Page 185 - My FlipBook
P. 185
Research Fellow 蘇克毅 研
究
Keh-Yih Su 人
員
Ph.D., Electrical Engineering, University of Washington, United States Faculty
T +86-2-2788-3799 ext. 1801 E kysu@iis.sinica.edu.tw
F +886-2-2782-4814 W www.iis.sinica.edu.tw/pages/kysu
・ Research Fellow, Institute of Information Science, Academia Sinica (2014/6-present)
・ President/CEO, Behavior Design Corporation (1998-2014)
・ Associate Professor, Professor, Dept. of Electrical Engineering, National Tsing Hua
University (1984-1998)
・ Chair, Asia-Paci c Chapter of the Association for Computational Linguistics (2021-2022)
・ Chair-elect, Asia-Paci c Chapter of the Association for Computational Linguistics (2018-
2020)
・ President, Asian Federation of Natural Language Processing (2011-2012)
・ Executive Member, Association for Computational Linguistics (2005-2007)
・ Ph.D., Electrical Engineering, University of Washington (1984)
Research Description
My research interests include Natural Language Understanding (NLU), Machine Reading, Question and Answering (Q&A), and Chinese
Language Processing. I have worked on statistical modeling and machine learning for natural language processing since 1986 (mainly
adopting principled approaches to solving NLP problems with domain knowledge, in contrast to the simple data-driven techniques).
My past experience mainly lies in: (1) Building a statistical semantic machine translation system, which conducts translation in semantic level
and automatically obtains associated parameters via semi-supervised learning. (2) Proposing a joint model to integrate Translation Memory
into a phrase-based machine translation system during decoding, which has achieved signi cant improvement. (3) Building both statistics-
based and DNN-based Math Word Problem Solvers, which tried to gure out the associated physical meaning of those quantities described
in the problem text for providing interpretable reasoning steps. They act as our test cases for studying the problems of Machine Reading and
Question Answering.
My current main projects are: (1) Building a platform to explore new knowledge from multiple documents, which is supported by the
Thematic Program of Academia Sinica, and (2) Designing/Constructing a Conversational Open Domain Document-based Natural Speech
Q&A (COSQA) system, which is supported by IIS.
The project of exploring new knowledge from multiple documents aims to refine the noisy information extracted from traditional
information extraction (IE), and to explore new knowledge across domains. On the other hand, the project of constructing COSQA system not
only acts as an ideal testbed for conducting NLU, but also has many real applications such as customer service, medical consultant system,
information-seeker, etc. This project will handle following di cult issues at the same time: free-text, multi-document rational, common-sense
reasoning, multi-hop inference, pragmatic-reasoning, and natural-speech. Conducting this project will force us to face the real challenge in
AI.
Publications
1. Qianlong Du, Chengqing Zong, and Keh-Yih Su, "Conducting 6. Chao-Chun Liang, Yu-Shiang Wong, Yi-Chung Lin and Keh-Yih Brochure 2020
Natural Language Inference with Word-Pair-Dependency and Su, "A Meaning-based Statistical English Math Word Problem
Local Context," ACM Transactions on Asian and Low-Resource Solver," Proceedings of NAACL-HLT 2018, New Orleans, LA,
Language Information Processing, volume 19, number 3, pages U.S.A, June 2018.
47:1-47:23, February 2020."
7. Hsin-Wei Yu, Chia-Hung Huang, and Keh-Yih Su, "Answering
2. Chao-Chun Liang and Keh-Yih Su, "ASNLU at the NTCIR-14 Yes/No Questions with a Decomposable Attention Model,"
FinNum Task: Incorporating Knowledge into DNN for Financial Proceedings of TAAI 2017, December 2017.
Numeral Classification," Proceedings of the 14th NTCIR
Conference on Evaluation of Information Access Technologies, 8. Chao-Chun Liang ,Yu-Shiang Wong ,Yi-Chung Lin and Keh-
Tokyo, Japan, June 2019. Yih Su, "A Goal-Oriented Meaning-based Statistical Multi-Step
Math Word Problem Solver with Understanding, Reasoning and
3. Yang Liu, Kun Wang, Chengqing Zong, and Keh-Yih Su, "A Explanation," Proceedings of IJCAI 2017, Melbourne, Australia,
Unified Framework and Models for Integrating Translation August 2017, System Demonstration.
Memory into Phrase-based Statistical Machine Translation,"
Journal of Computer Speech and Language, volume 54, pages 9. Qianlong Du, Chengqing Zong and Keh-Yih Su, "Integrating
176-206, March 2019. Structural Context with Local Context for Disambiguating Word
Senses," Proceedings of NLPCC-ICCPOL 2016, Kunming,
4. Meng-Tse Wu, Yi-Chung Lin, and Keh-Yih Su, "Supporting China, December 2016.
Evidence Retrieval for Answering Yes/No Questions,"
International Journal of Computational Linguistics and Chinese 10. Chao-Chun Liang, Shih-Hong Tsai, Ting-Yun Chang, Yi-Chung
Language Processing, volume 23, number 2, pages 47-66, Lin and Keh-Yih Su, "A Meaning-based English Math Word
December 2018, Its brief version also won the best paper award Problem Solver with Understanding, Reasoning and Explanation,"
in Rocling-2018. Proceedings of COLING 2016, Osaka, Japan, December 2016,
System Demonstration.
5. Qianlong Du, Chengqing Zong, and Keh-Yih Su, "Adopting the
Word-Pair-Dependency-Triplets with Individual Comparison for 183
Natural Language Inference," COLING 2018, Santa Fe, New
Mexico, USA, August 2018.
