Artificial Intelligence

Towards Semantically Interpretative and Contextually Sensitive Text-Mining

Wednesday, September 12, 2018, 11:00am - 12:00pm PDTiCal
6th Floor Conf Rm (#689)
This event is open to the public.
AI Seminar - Interview Talk
Vishrawas Gopalakrishnan

With the growth of the world wide web and large-scale digitization of data, text-mining has assumed a pivotal role in data analytics. In conjunction with the fact that 'text' is a ubiquitous means of representation, text-mining has also grown to be an inter-disciplinary field with applications ranging from social media analytics for product promotion to biomedical text-mining for new drug discovery and even to politics. This has led researchers to develop generalized algorithms and frameworks and move towards making the individual domain application more of a 'plug-and-play'. While this is desirable, as it promotes scalability across domains, the fact that many of the intermediary steps of such algorithms are not semantically interpretable by end-users, stands in conflict with the requirements of users in high-risk areas like medicine.


In this talk, I present a suite of approaches, especially aimed for short texts, that target various components of an end-to-end pipeline of a text-mining frame-work.  The proposed algorithms are generic, language-independent, semantically interpretable and contextually sensitive that can easily be ported across application domains. Furthermore, the developed algorithms require none to minimal human interventions and yields results that are at par with those of existing state-of-the-art supervised techniques. Towards the latter half of the talk, I shall present some application specific settings where these algorithms are dovetailed and adapted to meet the requirements.




Dr. Vishrawas Gopalakrishnan is a data scientist in the 'Analytics Center of Excellence' group in IBM Watson Health. His primary research interest is in text mining, and his thesis focused on developing unsupervised and contextually sensitive algorithms for both core tasks in a text mining pipeline as well as in application-related settings. He has actively participated in other research topics as well, with topics ranging from multi-source learning in biological networks to classical NLP and IR problems like extraction of multi-word expressions in transliterated texts.

At IBM Watson Health, he is responsible for NLP and semantics related projects within his team and closely works with IBM Research in maturing research assets and innovating commercial products.

Vishrawas graduated in 2017 from State University of New York at Buffalo where he worked under the tutelage of Prof. Aidong Zhang. He has published in top-tier conferences and journals like VLDB, CIKM, KDD, ICDM, Bioinformatics and TKDE. He serves as a reviewer for many of these conferences and journals and is currently a PC member for this year’s KDD Workshop on Machine Learning for Medicine and Healthcare.


« Return to Events