Fuzzy semantic annotation of XML documents

The e.dot project consists in the semi-automatic construction of an XML data warehouse in the field of food safety. The construction and the querying of the data warehouse are guided by an ontology. We are working on the annotation of tables extracted from scientific articles in microbiology with terms of the ontology. Terms from those articles may be connected to several terms of the ontology, but each connection is uncertain. We represent the annotations as possibility distributions: we associate each term of the ontology with the degree of possibility that this term represents the original term from the article. In this paper, we present two ways of computing this degree of possibility. One is based on a syntactic comparison, each word of an ontology term being weighted according to its “semantic power” in the term. The second way of computing the degree of possibility is based on the first one, but it also uses the hierarchy defined in the ontology