COLING-2008 Workshop:
KRAQ08: Knowledge and Reasoning for Answering Questions

Manchester, 23 August, 2008

Held in conjunction with COLING-2008 the 22nd International Conference on Computational Linguistics
18-22 August, 2008

This workshop is the fourth of a successful series, it follows :
KRAQ05 (IJCAI 2005 Edinburgh, UK).
KRAQ06 (EACL 2006, Trento, Italy).
KRAQ07 (IJCAI 2007, Hyderabad, India).


9.30-10.00 Semantic Chunk Annotation for complex questions using Conditional Random Field
Shixi Fan, Yaoyun Zhang, Wing W. Y. Ng, Xuan Wang and Xiaolong Wang

10.00-10.30 Context Inducing Nouns
Charlotte Price, Valeria de Paiva and Tracy Holloway King


11.00 - 12.30 Panel: Multimedia question answering
Bonnie Webber, University of Edinburgh, UK
Dragomir Radev, University of Michigan, USA
Silvia Quarteroni, University of Trento, Italy
moderator: Marie-Francine Moens, Katholieke Universiteit Leuven, Belgium

See outline below

Lunch break

2.00-3.00 Invited talk: Bonnie Webber : Complex Factoid Questions
Abstract: We take "complex questions" to be ones that require a range of evidence to answer. Opinion questions and what in TREC have been called "Other" questions obviously belong to this class. But, as some people have already noticed, even factoid questions may require more than a single piece of evidence to answer. In this talk I'll focus on complex factoid questions and various strategies that have been used for answering them.

3.00 - 3.30 Know-Why Extraction from Textual Data
Chaveevan Pechsiri, Phunthara Sroison and Uraiwan Janviriyasopa


3.30- 4.00 Context Modeling for IQA: The Role of Tasks and Entities
Raffaella Bernardi and Manuel Kirschner

4.00 - 4.30 Personalized, Interactive Question Answering on the Web,
Silvia Quarteroni

4.30 - 5.00 Creating and Querying a Domain dependent Know-How Knowledge Base of Advices and Warnings
Lionel Fontan and Patrick Saint-Dizier

Panel Outline:
We explore different current topics of multimedia QA among which are: Searching multimedia data (e.g., images, video) with natural language expressions, and inference over content recognized in the language expressions and in the other media; the links between speech and QA: this relates to both voice-activated QA and QA performed on spoken data; applications of graph-based approaches to QA.

Call for papers

Confronted with large information databases there is an urgent need for retrieving concise answers to information needs, possibly fused from different sources. Question answering (QA) systems that intelligently and cooperatively respond to the user’s questions posed in natural language aim at realizing this goal. Reasoning is here regarded as an important technology. Although quite advanced reasoning models are around for several decades in closed domain QA, it is only recently that open domain QA has recognized the potential of automated reasoning. Performances in the recent TREC-QA tracks show that inferencing substantially improves the response relevance and accuracy. There is still a long way to go before we can consider our document repositories (such as World Wide Web) as a huge knowledge base with question answering technology acting as the ultimate expert system. However, recent foundational, methodological and technological developments in knowledge representation (e.g. ontologies e.g. in the frame of the Semantic Web, knowledge bases incorporating various forms of incompleteness or uncertainty), in advanced reasoning forms (e.g. data fusion-integration, argumentation, decision theory, fuzzy logic, incomplete knowledge bases, probabilistic inferencing, etc.), in advanced language processing resources and techniques (for question processing, generating responses and textual inferencing), and recent progress in human language technology and formal pragmatics (user models, intentions, etc.) help us progressing towards this goal.

There have been several QA workshops in the past. Some of them were oriented towards the processing of massive sources of data like the Web. A few concentrated on user profiling, on a specific application (medical, opinion) or on the processing of the questions. The KRAQ series of workshops alternately organized at AI and NLP conferences address more fundamental problems of QA. More specifically its 2008 edition will investigate models where answers might be retrieved and fused from different types of data repositories (text, multimedia, web pages etc.) containing either open or closed domain content and the answers might be accompanied by explanations and arguments on how the system came to an answer.

We invite papers on any research topic related to question answering and reasoning. In this call we especially focus on:
  1. Reasoning aspects: information fusion-integration, summarization and intensional answers, detecting and resolving query failure, reasoning under uncertainty or with incomplete knowledge, probabilistic inference, models for explanation production and argumentation, levels of knowledge involved (e.g. ontologies, domain knowledge).
  2. Textual inference: question decomposition into candidate inference components, information alignment, entity linking within and across documents, entailment and paraphrasing.
  3. Innovative applications: multimedia question answering, where you question a more or less formal representation of the media objects, spoken question answering (increasing uncertainty caused by the speech recognition), question answering of semi-structured documents such as Wikipedia and legislation, XML question answering.

We also welcome papers on more traditional QA topics such as:

The goal of the KRAQ 2008 workshop is to enhance cooperation between participants with an AI background and the NLP and question-answering communities. Contributors must be opened to interactions with the different workshop areas. The programme committee will take care of having a balanced number of papers from the different areas concerned.

Submission format:

We welcome short papers (max. 4 pages) describing projects or ongoing research and long papers (max. 8 pages), that include established results. Papers can be uploaded in pdf format at The format to use for papers and abstracts is the same as for Coling ( All papers are blinded for review. The title page includes the following information:
Topic(s) of the above list, as appropriate
Abstract (short summary up to 5 lines).


May 5th: paper submissions
June 6th: acceptance/rejection notification
July 1st: final papers due, camera-ready manuscript sent for printing by organizers.

All accepted papers (long and short) will be published in the workshop proceedings. We are currently negotiating a special journal issue or book publication for the best long papers.

Patrick Saint-Dizier CNRS
IRIT Toulouse

Marie-Francine Moens
Department of Computer Science
Katholieke Universiteit Leuven

Programme Committee

Leila Amgoud, IRIT CNRS, France,
Sivaji Badhyopadyay, Jadavpur University, Kolkata, India
Johan Bos, Università di Roma, Italy
Gosse Bouma, University of Groningen, The Netherlands
Sanda Harabagiu, University of Dallas, USA
Jerry Hobbs, University of Southern California/ISI, USA
Kentaro Inui, NAIST, Nara, Japan
Asanee Kawtrakul, Kasetstart University, Bangkok, Thailand
Dietrich Klakow, Universität des Saarlandes, Germany
Jochen L. Leidner, Jochen L. Leidner, Thomson Corporation , USA
Anne-Laure Ligozat, LIMSI, France
Marie-Francine Moens, K.U.Leuven, Belgium (co-chair)
Matteo Negri, ITC-irst, Italy
Silvia Quarteroni, University of Trento, Italy
Maarten de Rijke, University of Amsterdam, The Netherlands
Patrick Saint-Dizier, IRIT-CNRS, France (co-chair)