Including category information as supplements in latent semantic analysis of Hindi documents
by Karthik Krishnamurthi; Vijayapal Reddy Panuganti; Vishnu Vardhan Bulusu
International Journal of Computational Science and Engineering (IJCSE), Vol. 15, No. 1/2, 2017

Abstract: Latent semantic analysis (LSA) is a mathematical model that is used to capture the semantic structure of documents by using the correlations between the textual elements in them. LSA captures the semantic structure very well being independent of external sources of semantics. However, the model's performance increases when it is supplemented with extra information. The work presented in this paper is to modify the model to analyse word correlations in documents by considering the document category information as supplements in the process. This enhancement is called supplemented latent semantic analysis (SLSA). SLSA's performance is empirically evaluated in a document classification application by comparing the accuracies of classification against plain LSA for various term weighting schemes. An increment of 1.14%, 1.30% and 1.63% is observed in the classification accuracies when SLSA is compared with plain LSA for tf, idf and tfidf respectively in the initial term-bydocument matrix.

Online publication date: Mon, 21-Aug-2017

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 Computational Science and Engineering (IJCSE):
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