Contribution à l’acquisition automatique de nouveaux sens

Abstract

This paper tackles the question of automated meaning acquisition. In several cases, meaning change can be considered as a process spreading both in time and through domains. In addition, this phenomenon relies on interactions between word meaning in discourse and dictionary definitions. We outline the main steps of a procedure to highlight the spread of a new word meaning in corpus, to extract semantic content represented by domain labels and to match corpus and dictionary information. The methodology is illustrated by experiments on several lexical units observed in a press corpus ranging from 2004 to 2010 and divided into several domains.