About me

Hi, I’m Debasis Ganguly, anagram of led by assuaging or for that matter, ably degaussing. By profession, I’m a lecturer in data science working in the IDA (Information and Data Analytics) group at the University of Glasgow. Formerly, I was a research staff member at IBM Research Europe, Dublin, Ireland. By hobby, I’m a chess player, a long distance runner, a nature enthusiast, and an aspiring globetrotter.

Generally speaking, my research activities span topics on Information Retrieval (IR) and Natural Language Processing (NLP). More specifically, I’m interested in applying semantic relationships between text units (e.g. by embedding the text units in a vector space over reals) for improving various IR and NLP tasks.

My PhD thesis was on the topic of improving retrieval models based on the information utilised at the level of sub-documents, e.g., topics or passages.

My recent research interests include application of multi-objective neural networks for bias removal of classification systems. Check out our recent AAAI paper on debiasing and its follow up CIKM’21 paper Multi-objective Few-shot Learning for Fair Classification.

Recently, I’m interested in investigating trustworthiness as an additional dimension of effectiveness of IR models; check out our recent CIKM’22 paper Measuring and Comparing the Consistency of IR Models for Query Pairs with Similar and Different Information Needs.

At IBM, my primary research project involved working towards an AI solution that will be able to make predictions about the likelihood of the success of a study (i.e. how likely it is that a given user group with certain characteristics with a set of registered interventions) is likely to change their behaviour.

In addition to the primary research project, I maintain a strong partnership with the academia. Currently I’m co-supervising one PhD student at the Indian Statistical Institute, the area of his research being privacy preserving machine learning.