Postdoctoral Position at the University of Lyon

<please excuse multiple copies>

The University of Lyon and the INSA of Lyon announce a position 
for a one year postdoctoral fellowship at the University of Lyon. 

Period : 		
12 months, beginning sept. 2001

Team : 		
Cognition & Experience Team (

Lab : 		
Lyon 1 University ( 

Place : 	
Lyon, France

Keywords : 	
Annotation, Document, Experience, Knowledge representation, RDF, 
Semantic Indexing, Task, Usage

Description : 	

For several years, our research team has been involved in the 
issue of helping people to find knowledge sources that could 
be reused for his/her particular task. This question was firstly 
tackled through research on the Case Based Reasoning paradigm 
(CBR). CBR consists in finding in a case base former cases that 
could be reused to solve new problems. Several projects have 
issued from this research effort, such as PAD’IM and DESIGNER 
[1,2], ACCELERE [3], which were about very precise contexts, 
generally some very well known tasks. Later, this approach was 
applied to the particular task of information retrieval (RADIX 
project [4]) which demonstrated that the information retrieval 
task is in fact mostly influenced by the task (generally unknown) 
for which it is necessary to find this information. This is not 
an original statement, and many researchers try to cope with it 
building "user profiles" (which we think would gained being 
called "usage profile"). Indeed the necessity for a general model 
for information retrieval task has arisen, and it appeared it was
necessary to take into account the real user’s task context.

We gave this orientation to our research effort along two projects, 
toward a "usage indexing" for contents that can be accessed from 
one’s computer. In SESAME [5], we developed an annotation model 
for audio-visual content description (Annotation Interconnected 
Strata : AI-Strata). This approach is based on the principle 
that annotating a document is in fact expressing the ongoing
exploitation task for this document (whichever that task can 
be: edition, indexing, analysis, research). The approach is 
basically not a rigid one, and also gives possibility to express 
knowledge as in knowledge based systems and formal ontologies. 
We set up an annotation graph expressing document description, 
and consider that usage of the graph occurs in a contextual way 
(as path expression from known nodes). Several tools have been 
designed and are being developed allowing to exploit the graph. 
Indeed, we do consider that useful experience (from document 
exploitation) lies in the description graphs, and all that it 
is necessary being able to exploit this experience through 
dedicated user’s tasks help ingassistants.

Our team is working upon these assumptions in the RECIS project 
(INRIA / France Telecom / LISI [6]). Our goal is to design 
efficient tools for users searching for audio-visual documents 
on the basis of audio (signature descriptors), visual (images 
and video signature descriptors) or semantic descriptions 
(textual, semantic graph descriptors). Our contribution consists 
in designing software assistants allowing the user to retrieve 
audio-visual sequences on the base of descriptors graphs capable 
on describing a "task signature". For instance, elaborating a 
request graph in the context of information retrieval is very 
similar to elaborating a description graph in an indexing task: 
in any case, the purpose is to annotate "in context" what is 
searched or indexed. The context is the context of the task 
for which descriptors are being supposed making sense, and in 
the framework of which they are supposed to have sense. This 
context is expressed with task signatures (that can be abstract 
or correspond to real descriptions). 

Our team is committed to an effort of formalisation, generalisation, 
and normalised representation of this approach, and we want to 
associate a young, non French, researcher to this task for a one 
year period. Beyond the RECIS project, other researchers of our 
team will be associated to this effort, trying to apply it to other 
issues (design assistance, learning assistance, etc.). Applicants 
should have a PhD in Computer Science in the area of knowledge 
representation, document indexing, information retrieval, content 
and document structure description languages (XML, RDF, MPEG7). 
Interests for linguistics, semiotics and interpretative systems 
will be appreciated. 

[1] Alain Mille, Jean Marc Chartres, Eric Niel, Béatrice Fuchs, 
and Benoit Chiron (1995). Intelligent workstation for immediate 
decision helping in process supervision. Computers and Chemical 
Engineering, Vol 19, 1995, pages s815-s821, 1995. 
[2] B. Fuchs, A. Mille, and B. Chiron (1995) Operator decision 
aiding by adaptation of supervision strategies. In Lecture Notes 
in Artificial Intelligence vol 1010, First International Conference 
on Case-Based Reasoning, ICCBR'95, pages 23--32, Sesimbra, 
Portugal, 1995. Springer-Verlag, Berlin, Germany.
[3] Olivier Herbeaux and Alain Mille. (1998) Accelere: a case-based 
design assistant for closed cell rubber industry. In R. Milne, 
A. Mac Macintosh, and M. Bramer, editors, Proceedings of ES98, 
the Eighteenth Annual International Conference of the British 
Computer Society, Specialist Group on Expert Systems, pages 69--82. 
BCS Conference Series, Springer-Verlag, London, 1998.
[4] Françoise Corvaisier, Alain Mille, and Jean Marie Pinon. (1997) 
Information retrieval on the world wide web using a decision making 
system. In RIAO 97, Montréal, Juin 1997, pages 284--295, 1997.  
[5] Yannick Prié, Alain Mille et Jean-Marie Pinon (1999) A Context-
Based Audiovisual Representation Model for Audiovisual Information 
Systems, Context'99, Second International and Interdisciplinary 
Conference on Modeling and using Context, Trento, sept. 1999. 
Lecture Notes in Artificial Intelligence, vol. 1688, pp. 296-309.
[6] Elöd Egyed-Zsigmond, Yannick Prié, Alain Mille et Jean-Marie 
Pinon (2000) A graph-based audiovisual document annotation and 
browing system. RIAO'2000, Paris, apr. 2000, vol. 2, pp. 1381-1389.

Location information :
City of Lyon : 

Rhônes-Alpes Region : 

Fee/month : 	~10,000 FRF - 1,524 EUR
Contact : 	

Alain Mille - Professor 

Yannick Prié - Associate Professor 

Send curriculum + references to 

Pr. Alain Mille 
LISI - Bât Nautibus (710) 
UFR Informatique 
Université Claude Bernard 
F-69622 Villeurbanne Cedex

Received on Wednesday, 23 May 2001 07:43:48 UTC