INVITED SPEAKERS (in alphabetic order)
			Chu-Ren HUANG, Academia Sinica Kiyong LEE, Korea University Yuji MATSUMOTO, Nara Institute of Science and Technology

Invited Talk 1 Day 1 : December 8th Wednesday 9:45-10:45

Processisng and Representing Temporally Sequential Events, Kiyong LEE, Korea University

• Abstract:
  In a discourse, events are narrated one by one linearly. But the temporal sequence of events does not always match the linear sequence of narration. One of such cases involves the mixed occurrences of durative and punctual events, as illustrated by "Mia took aspirin and slept, for she was ill. She then fell off a cliff in her dream." In this narration, five events are reported: three are durative events of sleeping, being ill and having a dream and two are punctual events of taking aspirin and falling off a cliff.

  This presentation aims at establishing some systematic way of processing such events and representing them in a reasonably understandable temporal sequence. For this, events are analyzed in terms of an interval semantics that allows them to be anchored to appropriate temporal intervals and be ordered in an appropriate temporal sequence. In order to provide a simple syntactic basis, the presentation attempts to develop a small computational program that derive representations in feature structure by analyzing a small fragment of Korean.

• Dr. Kiyong Lee's Biographic Sketch:
  Our first speaker Dr. Kiyong Lee is professor emeritus of Korea University. Recently he was awarded for his academic achievements in Humanities and Social Sciences by the Korean National Academy of Sceinces. As a Montague grammarian, he has been specializing in formal and computational semantics. He is also actively engaged in international activities of ISO, the International Standardization Organization, as Convenor and Project Leader.

  Professor Kiyong Lee was formerly president of the Linguistic Society of Korea and also president of the Korean Society for Cognitive Science. With Fulbright and German DAAD, de-a-de, scholarships, he visited Stanford University and Erlangen University in Germany. His educational background includes a Ph.D. in Linguistics from the University of Texas, an M.A. in English from Chonnam National University and a B.A. from Saint Louis University in St. Louis with an emphasis on classical study, mathematics and philosophy.
  

Invited Talk 2 Day 2 : December 9th Thursday 9:30-10:30

Machine Learning based NLP: Experiences and Supporting Tools, Yuji MATSUMOTO, Nara Institute of Science and Technology

• Abstract:
  Coprus-based approaches to natural language analysis that utilize recent sophisticated machine learning algorithms have now become to achieve very good performance. In this talk I will overview and categorize machine learning based natural language processing tasks and our experiences of using machine learning to various tasks such as segmentation, POS tagging, phrase and NE chunking, and syntactic parsing. I then discuss pros and cons of machine learning approaches and future issues. Finally, I will introduce an ongoing project of annotated corpus maintenance tools for developing consistent data for corpus and machine learning based NLP research.
  
  
• Dr. Yuji MATSUMOTO's Biographic Sketch
  Yuji Matsumoto is currently a Professor of Graduate School of Information Science, Nara Institute of Science and Technology. He received his M.S. and Ph.D. degrees in information science from Kyoto University in 1979 and in 1989. He joined Machine Inference Section of Electrotechnical Laboratory in 1979. He has then experienced an academic visitor at Imperial College of Science and Technology, a deputy chief of First Laboratory at ICOT, and an associate professor at Kyoto University. His main research interests are natural language understanding and linguistic knowledge acquisition.

Invited Talk 3 Day 3 : December 10th Friday 13:30-14:30

Text-based Construction and Comparison of Domain Ontology: A Study Based on Classical Poetry, Chu-Ren HUANG, Academia Sinica

• Abstract:
  The fact that people from different backgrounds may have knowledge structures unlike ours is a crucial issue to be addressed in knowledge engineering. In order to become sharable and reusable knowledge, all extracted information must first be correctly situated in a knowledge structure. In addition, the situated information must be allowed to transfer from knowledge structure to knowledge structure without losing its meaningful content. This is the vision behind the Suggested Upper Merged Ontology (SUMO, http://www.ontologyportal.org) proposed by an IEEE working group. A shared upper ontology will both anchor the structured transfer of knowledge as well as set a standard for the construction of a middle and lower level ontology for each domain. This vision also has promising applications in the Semantic Web.

