Precise information retrieval in semantic scientific digital libraries

When scientists are looking for information in document collections, they generally have a precise objective in mind. They try to answer specific needs such as finding the definition of a concept, checking whether an idea has already been tested, or comparing the scientific conclusions of several documents. To build better information system model it is important to understand the needs of scientists in terms of information and their seeking information behaviour. The contributions of this thesis are to a) offer a user model; b) propose a generic annotation model (SciAnnotDoc) for scientific documents; c) demonstrate that the annotation model is realistic enough to be used by both manual and automatic methods of annotation; d) build a faceted search interface (FSAD) and e) evaluate the model with scientists and show that the FSAD interface is more effective than a keywords-search, notably in terms of user preference, recall and F1 measure