I
Q6HSWHPEHUWKH$OOHQ,QVWLWXWHIRU$UWL¿FLDO,QWHOOLJHQFH$,XQYHLOHGD
FRPSXWHUSURJUDPFDOOHG$ULVWRWKDWFRXOGFRUUHFWO\DQVZHUPRUHWKDQ
SHUFHQWRIWKHTXHVWLRQVRQDQHLJKWKJUDGHVFLHQFHWHVW3DVVLQJDPLGGOH
VFKRROH[DPPLJKWVRXQGPXQGDQHEXWLW¶VFRPSOLFDWHGIRUFRPSXWHUV
$ULVWRZDVDEOHWR¿QGWKHDQVZHUVZLWKLQELOOLRQVRIGRFXPHQWVE\XVLQJ
natural language processing (NLP),DEUDQFKRIFRPSXWHUVFLHQFHDQG
DUWL¿FLDOLQWHOOLJHQFHWKDWHQDEOHVFRPSXWHUVWRH[WUDFWPHDQLQJIURP
XQVWUXFWXUHGWH[W7KRXJKZH¶UHVWLOODORQJZD\IURPPDFKLQHVWKDWFDQ
XQGHUVWDQGDQGVSHDNKXPDQODQJXDJH1/3KDVEHFRPHSLYRWDOLQPDQ\
DSSOLFDWLRQVWKDWZHXVHHYHU\GD\LQFOXGLQJGLJLWDODVVLVWDQWVZHEVHDUFK
HPDLODQGPDFKLQHWUDQVODWLRQ
WORDS ARE HARD
5HSOLFDWLQJWKHODQJXDJHSURFHVVLQJFDSDELOLWLHVRIWKHKXPDQPLQGLVDKLVWRULF
SDLQSRLQWIRUDUWL¿FLDOLQWHOOLJHQFH,PDJLQHDQ$,DJHQWWKDWPXVWUHVSRQGWR
ZHDWKHUFRQGLWLRQTXHULHVLWKDVWRXQGHUVWDQGDOOWKHGL̆HUHQWZD\VVRPHRQH
FDQDVNDERXWWKHZHDWKHU
- +RZLVWKHZHDWKHUWRGD\"
- :LOOLWUDLQWRPRUURZ"
- :KHQZLOOLWVWRSUDLQLQJ"
- ,VLWVXQQ\LQ&KLFDJR"
- :LOOLWEHZDUPHUWRPRUURZ"
- :KLFKGD\VDUHVXQQ\QH[WZHHN"
$QGODQJXDJHRIWHQFDUULHVKLGGHQPHDQLQJVWKDWLPSO\JHQHUDONQRZOHGJH
DERXWWKHZRUOGDQGKRZREMHFWVUHODWH&RQVLGHUWKHIROORZLQJTXHULHV
- :LOOWKHZHDWKHUEHJRRGIRUVRFFHUWRPRUURZ"
- ,VLWVQRZLQJLQWKHNLWFKHQ"
$Q\KXPDQKHDULQJWKH¿UVWVHQWHQFHZLOONQRZWKDW\RX¶UHLPSOLFLWO\DVNLQJ
ZKHWKHULWZLOOEHVXQQ\WRPRUURZ²RUSHUKDSVMXVWZKHWKHULWZRQ¶WUDLQ$VIRU
WKHVHFRQGVHQWHQFHSHRSOHNQRZLWGRHVQ¶WVQRZLQWKHNLWFKHQ%XWHQFRGLQJ
WKLVNLQGRIEDFNJURXQGNQRZOHGJHDQGUHDVRQLQJLQDUWL¿FLDOLQWHOOLJHQFH
V\VWHPVKDVDOZD\VEHHQDFKDOOHQJHIRUUHVHDUFKHUV