Abhirut Gupta

Research Software Engineer
IBM Research
Email: abhirutgupta@in.ibm.com

I am a Natural Language Processing Researcher at IBM Research, and I work on automating technical support. Only around 19% of questions in technical support are simple "factoid". Around 50% questions require complex understanding of the problem, the setup which led to the problem, and attempts by the user to mitigate it. I spend my days building algorithms for "understanding" these complex tech support problems. Over the last 4 years I've also worked on the following problems -

  • Improving retrieval performance on tech support Question Answering
  • Building a support ontology and Knowledge Graph
  • Automating construction of support chatbots from content
  • Extracting procedures from support documents


Research Interest

My research interest lies in exploring Natural Language Understanding and Reasoning for real world applications. In particular, problems where pattern matching approaches fail to meet human performance and require a thorough understanding of unstructured text to solve, interest me. Problems with very limited supervised task specific data are also of interest.



  1. Prateeti Mohapatra, Yu Deng, Abhirut Gupta,Gargi Dasgupta, Amit Paradkar, Ruchi Mahindru, Daniela Rosu, Shu Tao, Pooja Aggarwal. Domain Knowledge Driven Key Term Extraction for IT Services. 16th International Conference on Service-Oriented Computing, ICSOC 2018
  2. Abhirut Gupta, Abhay Khosla, Gautam Singh, Gargi Dasgupta. Mining Procedures from Technical Support Documents. arXiv preprint. arXiv:1805.09780
  3. Abhirut Gupta, Anupama Ray, Gautam Singh, Gargi Dasgupta, Pooja Aggarwal, and Prateeti Mohapatra. Semantic Parsing for Technical Support Questions. COLING 2018

  4. 2016

  5. Abhirut Gupta, Arjun Akula, Gargi Dasgupta, Pooja Aggarwal, and Prateeti Mohapatra. Desire: Deep Semantic Understanding and Retrieval for Technical Support Services. ICSOC 2016 Workshops


  1. System and Method for Instance Specific Aspect Based Cross Documents Sentiment Aggregation. Application Number 14/937551
  2. Deep Learning based Unsupervised Event Learning for Economic Indicator Prediction. Application Number 15/258176