P
Pushpak Bhattacharyya
Researcher at Indian Institute of Technology Patna
Publications - 576
Citations - 8724
Pushpak Bhattacharyya is an academic researcher from Indian Institute of Technology Patna. The author has contributed to research in topics: Machine translation & WordNet. The author has an hindex of 38, co-authored 576 publications receiving 6465 citations. Previous affiliations of Pushpak Bhattacharyya include Xerox & IBM.
Papers
More filters
Evaluating the Performance of Back-translation for Low Resource English-Marathi Language Pair: CFILT-IITBombay @ LoResMT 2021
TL;DR: In this paper, the authors discuss the details of the various machine translation (MT) systems that have submitted for the English-Marathi LoResMT task and explore the performance of these NMT systems between English and Marathi languages.
Proceedings Article
Lexical Resources for Semantics Extraction
TL;DR: On the challenging problem of generating interlingua from domain and structure unrestricted English sentences, it is able to demonstrate that the use of these lexical resources makes a difference in terms of accuracy figures.
Proceedings Article
Synset Ranking of Hindi WordNet
Sudha Bhingardive,Rajita Shukla,Jaya Saraswati,Laxmi Kashyap,Dhirendra Singh,Pushpak Bhattacharyya +5 more
TL;DR: This work is presented on creating the WFS baseline for Hindi language by manually ranking the synsets of Hindi WordNet by using a ranking tool developed where human experts can see the frequency of the word senses in the sense-tagged corpora and have been asked to rank the senses of a word by using this information and also his/her intuition.
Posted Content
Role of Language Relatedness in Multilingual Fine-tuning of Language Models: A Case Study in Indo-Aryan Languages
Tejas I. Dhamecha,V. Rudra Murthy,Samarth Bharadwaj,Karthik Sankaranarayanan,Pushpak Bhattacharyya +4 more
TL;DR: This paper explored the impact of leveraging the relatedness of languages that belong to the same family in NLP models using multilingual fine-tuning and found that careful selection of subset of related languages can significantly improve performance than utilizing all related languages.
Book ChapterDOI
Authorship Attribution Using Capsule-Based Fusion Approach
TL;DR: In this paper, a fusion-based convolutional neural network model was proposed for authorship attribution on Twitter, where three different types of features were extracted from the input tweet samples: capsule, LSTM, and GRU.