Automatic keyphrase extraction using graph-based methods

Josiane Mothe (1, 2) , Faneva Ramiandrisoa (1) , Michael Rasolomanana (3)

1 IRIT-SIG - Systèmes d’Informations Généralisées
2 UT2J - Université Toulouse - Jean Jaurès
3 Université d'Antananarivo

This paper analyses various unsupervised automatic keyphrase extraction methods based on graphs as well as the impact of word embedding. Evaluation is made on three datasets. We show that there is no differences when using word embedding and when not using it.

Mots clés: Information systems; Information retrieval; Automatic keyphrase extraction; Graph-based methods; Word-embedding; Word2Vec

Domaines: Informatique et langage [cs.CL]; Recherche d'information [cs.IR]; Théorie de l'information [cs.IT]

Josiane Mothe, Faneva Ramiandrisoa, Michael Rasolomanana. Automatic keyphrase extraction using graph-based methods. 33th ACM Symposium on Applied Computing (SAC 2018), ACM, Apr 2018, Pau, France. pp.728-730, ⟨10.1145/3167132.3167392⟩. ⟨hal-02640988⟩