Information visualisation workshop

Teachware on demand - the role of semantic annotations for individualised courses (Peter Fankhauser, GMD-IPSI)

e-mail: [e-mail]
URL: http://www.darmstadt.gmd.de/~fankhaus

Semantic user and content modeling for an adaptive distance learning environment (Luca Botturi, Universita della Svizzera Italiana)

The goal of the talk is to outline research issues in information modeling starting from a short analysis of distance learning developments, potentialities and needs in different contexts.

It starts identifying the major distance learning options such as self-learning, collaboration, virtual simulation, etc. from both cognitive and social perspectives. These options are then embodied in different application contexts (corporate training, learning institutions, etc.), trying to identify some focus points for possible interventions in learning environments.

The claim is that we need now flexible distance learning development environments (as opposite to course-products or teaching materials) that exploit semantic technologies such as adaptive hypermedia. Moreover, these development environments should not be edge technologies, but simple and user-friendly, providing defined functions as answers to particular needs.

In order to do that, two kinds of information should be formalized: the user profile and the content to be taught. Some models coming from our research are presented.

e-mail: [e-mail]
URL: http://www.lu.unisi.ch

Open Learning Repository: a project idea in the field of computer science and cognitive science (Heidrun Allert, University of Hannover)

e-mail: [e-mail]
URL: http://www.kbs.uni-hannover.de, http://www.etl.uni-hannover.de

PhotoRDF and rdfpic,an application of RDF to holiday photos (Bert Bos, W3C/INRIA)

e-mail: [e-mail]
URL: http://www.w3.org/people/bos

A tool for pictorial contents retrieval (Zdenek Mikovec, Czech Technical University)

As the complexity of pictures increases and the need for more detailed information about individual images is steadily increasing, it is necessary to use new type of information for picture description. This sort of information usually includes description of relations among various objects in the picture. These relations may have different forms and meanings beginning from purely geometric relations (like an object is below another object) and ending up with relations that have semantic character (like a person talks to another person).

Both types of information might be in general linked up in various manners. This fact requires an easy switching from one type of relations to different type of relation. Another important topic is the possibility to filter out the information that is not relevant in given context. This feature would ease the browsing in a picture as the number of relations could be limited by means of proper filter. In principle the filtering process could be interpreted as data transformation process. The need for data transformation is more general in picture databases. The solution to this problem is highly influenced by the implementation tool used.

The tool that has all features necessary is XML where data transformations could be easily specified by means of XSL transformation sheets. We have implemented experimental system for picture description creation, filtering and browsing based on these standards (XML, XSL).

Due to generality of XSL it is possible to describe the querying process in various stages of data transformation (e.g. picture data input, internal data transformations, picture data output). It is also possible to define the way the query results are visualized.

e-mail: [e-mail]
URL: http://www.cgg.cvut.cz; http://www.cgg.cvut.cz/Research/bis
Other contact: Pavel Slavik, [e-mail]

Semantics for voice recognition systems - a project idea (Christian Gronoff, Semiosphere)

WHAT we wish to achieve:

• 
  The Intelligent Conversation Management Agent is the missing link for users and the building component for today’s vocal applications.
  
• 
  The agent dialogues with the user thanks to text-to-speech and speech recognition, associated with semantic engine, until it has an acceptable response to establish a sharp and balanced search.
  
• 
  Query is transfered to the application which delivers required results.
  
• 
  IMCA takes advantage of the VoiceXML standard.
  

Benefits:

• 
  SAIMCA will be configured as to be engine vendors independant (text to speech & recognition)
  
• 
  Key component to Natural Langage understanding
  
• 
  Major innovation for Multilingual Voice portal.
  
• 
  Improve efficiency of vocal browser applications
  
• 
  Voice technology nearer to match Write technology
  

Applications

• 
  NEW MULTILINGUAL MAIL READER GENERATION: Retrieve mail over the phone using powerfull filters.(name, date, priority…)
  
• 
  MULTILINGUAL VOCAL CONFIGURATOR: Perform configuration over the phone, pricing, real estate activity….etc…
  

e-mail: [e-mail]

Knowledge based indexing (Ralph Traphoener, tecinno GmbH)

e-mail: [e-mail]
URL: http://www.tecinno.com

OntoMap: facilitating easier access, understanding and comparison of the existing upper-level models (Cyc, SENSUS, WordNet, EuroWordnet, SIMPLE, GUM) (Atanas Kiryakov, OntoText Lab., Sirma AI Ltd. )

The goal of the project is to facilitate an easy access, understanding, and reuse of upper-level and general purpose ontologies as well as lexical knowledge bases. The guideline of OntoMap is that currently the evaluation of the existing upper-level models is too expensive mostly because of technical problems such as different representations and terminologies used. Additionally, the situation could be improved by making available formal mappings between the upper-level ontologies that will make easier any kind of comparative studies.

Technically the result will be a web site that provides a number of upper-level ontologies and mappings between them in a number of different formats — PROLOG, KIF, SQL, OIL, XML, etc. An application server will be developed giving a uniform access to the resources via widely accepted standards such as OKBC (on a semantic and pragmatic level) and RMI, SOAP, and CORBA (on implementation level). A conceptual model (meta-ontology) will be developed inspired by the property classification of Guarino and Welty, ODE (Methontology) and Cyc. All the ontologies and mappings will be classified according to the KIF (ANSI draft proposal) conformance dimensions.

After the end of the project the server will be further supported and developed by OntoText Lab. Paid services for ontology conversion, classification, hosting and mapping based on OntoMap will be provided for both upper-level and domain specific ontologies.

e-mail: [e-mail]

URL: http://www.ontotext.com