635-646, 2004. (in Japanese)
Top5. Since this approach relies on a morphological
Satoshi Sekine, Kiyoshi Sudo, Yusuke Shinyama,analyzer, applying the QBTE Model 1 to QA tasks
Chikashi Nobata, Kiyotaka Uchimoto, and Hitoshi Isa-of other languages is our future work.
hara, NYU/CRL QA system, QAC question analysisand CRL QA data,in Working Notes of NTCIR Work-shop 3(2002).Acknowledgment
Jun Suzuki, Yutaka Sasaki, and Eisaku Maeda: SVM An-This research was supported by a contract with the
swer Selection for Open-Domain Question Answer-National Institute of Information and Communica-
ing,Proc. of Coling-2002, pp. 974–980 (2002).tions Technology (NICT) of Japan entitled, “A study
Jun Suzuki, Hirotoshi Taira, Yutaka Sasaki, and Eisakuof speech dialogue translation technology based on
Maeda: Directed Acyclic Graph Kernel,Proc. of ACLa large corpus”.
2003 Workshop on Multilingual Summarization andQuestion Answering - Machine Learning and Beyond,pp. 61–68, Sapporo (2003).References
Ingrid Zukerman and Eric Horvitz: Using MachineAdam L. Berger, Stephen A. Della Pietra, and Vincent J.Learning Techniques to Interpret WH-Questions,Della Pietra: A Maximum Entropy Approach to Nat-Proc. of ACL-2001, Toulouse, France, pp. 547–554ural Language Processing,Computational Linguistics,(2001).Vol. 22, No. 1, pp. 39–71 (1996).Question Type #Q MRR T5 MRR’ T5’Appendix: Analysis of Evaluation Results w.r.t.
RANK 7 0.18 0.29 0.54 0.71Question Type — Results of QBTE from the first-
BOOK 6 0.31 0.50 0.47 0.67ranked paragraph (NB: No information about these
AWARD 9 0.17 0.33 0.34 0.56N LOCATION 2 0.10 0.50 0.10 0.50question types was used in the training phrase.)
VEGETABLE 10 0.31 0.50 0.34 0.60COLOR 5 0.20 0.20 0.20 0.20Question Type #Qs MRR T5 MRR’ T5’NEWSPAPER 7 0.61 0.71 0.61 0.71GOE 36 0.30 0.36 0.41 0.53WORSHIP 8 0.47 0.62 0.62 0.88GPE 4 0.50 0.50 1.00 1.00SEISMIC 1 0.00 0.00 1.00 1.00N EVENT 7 0.76 0.86 0.76 0.86N PERSON 72 0.30 0.39 0.43 0.60EVENT 19 0.17 0.21 0.41 0.53PERSON 282 0.18 0.21 0.46 0.55GROUP 74 0.28 0.35 0.45 0.62NUMEX 19 0.32 0.32 0.35 0.47SPORTS TEAM 15 0.28 0.40 0.45 0.73MEASUREMENT 1 0.00 0.00 0.00 0.00BROADCAST 1 0.00 0.00 0.00 0.00P ORGANIZATION 3 0.33 0.33 0.67 0.67POINT 2 0.00 0.00 0.00 0.00P PARTY 37 0.30 0.41 0.43 0.57DRUG 2 0.00 0.00 0.00 0.00GOVERNMENT 37 0.50 0.54 0.53 0.57SPACESHIP 4 0.88 1.00 0.88 1.00N PRODUCT 41 0.25 0.37 0.37 0.56ACTION 18 0.22 0.22 0.30 0.44PRODUCT 58 0.24 0.34 0.44 0.69MOVIE 6 0.50 0.50 0.56 0.67WAR 2 0.75 1.00 0.75 1.00MUSIC 8 0.19 0.25 0.56 0.62SHIP 7 0.26 0.43 0.40 0.57WATER FORM 3 0.50 