Determining Termhood for Learning Domain Ontologies using Domain Prevalence and Tendency

In the course of reviewing existing automatic term recognition techniques for applications in ontology learning, we came across four issues which can be improved upon. We proposed a new mechanism that incorporates both statistical and linguistic evidences for the computation of a final weight defined as Termhood (TH) for ranking term candidates. The analysis of the frequency distributions of the term candidates during our initial experiments revealed three advantages for higher quality term recognition.

• Type: Conference
• Year: 2007
• Where: AUSDM
• Authors: Wilson Wong, Wei Liu, Mohammed Bennamoun
• Keywords: AUSDM 2007, Data Mining, Term Candidates, Term Recognition, Term Recognition Techniques