GDTR: A graph-based domain terms ranking model

GDTR-a graph-based d,om,a,in terms ranking model is proposed, in thispaper. The model first constructs a similar graph on candidate domain terms.Then it computes the centrality score of each candidate d,om,a,in term, by a,graph centrality algorithm,. Finally, all the candidate d,om,a,in terms areranked, by a, combined, score of their frequency and, centrality score. Theranked, domain terms are evaluated, by a, standard, domain term, list. The GDTRmodel is tested, and, compared with related, work on two domain corpus, theexperimental results show that GDTR obtains competitive results with thestare-of-the-art domain terms ranking methods. Our contribution is two-fold, oneis a, graph-based, domain terms ranking model GDTR, the other is a, new term,similarity measure SP, which computes term, similarity by their appearance inlonger phrases. ICIC International © 2010.