  The most salient factors dictating variations in knowledge structures are time, space, and domain. These factors are compounded with language, which is both the product and conduit of the conceptual structure of its speakers. In order to demonstrate the felicity of the shared upper ontology approach, we need to show that it can successfully applied to comparative studies of different knowledge structures regardless of their ontological variations. We apply the shared ontology proposal to the interpretation historical texts by adopting the Shakespearean-garden approach towards construction of historical ontology. The Shakespearean-garden refers to the common practice in western museums of collecting in a garden all the plants referred to in the Shakespearean texts. This garden then illustrates the flora of the Shakespearean England and will give us the context to interpret his work. In our approach, we first segment and collect all words from the set of target texts. Once the comprehensive lexicon of that period is collected, a lexical interface based on Sinica BOW (http://bow.sinica.edu.tw) can be applied. It links each word to a conceptual location on the SUMO ontology, or a synset in WordNet. Since the lexicon from the text represents instantiated concepts, we use the linked conceptual nodes to construct an ontology for that text. The constructed ontology allows us to both interpret the conceptual structure of that text as well compare its knowledge with our contemporary knowledge.

  The two text collections studied are the 300 Tang poems (唐詩三百首) and the collection of poems by Su Shi (蘇軾詩). The ontologies constructed from both text collections allow us to compare and study the knowledge structure of two different historical periods and gain perspective understanding of the different culture and time.

• Dr. Chu-Ren Huang's Biographic Sketch and Research Outline:
  Dr. Chu-Ren Huang is a research fellow and associate director of the Institute of Linguistics, Academia Sinica. He is a member of the International Committee on Computational Linguistics (ICCL). the vice president of the Linguistic Society of Taiwan (LST), founding council member of the Open Language Archives Community (OLAC), and founding co-chair of the Asian Language Committee. He currently serves on the editorial boards of Journal of Chinese Linguistics, Language and Linguistics, and Computational Linguistics and Chinese Language Processing.

  He received his Ph.D. in linguistics from Cornell University in January 1987 and started his research position at Academia Sinica in the same year. He has held adjunct or visiting positions in various international academic institutions, including City University of Hong Kong, CNRS, CSLI, Peking University, University of Pennsylvania, and UCSB. He served as president and secretary of ROCLING/ACLCLP and was a founding executive board member of ROCLING, LST, and the International Association of Computational Linguistics (IACL).

  Over the past 17 years, he has played an active role to promote research on Chinese computational and corpus linguistics. He has directed or co-directed the successful construction of the following Chinese language resources: CKIP lexicon, Sinica Corpus, Classical Chinese Corpora, Sinica Treebank, and Academia Sinica Bilingual Ontological Wordnet. His linguistic research focus shifted from earlier work on GPSG and LFG to recent emphasis on lexical semantics, which led to the development of the MARVS theory. Both lines of research lead to his current work on Chinese WordNet as well as merging CWN with English WordNet and Upper Ontology. His research vision is to develop lexical knowledge framework and infrastructure that will both anchor web-based knowledge engineering as well as shed light on the internal representation of human knowledge.

• List of Publications:
  http://corpus.ling.sinica.edu.tw/member/churen/
  
  
• Research Accomplishments

1. Academia Sinica Bilingual Ontological Wordnet, Sinica BOW
   http://BOW.sinica.edu.tw
2. Sinica Corpus
   http://www.sinica.edu.tw/SinicaCorpus/
3. SouWenJieZi: A Linguistic KnowledgeNet
   http://www.sinica.edu.tw/~words/
4. Adventures in Wen-Land:
   http://http://www.sinica.edu.tw/wen/
5. Early Mandarin Corpus
   http://www.sinica.edu.tw/Early_Mandarin/
6. Language Archives and Linguistic Anchoring
   http://corpus.ling.sinica.edu.tw/project/LanguageArchive/