Unsupervised learning of semantic relations of a morphologically rich language
by J. Balaji; P. Ranjani; T.V. Geetha
International Journal of Information and Communication Technology (IJICT), Vol. 8, No. 4, 2016

Abstract: Use of semantic concepts and relations for NLP applications including information retrieval and web search is a major area of research. In this context, semantic relation extraction from open domain web documents is important not only for English but also for other languages where hand-crafted rules covering the variability in expressing semantic relations or semantically tagged corpora are not available. To meet this crucial need, an unsupervised approach to learn semantic relations between concepts specifically for morphologically rich, relatively free word order languages gains importance. Unlike previous approaches that used word order and morpho-syntactic features, in this paper, we use morpho-semantic features with a minimal amount of co-occurrence features to extract semantic relations. The features are used to learn whether a concept node is a source of a concept-relation-concept subgraph, what the relation associated with the source is and which is the destination node using appropriate probabilities. Determining the destination node requires a novel source-destination probability because of the relatively free word order nature of the language. The approach was evaluated using 20,000 document corpus from both tourism and news domain. The results showed that the approach gave F-measure 0.50 for a morphologically rich language without using syntactic features.

Online publication date: Wed, 01-Jun-2016

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Information and Communication Technology (IJICT):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com