Dr. Anoop Sarkar, from the Natural Language Lab at Simon Fraser University recently won the IBM Faculty Award for his research in computational linguistics and natural language processing.

 The IBM faculty award is a competitive worldwide program given to faculty members having an outstanding reputation for contributions in a research area that is potentially useful to IBM Research and Development. The candidate for the award is nominated by an IBM employee, and the candidate submits a proposal for a research program whose merits are then evaluated by IBM.

 Dr. Sarkar's research deals with implicitly training a computer to understand, extract information from or translate natural languages that are spoken and written by humans such as English, Spanish, etc.He says, "Humans don't get explicitly trained in language, but get implicitly trained". For example, one aspect of natural language processing is to implicitly train a computer to translate between languages using examples of previously translated sentences. The idea is to feed a computer algorithm, lots of examples of translations from a source language (say, Chinese) to a target language (say, English), and expect the computer to learn from these examples.

 Specifically, Dr. Sarkar's research focuses on how much training needs to be given to the computer. For example, let's assume that fifty million sentences of Chinese to English translated texts are currently being used to train a translation system. But, what happens when there is a language pair (of source and target languages) for which there isn't enough text to train the system? His research attempts to find out whether it's required to feed the computer all the examples needed to do the translations, or whether you can get away with giving only a few examples, and have the computer infer from this limited amount of supervision.

 For more information on the Natural Language Lab, please visit http://natlang.cs.sfu.ca.

 To learn more about Dr. Anoop Sarkar's work, please refer to his website: http://www.cs.sfu.ca/~anoop

 Written by Salima Vastani