Papers by Partha Talukdar

TwiRGCN: Temporally Weighted Graph Convolution for Question Answering over Temporal Knowledge Graphs (2023.eacl-main)

Challenge:	Existing methods for generating Open Knowledge Graphs have been criticized for not achieving canonicalization of OpenKGs.
Approach:	They propose to use Open Information Extraction methods to extract triples from text . they propose to learn embeddings of NPs and RPs present in the graph .
Outcome:	The proposed methods improve OpenKG embeddings and bootstrap OpenKGs from text corpus.

Copied to clipboard

Aditya Sharma, Apoorv Saxena, Chitrank Gupta, Mehran Kazemi, Partha Talukdar, Soumen Chakrabarti

Overlap-based Vocabulary Generation Improves Cross-lingual Transfer Among Related Languages (2022.acl-long)

Challenge:	Recent years have witnessed interest in Temporal Question Answering over Knowledge Graphs (TKGQA) but these methods are highly engineered and do not automatically discover relevant parts of the KG during multi-hop reasoning.
Approach:	They propose a scheme to modulate the messages passed through a KG edge during convolution based on the relevance of its associated period to the question.
Outcome:	The proposed system outperforms state-of-the-art models on a recent challenging dataset for multi-hop complex temporal QA called TimeQuestions.

Copied to clipboard

Vaidehi Patil, Partha Talukdar, Sunita Sarawagi

Low-Resource NLP

Challenge:	Pre-trained multilingual models have shown great potential for zero-shot cross-lingual transfer to low web-resource languages (LRLs).
Approach:	They propose a vocabulary generation algorithm which enhances lexical overlap across related languages by generating a token that increases the representation of LRLs.
Outcome:	The proposed approach improves cross-lingual transfer accuracy without reducing HRL representation and accuracy.

Syntax-Guided Controlled Generation of Paraphrases (2020.tacl-1)

Copied to clipboard

Ashutosh Kumar, Kabir Ahuja, Raghuram Vadapalli, Partha Talukdar

Text Generation

Challenge:	Recent work has explored the incorporation of complex syntactic-guidance as constraints in the task of controlled text generation.
Approach:	They propose an end-to-end framework for controlled paraphrase generation that incorporates complex syntactic-guidance constraints into the task.
Outcome:	The proposed framework generates syntax-conforming sentences while not compromising on relevance.

AD3: Attentive Deep Document Dater (D18-1)

Copied to clipboard

Swayambhu Nath Ray, Shib Sankar Dasgupta, Partha Talukdar

Information Extraction & Text Mining

Challenge:	Existing methods to predict creation time of documents are based on time-stamp metadata, but none are available.
Approach:	They propose an attention-based neural document dating system which utilizes both context and temporal information in documents in a flexible and principled manner.
Outcome:	The proposed system outperforms neural and non-neural baselines on multiple real-world datasets.

ELDEN: Improved Entity Linking Using Densified Knowledge Graphs (N18-1)

Copied to clipboard

Priya Radhakrishnan, Partha Talukdar, Vasudeva Varma

Salient Span Masking for Temporal Understanding (2023.eacl-main)

Challenge:	Entity Linking (EL) systems aim to automatically map mentions of an entity in text to the corresponding entity in Knowledge Graph (KG).
Approach:	They propose to densify the Knowledge Graph (KG) with co-occurrence statistics and then use the densified KG to train entity embeddings.
Outcome:	The proposed system outperforms state-of-the-art EL systems on benchmark datasets and outperformed state- of-the art systems on sparsely connected entities in the KG.

Copied to clipboard

Jeremy R. Cole, Aditi Chaudhary, Bhuwan Dhingra, Partha Talukdar

When is BERT Multilingual? Isolating Crucial Ingredients for Cross-lingual Transfer (2022.naacl-main)

Challenge:	Salient Span Masking (SSM) has shown to be effective for closed-book question answering . authors of this study found that SSM alone improves performance on temporal tasks .
Approach:	They introduce Temporal Span Masking (TSM) to improve performance on temporal tasks . they find that SSM alone improves the downstream performance by +5.8 points .
Outcome:	The proposed approach improves performance on three temporal tasks by +5.8 points . the additional targeted spans achieved by adding the TSM task are the best .

Copied to clipboard

Ameet Deshpande, Partha Talukdar, Karthik Narasimhan

Evaluating the Diversity, Equity, and Inclusion of NLP Technology: A Case Study for Indian Languages (2023.findings-eacl)

Challenge:	Recent work on multilingual language models has demonstrated their capacity for cross-lingual zero-shot transfer on downstream tasks.
Approach:	They conduct a large-scale empirical study to isolate the effects of various linguistic properties by measuring zero-shot transfer between four different natural languages.
Outcome:	The proposed model exhibits decent cross-lingual zero-shot transfer, with no significant differences in word order and embedding alignment.

Copied to clipboard

Simran Khanuja, Sebastian Ruder, Partha Talukdar

Graph-based Deep Learning in Natural Language Processing (D19-2)

Challenge:	In order for NLP technology to be widely applicable, fair, and useful, it needs to serve a diverse set of speakers across the world’s languages, be equitable, not unduly biased towards any particular language, and be inclusive of all users.
Approach:	They propose to use Gini coefficient to assess NLP across all three dimensions to assess diversity, equity, and inclusion across all languages.
Outcome:	The proposed evaluation paradigm assesses NLP technologies across all three dimensions and identifies the need for regional-specific choices in model building and dataset creation.

