HYBRIDRANK: A COLLABORATION BETWEEN SUPERVISED AND UNSUPERVISED APPROACHES FOR KEYPHRASE EXTRACTION

Abstract

Traditionally, keyphrases (or keywords) have been manually assigned to documents by their authors or by human indexers. This, however, has become impractical due to the massive growth of documents|particularly short articles (e.g. microblogs, abstracts, snippets)|on the Internet each day, thus creating a need for systems that automatically extract keyphrases from documents. Automatic keyphrase extraction methods have generally taken either supervised or unsupervised approaches. Supervised methods extract keyphrases by using a training document set, thus acquiring knowledge from a global collection of texts. Conversely, unsupervised methods rank phrases by their importance in a single-document context, without prior learning. We present a hybrid keyphrase extraction method for short articles, HybridRank, which leverages the benets of both approaches. Our system implements modied versions of the TextRank [6] (unsupervised) and KEA [16] (supervised) methods, and applies a merging algorithm to produce an overall list of keyphrases. We have tested HybridRank on more than 900 abstracts belonging to a wide variety of subjects, including engineering, science, physics and IT, and show its superior eectiveness. It is observed that knowledge collaboration between supervised and unsupervised methods can produce higher-quality keyphrases than applying these methods individually.