Thisuri Chamathka Lekamge and TGI Fernando published a research paper titled “Finding the Gender of Personal Names and Finding the Effect of Gana on Personal Names with Long Short Term Memory” at IEEE ICTer conference 2019 organized by University of Colombo School of Computing (UCSC) held on September 03, 2019 at UCSC.
Abstract:
Naming a baby is a very integral part of our life. As well as, naming a business with an attractive and suitable name takes the first place in business world. In Asian countries, especially in Sri Lanka, people tend to follow rituals and customs introduced by astrology when naming a baby or a business. Because they believe that good luck can be achieved by following astrological concepts like ‘Mathra’ and ‘Gana’ in naming. As a result of that, people tend to spend more money in finding a suitable name which follows astrological concepts. Also, when naming a new born according to gender, parents think whether that name is suitable for their baby boy/girl or not. On the other hand, as web services grow, personal names have become one of the most abundant sources of data on the web. So, if one can predict or suggest proper gender for personal names; web or mobile applications can benefit from knowing each user’s gender. Research studies on predicting the gender of personal names and the effect of ‘Gana’ on personal names with deep learning algorithms have not been studied. Therefore, our main goal was to implement separate models for gender classification and finding the effect of ‘Gana’ on personal names (forenames/most use name) which are in text format using Recurrent Neural Networks. As a result of that, we developed two LSTMs (a special type of Recurrent Neural Networks) that produced promising results.
Web-based Interface Tool – EXYONA