Résumé
We present different approaches to construct n-Ary relation instances from scientific articles in experimental domains (i.e. food packaging) driven by a Termino-Ontological Resource dealing with the food packaging domain. Our method starts with the identification of partial n-Ary relations in the document tables and seeks to complete them with argument instances from text of the articles. Our approach is based on a new multi-feature representation (i.e. STaRe - Scientific Table Representation) and the n-Ary relation reconstruction based on text-mining methods.