Bishal is a Ph.D. Scholar at the Department of Computer Science and Engineering, Indian Institute of Technology Kharagpur from July, 2018. He is working under the supervision of Prof. Pawan Goyal. His research interest involves building new machine learning models for Chatbots and exploring methods for using large scale Knowledge Graphs in other downstream applications. He is a member of the Complex Network Research Group(CNeRG) as well.
He did his B. Tech and M. Tech in Electronics and Electrical Communication Engineering (spcl. in Visual Information Processing and Embedded Systems) from Indian Institute of Technology, Kharagpur. He also has a minor in Computer Science and Engineering.
PhD Advisor: Prof. Pawan Goyal
Table of Contents
- Curriculum Vitae
- Active Research Projects
- Courses done during PhD
- Social Networking
Curriculum Vitae (Link)
Active Research Projects
Probing Neural Dialogue Models: Although neural network based models give great numerical accuracy, they are not well known for their interpretability. In this project, we are designing probes & experiments to understand the inner workings of neural network based dialog systems.
Task-Oriented Dialogue System: In this project, our aim is to design better architectures for building dialogue systems in specific domains and for well-defined tasks. Unlike open-domain dialogues, the main focus of task-oriented dialogues is to satisfy the user’s goals.
Development of Evaluation Methods: Evaluating open-domain dialog systems is a very challenging task due to the versatile and subjective nature of the idea of quality of a conversation. Even harder is representing them mathematically. We are working on developing aspect-specific (like dialog acts, intent, etc.) metrics for evaluating dialogs.
Long-Term Memory in Language Learning Models: Modern natural language understanding (NLU) frameworks relying on deep neural networks has a very limited working memory, upto one single document or a fixed-sized structured database. This limits the usability of these NLU models in interactive environments (e.g. chatbots). We are working to developing learning algorithms for inducing the capability of long-term memory in neural network based language understanding models.
- Post-Hoc Explanations for Recommendation Systems: The best performing models for any task are often based upon very complicated neural architectures. The same is also true for recommendation systems. So often a separate simpler and more interpretable model is used for generating explanations. In this project, we are developing ways to generate post-hoc explanations for movie recommendation systems using open knowledge-graphs (e.g. DBPedia) and to measure the power of these post-hoc methods in terms of their explainability and faithfulness to the original model.
Courses done during PhD
- Information Retrieval
- Scalable Data-Mining
- Advanced Graph Theory
[Google Scholar], * Equal Contributions
Bishal Santra, Sumegh Roychowdhury, Aishik Mandal, Vasu Gurram, Atharva Naik, Manish Gupta, Pawan Goyal. Representation Learning for Conversational Data using Discourse Mutual Information Maximization. NAACL 2022 (Long Paper) [PDF]
Computational Linguistics (2021)
Amrith Krishna, Bishal Santra, Ashim Gupta, Pavankumar Satuluri and Pawan Goyal. A Graph Based Framework for Structured Prediction Tasks in Sanskrit. Computational Linguistics Vol. 46, No. 4, December 2020. [PDF]
Arxiv Preprint 2020
Bishal Santra, Prakhar Sharma, Sumegh Roychowdhury, Pawan Goyal. Exploring Effects of Random Walk Based Minibatch Selection Policy on Knowledge Graph Completion. On Arxiv. [PDF]
Soumya Sharma*, Bishal Santra*, Abhik Jana, T Y S S Santosh, Niloy Ganguly and Pawan Goyal. Incorporating Domain Knowledge into Medical NLI using Knowledge Graphs. EMNLP-IJCNLP 2019, November 3-7, Hong Kong (short paper). [PDF]
Amrith Krishna, Vishnu Dutt Sharma, Bishal Santra, Pavan Kumar Satuluri and Pawan Goyal (2019). Poetry to Prose Conversion in Sanskrit as a Linearisation Task: A case for Low-Resource Languages. ACL, July 28th - August 3rd, Florence, Italy (short paper). [PDF]
Abhinav Kumar*, Aishwarya Gupta*, Bishal Santra*, Lalitha Srinivasan*, Manasa Kolla*, Mayank Gupta* and Rishabh Singh*, VPDS: An AI-based Automated Vehicle Occupancy and Violation Detection System, IAAI (application conference of AAAI), 2019 [PDF]
Amrith Krishna, Bishal Santra, Sasi Prasanth Bandaru, Sahu, Gaurav, Sharma, Vishnu Dutt, Satuluri, Pavan Kumar and Pawan Goyal (2018). Free as in Free Word Order: An Energy Based Model for Word Segmentation and Morphological Tagging in Sanskrit. EMNLP, Brussels, Belgium, October 31-November 4, 2018. [PDF]
Amrith Krishna, Bishal Santra, Pavan Kumar Satuluri, Sasi Prasanth Bandaru, Bhumi Faldu, Yajuvendra Singh and Pawan Goyal (2016). Word Segmentation in Sanskrit Using Path Constrained Random Walk. 26th International Conference on Computational Linguistics (Coling), Osaka, Japan, December 11-16, 2016 (Poster). [PDF]
M.Tech. Project (2017)
Bishal Santra. Supervised by (Prof. Priyadip Ray and Prof. Saswat Chakrabarti). Image Super-resolution with Sparse Image Representation using Dictionary Learning. [PDF]
Or else in General Research Lab, Computer Science and Engineering Dept., IIT Kharagpur, India