究
Keh-Yih Su 人
員
Ph.D., Electrical Engineering, University of Washington, United States Faculty
T +86-2-2788-3799 ext. 1801 E kysu@iis.sinica.edu.tw
F +886-2-2782-4814 W www.iis.sinica.edu.tw/pages/kysu
・ Research Fellow, Institute of Information Science, Academia Sinica (2014/6-present)
・ President/CEO, Behavior Design Corporation (1998-2014)
・ Associate Professor, Professor, Dept. of Electrical Engineering, National Tsing Hua
University (1984-1998)
・ Chair, Asia-Paci c Chapter of the Association for Computational Linguistics (2021-2022)
・ Chair-elect, Asia-Paci c Chapter of the Association for Computational Linguistics (2018-
2020)
・ President, Asian Federation of Natural Language Processing (2011-2012)
・ Executive Member, Association for Computational Linguistics (2005-2007)
・ Ph.D., Electrical Engineering, University of Washington (1984)
Research Description
My research interests include Natural Language Understanding (NLU), Machine Reading, Question and Answering (Q&A), and Chinese
Language Processing. I have worked on statistical modeling and machine learning for natural language processing since 1986 (mainly
adopting principled approaches to solving NLP problems with domain knowledge, in contrast to the simple data-driven techniques).
My past experience mainly lies in: (1) Building a statistical semantic machine translation system, which conducts translation in semantic level
and automatically obtains associated parameters via semi-supervised learning. (2) Proposing a joint model to integrate Translation Memory
into a phrase-based machine translation system during decoding, which has achieved signi cant improvement. (3) Building both statistics-
based and DNN-based Math Word Problem Solvers, which tried to gure out the associated physical meaning of those quantities described
in the problem text for providing interpretable reasoning steps. They act as our test cases for studying the problems of Machine Reading and
Question Answering.
My current main projects are: (1) Building a platform to explore new knowledge from multiple documents, which is supported by the
Thematic Program of Academia Sinica, and (2) Designing/Constructing a Conversational Open Domain Document-based Natural Speech
Q&A (COSQA) system, which is supported by IIS.
The project of exploring new knowledge from multiple documents aims to refine the noisy information extracted from traditional
information extraction (IE), and to explore new knowledge across domains. On the other hand, the project of constructing COSQA system not
only acts as an ideal testbed for conducting NLU, but also has many real applications such as customer service, medical consultant system,
information-seeker, etc. This project will handle following di cult issues at the same time: free-text, multi-document rational, common-sense
reasoning, multi-hop inference, pragmatic-reasoning, and natural-speech. Conducting this project will force us to face the real challenge in
AI.
Publications
1. Qianlong Du, Chengqing Zong, and Keh-Yih Su, "Conducting 6. Chao-Chun Liang, Yu-Shiang Wong, Yi-Chung Lin and Keh-Yih Brochure 2020
Natural Language Inference with Word-Pair-Dependency and Su, "A Meaning-based Statistical English Math Word Problem
Local Context," ACM Transactions on Asian and Low-Resource Solver," Proceedings of NAACL-HLT 2018, New Orleans, LA,
Language Information Processing, volume 19, number 3, pages U.S.A, June 2018.
47:1-47:23, February 2020."
7. Hsin-Wei Yu, Chia-Hung Huang, and Keh-Yih Su, "Answering
2. Chao-Chun Liang and Keh-Yih Su, "ASNLU at the NTCIR-14 Yes/No Questions with a Decomposable Attention Model,"
FinNum Task: Incorporating Knowledge into DNN for Financial Proceedings of TAAI 2017, December 2017.
Numeral Classification," Proceedings of the 14th NTCIR
Conference on Evaluation of Information Access Technologies, 8. Chao-Chun Liang ,Yu-Shiang Wong ,Yi-Chung Lin and Keh-
Tokyo, Japan, June 2019. Yih Su, "A Goal-Oriented Meaning-based Statistical Multi-Step
Math Word Problem Solver with Understanding, Reasoning and
3. Yang Liu, Kun Wang, Chengqing Zong, and Keh-Yih Su, "A Explanation," Proceedings of IJCAI 2017, Melbourne, Australia,
Unified Framework and Models for Integrating Translation August 2017, System Demonstration.
Memory into Phrase-based Statistical Machine Translation,"
Journal of Computer Speech and Language, volume 54, pages 9. Qianlong Du, Chengqing Zong and Keh-Yih Su, "Integrating
176-206, March 2019. Structural Context with Local Context for Disambiguating Word
Senses," Proceedings of NLPCC-ICCPOL 2016, Kunming,
4. Meng-Tse Wu, Yi-Chung Lin, and Keh-Yih Su, "Supporting China, December 2016.
Evidence Retrieval for Answering Yes/No Questions,"
International Journal of Computational Linguistics and Chinese 10. Chao-Chun Liang, Shih-Hong Tsai, Ting-Yun Chang, Yi-Chung
Language Processing, volume 23, number 2, pages 47-66, Lin and Keh-Yih Su, "A Meaning-based English Math Word
December 2018, Its brief version also won the best paper award Problem Solver with Understanding, Reasoning and Explanation,"
in Rocling-2018. Proceedings of COLING 2016, Osaka, Japan, December 2016,
System Demonstration.
5. Qianlong Du, Chengqing Zong, and Keh-Yih Su, "Adopting the
Word-Pair-Dependency-Triplets with Individual Comparison for 183
Natural Language Inference," COLING 2018, Santa Fe, New
Mexico, USA, August 2018.