦¡¡A¨D¨ú»y¨¥¼Ò¦¡°Ñ¼Æ¡C§ï¶i¶Ç²Îªº¾÷¾¹Â½Ä¶¨t²Î¡AÁÚ¦V·s¤@¥N¾÷¾¹Â½ Ķ¨t²Î©Ò°l¨Dªº¥Ø¼Ð¡C µ²»y ¥Ñ©ó¾÷¾¹Â½Ä¶ªº»Ý¨D¶V¨Ó¶V¢¤Á¡A¾÷¾¹Â½Ä¶¤Î¬ÛÃöªº¦ÛµM»y¨¥³B²z§Þ³N¡A¤w ¦C¤J¬ü¡B¤é³\¦hµÛ¦W¬ã¨s¾÷ºcªº«ÂI¬ì§Þ¤§¤º¡C¦]¦¹¡A°ê¤H¦³¥²n¦b²{¦³ªº °ò¦¤W¡AÄ~Äò¶}µo§ó¥ý¶iªºÂ½Ä¶§Þ³N¡C ¶Ç²Îªº³W«h¦¡¾÷¾¹Â½Ä¶¨t²Î¬[ºc¡A¤w³vº¥ÅãÅS¥X¨ä¥ý¤Ñ¤Wªº¨î¡C¦]¦¹¡A¥» ¤å¤¤©Ò´£¨ìªº¤@¨Ç«nijÃD¡A¥i¬°¦³¤ß¥[¤J¬ãµo¦æ¦Cªº¤H¡A´£¨Ñ¤@Ó©ú½Tªº ¤è¦V¡C²¨¥¤§¡A¥¼¨Óªº¾÷¾¹Â½Ä¶¨t²Î±N·|³vº¥±j½Õ¡G (1) °Ñ¼Æ¤Æªº¼Ò¦¡¡F (2) ¥H»y®Æ¬°°ò¦¡A¥H²Îp¬°¤u¨ã¡A«Øºc¦b°ª¼h»y¨¥ª¾ÃѤWªº»y¨¥¼Ò¦¡¡F (3) ¤@P©Êªºª[¸q¸Ñ¨M¤è¦¡»P«ÈÆ[ªºµû¤À¾÷¨î¡F (4) ±j¤ÆŲ§O¤O¤Î±j°·©Êªº¦Û°Ê¾Ç²ßµ¦²¤¡F (5) Âù¦V¦¡ªº¨t²Î³]p¡F (6) ¨Ì¨Ï¥ÎªÌ»Ý¨D½Õ¾ã¨t²Î¦æ¬°ªº¦^õX±±¨î¥\¯à¡C ¦Ó³o¨Ç¤è¦V¤]¬O BehaviorTran ³v¨B§¹¦¨¤Î°l¨Dªº¥Ø¼Ð¡C °Ñ¦Ò®Ñ¥Ø¡G [Chan 92a] Chang, J.-S., Y.-F. Luo and K.-Y. Su, "GPSM: A Generalized Probabilistic Semantic Model for Ambiguity Resolution," Proceedings of ACL-92, pp. 177--184, 30th Annual Meeting of the Association for Computational Linguistics, University of Delaware, Newark, DE, USA, 1992. [Chan 92b] Chang, J.-S. and K.-Y. Su, "A Corpus-Based Statistics-Oriented Transfer and Generation Model for Machine Translation," manuscript, 1992. [Chen 91] Chen, S.-C., J.-S. Chang, J.-N. Wang and K.-Y. Su, "ArchTran: A Corpus-Based Statistics-Oriented English-Chinese Machine Translation System," Proceedings of Machine Translation Summit III, pp. 33--40, Washington, D.C., USA, 1991. [Chia 92] Chiang, T.-H., Y.-C. Lin and K.-Y. Su, "Syntactic Ambiguity Resolution Using A Discrimination and Robustness Oriented Adaptive Learning Algorithm", Proceedings of COLING-92, vol. I, pp. 352--358, 14th Int. Conference on Computational Linguistics, Nantes, France, 1992. [Hutc 86] W.J. Hutchins, Machine Translation: Past, Present, Future, Ellis Horwood Limited, West Sussex, England, 1986. [JEIDA 89] JEIDA, A Japanese View Of Machine Translation In Light Of The Considerations And Recommendations Reported By ALPAC, U.S.A., M. Nagao (chairman), Machine Translation System Research Committee, Japan Electronic Industry Development Association, 1989. [Lin 92] Lin, Yi-Chung, Tung-Hui Chiang and Keh-Yih Su, "Discrimination Oriented Probabilistic Tagging," Proceedings of ROCLING-V, ROC Computational Linguistics Conference V, pp. 87--96, 1992. [Liu 90] Liu, C.-L., J.-S. Chang and K.-Y. Su, "The Semantic Score Approach to the Disambiguation of PP Attachment Problem," Proceedings of ROCLING-III, pp. 253-270, 1990. [Su 87] Su, K.-Y., J.-S. Chang and H.-H. Hsu, "A Powerful Language Processing System for English-Chinese Machine Translation System," Proceedings of 1987 Int. Conf. on Chinese and Oriental Language Computing, 260--264, Chinese Language Computer Society, Chicago, Illinois, USA, 1987. [Su 88] Su, K.-Y. and J.-S. Chang, "Semantic and Syntactic Aspects of Score Function," Proc. of COLING-88, vol. 2, pp. 642--644, 12th Int. Conf. on Computational Linguistics, Budapest, Hungary, 1988. [Su 90] Su, K.-Y. and J.-S. Chang, "Some Key Issues in Designing MT Systems," Machine Translation, vol. 5, no. 4, pp. 265-300, 1990. [Su 91] Su, K.-Y., J.-N. Wang, M.-H. Su and J.-S. Chang, "GLR Parsing with Scoring," In M. Tomita (ed.), Generalized LR Parsing, Chapter 7, pp. 93-112, Kluwer Academic Publishers, 1991. [Su 92a] Su, K.-Y and J.-S. Chang, "Why Corpus-Based Statistics-Oriented Machine Translation," Proceedings of TMI-92, pp. 249--262, 4th Int. Conf. on Theoretical and Methodological Issues in Machine Translation, Montreal, Canada, 1992. [Su 92b] Su, K.-Y., M.-W. Wu and J.-S. Chang, "A New Quantitative Quality Measure for Machine Translation Systems," Proceedings of COLING-92, vol. II, pp. 433--439, 14th Int. Conference on Computational Linguistics, Nantes, France, 1992.