Advertisement

Help
You are here: Rediff Home » India » News » PTI
Search:  Rediff.com The Web
Advertisement
  Discuss this Article   |      Email this Article   |      Print this Article

Gujarat girl to present paper on Indian fonts in US
Get news updates:What's this?
Advertisement
September 25, 2007 09:31 IST

A young computer wiz from Gujarat will present a paper on Indian font recognition softwares at a global conference in the US next month.

Sandhya Sitaraman (20), studying national language processing at an institute in Surat, is the only representative from India invited to present a paper at the forum, which is specially for undergraduate women pursuing careers in computer sciences.

Personal computers and laptops have become an everyday tool for professionals in all walks of life across India for some years now, but the lack of adequate font recognition softwares has limited their usage mostly to those who understand English.

Sitaraman is gearing up to overcome this handicap and will present a paper on 'Artificial Intelligence Recognition for Indian languages' at the conference at Carnegie Mellon University.

"I was very happy that my paper has been selected and I will be given time to talk about my subject amid such a recognised panel of experts in the computer field," Sitaraman told PTI.

Clearing the air about artificial intelligence, Sitaraman said "Many people have the misconception that AI is all about cyborgs and inserting chips into human beings which is not true. A lot of AI today is just about fields like human computer interaction and natural language processing to make communication smoother to help solve complex problems easily."

She said the aim of her paper is to devise a system that automatically determines which of a given set of fonts a Gujarati document uses and then concerts the document into a standard Gujarati font for viewing.

"The problem of font recognition was treated as a problem of solving random substitution cipher. Single character frequency analysis was used to arrive at frequency distribution of the fonts and the Euclidean distance between the distributions was used as a similarity measure.

The approach was tested on 153 documents using six different Gujarati fonts and 100 per cent accuracy was obtained in the recognition," Sitaraman said about the technicalities.

Among those who will be present when Sitaraman presents her paper is the winner of this year's AMC Turning award (considered to the Nobel prize [Images] in computing).

Since its inception, Carnegie Mellon University has grown into one of the leading educational institutes in computer sciences.


© Copyright 2007 PTI. All rights reserved. Republication or redistribution of PTI content, including by framing or similar means, is expressly prohibited without the prior written consent.
 Email this Article      Print this Article

© 2007 Rediff.com India Limited. All Rights Reserved. Disclaimer | Feedback