In the context of a virtual library of scientific papers, the work reported here addresses an original method of representing documents for information retrieval (IR) use. The approach we propose aims at exploiting the logical structure of documents in order to index them hierarchically. It also use ontology to form content-based and context-based annotations. These semantic annotations make it possible to associate information additional to the documents, to specify certain parts of this one and finally to share them within the framework of a working group. Finally, an adaptation technique for exploiting the scientific papers indexes is proposed in order to satisfy actors' needs.
展开▼