Semantic annotation: Mapping text to ontologies
Résumé
This position paper discusses various issues concerning the semantic web. We claim that reflection on formats has attracted the most important part of the research effort so far, whereas the web (including the future semantic web) is mainly made of texts. From our point of view, research should attach a major role to natural language properties and deal with complexity, ambiguity and polysemy. We examine a set of resources and search engines, which are good candidates to provide enhanced indexing capacities to existing tools. We give some examples of applications intended to provide a semantic annotation of different kinds of texts. Even if some of these technologies are already available, we show that they suppose to make more complex the initial model from T. Berners-Lee if we want them to be able to take into account language diversity.