Building models of documentary and factographic retrieval in digital libraries

The building of models of documentary and factographic retrieval is discussed in the context of digital libraries working with documents of a fairly arbitrary structure. An overview is given of a technology for extracting factographic information from scientific documents of a fairly arbitrary structure. A model is proposed for the classification of documents in a digital library. The model is based on the use of the tolerance relationship and takes into account the possible lack of a priori given classifiers. In creating factographic systems, it is suggested that the concept of a fact should be understood as a totality of relationships, as contained in the text and document metadata, between the entities described in the information system ontology. A simple model is proposed to describe the ontology of a factographic system.