Exploiting and Evaluating a Supervised, Multilanguage Keyphrase Extraction Pipeline for Under-Resourced Languages