The First KYOTO Workshop
Environmental Knowledge Transition and Exchange
February 2-3, 2009, Amsterdam, the Netherlands
Home
Organizers
Invited speakers
Participants
Program
Registration
Venue

Invited Speakers

Talk topic: The role of standards in our community - an ISO perspective
Speaker: Laurent Romary
Talk abstract (Slides)

My talk will focus on providing the methodological background for standardisation activities as carried out within ISO and more particularly in the committee dedicated to language resources (ISO/TC 37/SC 4).

I will present the basic standards related to character and language encoding and move on to demonstrate how recent standard developments related to the representation of linguistic content (such as TMF and LMF on terminologies and lexica respectively) may provide a coherent background for more exploratory works (e.g. on dialogue acts).

The speaker

Director of Scientific Information at CNRS.

Born in 1964, Laurent Romary got his PhD in computational linguistics in 1989 and his Habilitation thesis in 1999.

He is currently senior researcher at INRIA (Institut National de Recherche en Informatique et Automatique).

He leads the Langue et Dialogue team (http://www.loria.fr/equipes/led/) at Loria laboratory and has conducted various research activities on man-machine dialogue, multilingual document management and linguistic engineering.

His has participated in numerous national and international project related to the representation and dissemination of language resources and on man-machine interaction, and in particular coordinated the MLIS/DHYDRO and IST/MIAMM projects.

He has been the editor of ISO 16642 (TMF, Terminological Markup Framework) under TC37/SC3, and is the chairman of ISO committee (TC37/SC4) on Language Resource Management.

Reference Web site: http://www.loria.fr/~romary/index.php
Talk topic: Information access through textual inferences: the experience of the Qall-Me project
Speaker: Bernardo Magnini
Talk abstract (Slides)

Accessing information contained in large repositories, both structured and unstructured, is one of the most challenging task in the digital era.

Systems need to deal with multiple and heterogeneous information sources, need to cope with different languages, need to be aware of the context in which questions are posed and need effective and natural method for presenting information to users.

In this talk I address information access through natural language under three related perspectives: (i) the role of the ontology for representing domain knowledge in order to allow multilingual compatibility; (ii) the design of entailment engines for textual inferences in order to address language variability; (iii) the use of machine learning in order to ensure easy and fast prototyping to port the system from one domain to another.

Such solutions have been experimented in the in the context of the Qall-Me project.

The speaker

Bernardo Magnini joint the Natural Language Processing group at Irst in 1988, under the lead of Oliviero Stock. He collaborated to the Alfresco project, for which he developed the lexical-semantic analysis module. In 1994 he was visiting researcher at CRL at New Mexico State University. In that period he started a project for the development of an Italian version of the English WordNet. In 2000 he collaborated to develop WordNet Domains, an enrichment of WordNet with domain information.

He collaborated to the organization of Senseval, an evaluation campaign on word sense disambiguation, and in 2003 lunched QA@CLEF, an evaluation campaign for multilingual question answering systems. Since 2005 he is involved in the organization of the Recognizing Textual Entailment challenge and in 2007 he chaired Evalita-2007, the first evaluation campaign on NLP tools for Italian. In 2006 he served as local organizer co-chair of EACL, the 11th Conference of the European Chapter of the Association for Computational Linguistics, in Trento.

He has been involved in several research projects, including TAMIC-P (NLI to database), MEANING (word sense disambiguation), ONTOTEXT (ontology population from texts) and QALL-ME (question answering for mobile devices).

He is member of the Editorial Board of the Journal of Natural Language Engineering, of the John Benjamins Publishers' book series on Natural Language Processing and of the Advisory Board of the Text Analysis Conference run by NIST.

He is contract professor at the University of Trento (since 2002), at the Free University of Bolzano (since 2005) and he is Faculty member of the Master in Human Language Technology and Interface at University of Trento.

Since 2008 he co-heads the Human Language Technology research unit at FBK-irst.

Reference Web site: http://hlt.fbk.eu/people/magnini
Talk topic: The day to day information needs of a statutory nature conservation agency
Speaker: Bernard Fleming
Talk abstract (Slides part 1 - Slides part 2)

In the UK, national nature conservation agencies fulfil the role of statutory, but independent, advisers to government. In the context of development control, they are statutory consultees where sites or species of national or international importance are likely to be impacted upon (by proposed development). Taking implementation of the Habitats Directive as an example, they are responsible for advising decision-making bodies (eg local and regional government, etc) and therefore have to consider the impacts of both the development itself and in-combination with other plans or projects.

As an example, the Humber Estuary in England, extends over 37,000ha and supports internationally important populations of breeding, passage and wintering birds (granting Special Protection Area status), along with 13 littoral and subtidal habitats and species (granting candidate Special Area for Conservation status). It also forms the largest port complex in the country and much new development is proposed in terms of new port construction, petro-chemical developments and both on- and off-shore windfarms. The potential for conflict is clear but examples of best practice have seen the construction of major infrastructure developments without harm to the environment.

Central to the decision-making process is the assessment of impacts resulting from a variety of sources and pathways. In-house expertise is not able to master all these disciplines and so the task of obtaining and analysing relevant data and information is key to efficient and effective working. Difficulties in processing data and information arise when assessing complex proposals which can involve matters as diverse as noise, air pollution, coastal dynamics and water quality.

Typically, staff resort to internet searches via traditional search engines, discussion with in-house experts (if there are any) and (importantly) colleagues, library searches and the use of specialist NGOs. This does not always provide the compelling evidence required by the legislation. This can result in delay and conflict and threatens both commercial aspirations and the biodiversity of the site.

A better system is required that provides up to date, relevant information. Ideally it should be web-based but should not be seen as a replacement for talking to people.

The speaker

Mr Bernard Fleming is an experienced practitioner in the field of nature conservation with more than 20 years working in wildlife and environmental legislation and its implementation, site management, policy, stakeholder communication and issues relating to large sites (e.g. estuaries, etc).

He is both an independent consultant (with work experience in eastern Europe) and a full-time employee of Natural England.

In all aspects of his work he requires access to easily digested information and data about species, habitats and a range of related issues.

Reference Web site: http://www.naturalengland.org.uk/
Email address : Bernard.Fleming@naturalengland.org.uk
Talk topic: Semantic Wikis and the Web of Data
Speaker: Jens Lehmann
Talk abstract (Slides)

To turn the vision of the Semantic Web into reality, a wide deployment of semantic technologies is of paramount importance. Within the Web of Data initiative, best practices and tools are developed to transform the Web of documents into a Web of queryable and interlinkable knowledge bases.

A main component of this Data Web are Semantic Wikis, which extend Wiki technology and concepts with collaborative semantic knowledge engineering facilities. Besides extensions of text-oriented Wiki systems, such as Semantic MediaWiki and Ikewiki, we will present the more data-oriented OntoWiki platform.

A further central component of the Data Web is the DBpedia knowledge base, which contains facts extracted from Wikipedia and currently evolves into a crystallization point for interlinked open data within the Semantic Web.

The speaker

Jens Lehmann is a researcher at the Agile Knowledge Engineering and Semantic Web (AKSW) Group, Chair of Business Information Systems, Institute of Computer Science, Faculty of Mathematics and Computer Science, University of Leipzig.

Jens is one of the co-founders of the DBpedia project, founder of the DL-Learner machine learning toolkit and an intimate expert in the Data Web and Semantic Wiki realms.

Reference Web site: http://jens-lehmann.org/