Abstract
In this paper we introduce a minimalist hypothesis for keywords extraction: keywords can be extracted from text documents by considering concepts underlying document terms. Furthermore, central concepts are individuated as the concepts that are more related to title concepts. Namely, we propose five metrics, that are diverse in essence, to compute the centrality of concepts in the document body with respect to those in the title. We finally report about an experimentation over a popular data set of human annotated news articles; the results confirm the soundness of our hypothesis.
Author supplied keywords
Cite
CITATION STYLE
Colla, D., Mensa, E., & Radicioni, D. P. (2017). Semantic measures for keywords extraction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10640 LNAI, pp. 128–140). Springer Verlag. https://doi.org/10.1007/978-3-319-70169-1_10
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