0.67 0.50 0.67N ORGANIZATION 20 0.14 0.25 0.28 0.55CONFERENCE 17 0.14 0.24 0.46 0.65ORGANIZATION 23 0.08 0.13 0.20 0.30SEA 1 1.00 1.00 1.00 1.00SPEED 1 0.00 0.00 1.00 1.00PICTURE 1 0.00 0.00 0.00 0.00VOLUME 5 0.00 0.00 0.18 0.60SCHOOL 21 0.10 0.10 0.33 0.43GAMES 8 0.28 0.38 0.34 0.50ACADEMIC 5 0.20 0.20 0.37 0.60POSITION TITLE 39 0.20 0.28 0.30 0.44PERCENT 47 0.35 0.43 0.43 0.55REGION 22 0.17 0.23 0.46 0.64COMPANY 77 0.45 0.55 0.57 0.70GEOLOGICAL 3 0.42 0.67 0.42 0.67PERIODX 1 1.00 1.00 1.00 1.00LOCATION 2 0.00 0.00 0.50 0.50RULE 35 0.30 0.43 0.49 0.69EXTENT 22 0.04 0.09 0.13 0.18MONUMENT 2 0.00 0.00 0.25 0.50CURRENCY 1 0.00 0.00 0.00 0.00SPORTS 9 0.17 0.22 0.40 0.67STATION 3 0.50 0.67 0.50 0.67INSTITUTE 26 0.38 0.46 0.53 0.69RAILROAD 1 0.00 0.00 0.25 1.00MONEY 110 0.33 0.40 0.48 0.63PHONE 1 0.00 0.00 0.00 0.00AIRPORT 4 0.38 0.50 0.44 0.75PROVINCE 36 0.30 0.33 0.45 0.50MILITARY 4 0.00 0.00 0.25 0.25N ANIMAL 3 0.11 0.33 0.22 0.67ART 4 0.25 0.50 0.25 0.50ANIMAL 10 0.26 0.50 0.31 0.60MONTH PERIOD 6 0.06 0.17 0.06 0.17ROAD 1 0.00 0.00 0.50 1.00LANGUAGE 3 1.00 1.00 1.00 1.00DATE PERIOD 9 0.11 0.11 0.33 0.33COUNTX 10 0.33 0.40 0.38 0.60DATE 130 0.24 0.32 0.41 0.58AMUSEMENT 2 0.00 0.00 0.00 0.00YEAR PERIOD 34 0.22 0.29 0.38 0.59PARK 1 0.00 0.00 0.00 0.00AGE 22 0.34 0.45 0.44 0.59SHOW 3 0.78 1.00 1.11 1.33MULTIPLICATION 9 0.39 0.44 0.56 0.67PUBLIC INST 19 0.18 0.26 0.34 0.53CRIME 4 0.75 0.75 0.75 0.75PORT 3 0.17 0.33 0.33 0.67AIRCRAFT 2 0.00 0.00 0.25 0.50N COUNTRY 8 0.28 0.38 0.32 0.50MUSEUM 3 0.33 0.33 0.33 0.33NATIONALITY 4 0.50 0.50 1.00 1.00DISEASE 18 0.29 0.50 0.43 0.72COUNTRY 84 0.45 0.60 0.51 0.67FREQUENCY 13 0.18 0.31 0.19 0.38OFFENSE 9 0.23 0.44 0.23 0.44WEAPON 1 1.00 1.00 1.00 1.00CITY 72 0.41 0.50 0.53 0.65MINERAL 18 0.16 0.22 0.25 0.39N FACILITY 4 0.25 0.25 0.38 0.50METHOD 29 0.39 0.48 0.48 0.62FACILITY 11 0.20 0.36 0.25 0.55ETHNIC 3 0.42 0.67 0.75 1.00TIMEX 3 0.00 0.00 0.00 0.00NAME 5 0.20 0.20 0.40 0.40TIME TOP 2 0.00 0.00 0.50 0.50SPACE 4 0.50 0.50 0.50 0.50TIME PERIOD 8 0.12 0.12 0.48 0.75THEORY 1 0.00 0.00 0.00 0.00TIME 13 0.22 0.31 0.29 0.38LANDFORM 5 0.13 0.40 0.13 0.40ERA 3 0.00 0.00 0.33 0.33TRAIN 2 0.17 0.50 0.17 0.50PHENOMENA 5 0.50 0.60 0.60 0.802000 0.28 0.36 0.43 0.58DISASTER 4 0.50 0.75 0.50 0.75OBJECT 5 0.47 0.60 0.47 0.60CAR 1 1.00 1.00 1.00 1.00RELIGION 5 0.30 0.40 0.30 0.40WEEK PERIOD 4 0.05 0.25 0.55 0.75WEIGHT 12 0.21 0.25 0.31 0.42PRINTING 6 0.17 0.17 0.38 0.50
Bạn đang xem 635- - BÁO CÁO KHOA HỌC QUESTION ANSWERING AS QUESTION BIASED TERM EXTRACTION A NEW APPROACH TOWARD MULTILINGUAL QA DOC