Unsupervised Automatic Keyphrases Extraction Algorithms

Abstract

Massive volumes of images of museums or art collections, or made available by artists and photographers, more and more often, are available on the web, along with some metadata, essential for their characterization and retrieval. A set of (scored) keywords/keyphrases that characterize the semantic content of the documents should be, automatically or manually, extracted and/or associated. We present here a work-in-progress to evaluate different methods for the unsupervised keyword extraction to Italian and English datasets. In the paper datasets, algorithms and approaches are presented and discussed together with some preliminary results referred to relatedness of terms.