Context and Keyword Extraction in Plain Text Using a Graph Representation

Document indexation is an essential task achievedby archivists or automatic indexing tools. To retrieverelevant documents to a query, keywords describing thisdocument have to be carefully chosen. Archivists haveto find out the right topic of a document before startingto extract the keywords. For an archivist indexingspecialized documents, experience plays an importantrole. But indexing documents on different topics is muchharder. This article proposes an innovative method foran indexing support system. This system takes as inputan ontology and a plain text document and provides asoutput contextualized keywords of the document. Themethod has been evaluated by exploiting Wikipedia’scategory links as a termino-ontological resources.