Semantic text processing and applications
-----------------------------------
          
ABSTRACT
-----------------------------------
The availability of large and heterogeneous datasets has a strong impact 
on the way information is extracted and retrieved. In this presentation, 
I will first introduce Explicit Semantic Analysis (ESA), a text 
processing technique developed by E. Gabrilovich that is based on the 
exploitation of structured corpora such as Wikipedia. In ESA, words 
and/or texts are projected onto a large conceptual space and this 
projection is then used to assess their relatedness. Despite its 
simplicity - the techniques relied on a TF-IDF modeling of Wikipedia 
articles to link words to concepts - the richness of the underlying 
encyclopedic knowledge makes the method competitive in a number of 
applications.
A number of applications based on variations of ESA will constitute the 
core of the second part of the presentation. I will discuss the utility 
of ESA in image retrieval, text retrieval and bilingual lexicon creation.
The last part of the talk will be focused on open problems associated to 
explicit semantic models and to on related future works within the MUCKE 
project.
          
SPEAKER(S)
-----------------------------------
Adrian POPESCU, Researcher, PhD
CEA LIST
Franta
-----------------------------------
Adrian Popescu received his Engineering Degree and his 
PhD in Computer Science from TELECOM Bretagne, France. 
He is currently a researcher with the CEA LIST, France 
and his research interests span the areas of text 
processing, multimedia information retrieval and 
social networks.
-----------------------------------