Copied to clipboard

Shikhar Vashishth, Naganand Yadati, Partha Talukdar

OKGIT: Open Knowledge Graph Link Prediction with Implicit Types (2021.findings-acl)

Challenge:	This tutorial aims to introduce graph-based deep learning techniques such as Graph Convolutional Networks (GCNs) for Natural Language Processing (NLP)
Approach:	It provides a brief introduction to graph-based deep learning techniques such as Graph Convolutional Networks (GCNs) for Natural Language Processing (NLP).
Outcome:	This tutorial provides a brief introduction to graph-based deep learning techniques such as Graph Convolutional Networks (GCNs) for natural language processing (NLP).

Copied to clipboard

. Chandrahas, Partha Talukdar

Submodular Optimization-based Diverse Paraphrasing and its Effectiveness in Data Augmentation (N19-1)

Challenge:	Open Knowledge Graphs (OpenKGs) are sparse and not directly usable in an end task.
Approach:	They propose a method that bootstraps OpenKGs from a corpus using OpenIE tools.
Outcome:	The proposed method achieves state-of-the-art performance while producing type compatible NPs in the link prediction task.

Copied to clipboard

Ashutosh Kumar, Satwik Bhattamishra, Manik Bhandari, Partha Talukdar

Text Generation

Challenge:	Previous work focused on generating semantically similar paraphrases without considering diversity.
Approach:	They propose a method to obtain highly diverse paraphrases without compromising on paraphrasing quality by using monotone submodular function maximization.
Outcome:	The proposed method is effective on multiple tasks such as intent classification and paraphrase recognition.

Exploiting Language Relatedness for Low Web-Resource Language Model Adaptation: An Indic Languages Study (2021.acl-long)

Copied to clipboard

Yash Khemchandani, Sarvesh Mehtani, Vaidehi Patil, Abhijeet Awasthi, Partha Talukdar, Sunita Sarawagi

Multilinguality

Challenge:	Recent research in multilingual language models (LMs) has demonstrated their ability to effectively handle multiple languages in a single model.
Approach:	They propose to exploit relatedness among languages in a language family to overcome corpora limitations of LRLs.
Outcome:	The proposed model exploits relatedness among languages in a language family to overcome corpora limitations for low web-resource languages.

IndicGenBench: A Multilingual Benchmark to Evaluate Generation Capabilities of LLMs on Indic Languages (2024.acl-long)

Copied to clipboard

Harman Singh, Nitish Gupta, Shikhar Bharadwaj, Dinesh Tewari, Partha Talukdar

Towards Understanding the Geometry of Knowledge Graph Embeddings (P18-1)

Challenge:	IndicGenBench is the largest benchmark for evaluating large language models on user-facing generation tasks across a diverse set of 29 Indic languages .
Approach:	They evaluate large language models on user-facing generation tasks across 29 languages . they use human curation to provide multi-way parallel evaluation data for many under-represented languages a github repository .
Outcome:	IndicGenBench is the largest benchmark for evaluating LLMs on user-facing generation tasks across a diverse set of 29 Indic languages covering 13 scripts and 4 language families.

Copied to clipboard

null Chandrahas, Aditya Sharma, Partha Talukdar

Higher-order Relation Schema Induction using Tensor Factorization with Back-off and Aggregation (P18-1)

Challenge:	Knowledge Graph (KG) embedding has emerged as a very active area of research over the last few years, resulting in the development of several embeddable methods.
Approach:	They propose to use KG embedding methods to represent entities and relations as vectors in a high-dimensional space.
Outcome:	The proposed methods represent entities and relations in KGs as vectors in a high-dimensional space.

Copied to clipboard

Madhav Nimishakavi, Manish Gupta, Partha Talukdar

Self-Influence Guided Data Reweighting for Language Model Pre-training (2023.emnlp-main)

Challenge:	Relation Schema Induction (RSI) is a problem of identifying type signatures of arguments from unlabeled text.
Approach:	They propose a framework for inducing higher-order relation schemata from unlabeled text.
Outcome:	The proposed framework helps in dealing with sparsity and induces higher-order relation schemata.

Copied to clipboard

Megh Thakkar, Tolga Bolukbasi, Sriram Ganapathy, Shikhar Vashishth, Sarath Chandar, Partha Talukdar

Dating Documents using Graph Convolution Networks (P18-1)

Challenge:	Language Models (LMs) pre-trained with selfsupervision on large text data are the default starting point for developing models for various downstream tasks.
Approach:	They propose a method for jointly reweighting samples by leveraging self-influence scores as an indicator of sample importance and pre-training.
Outcome:	The proposed method promotes novelty and stability for model pre-training.

Copied to clipboard

Shikhar Vashishth, Shib Sankar Dasgupta, Swayambhu Nath Ray, Partha Talukdar

Structured Data in NLP

Challenge:	Existing approaches for document dating assume accurate knowledge of document date, but this is not always available for arbitrary documents from the Web.
Approach:	They propose a Graph Convolutional Network (GCN) based document dating approach which exploits syntactic and temporal graph structures of document in a principled way.
Outcome:	The proposed approach outperforms state-of-the-art models on real-world datasets by 19% absolute accuracy points.

A Re-evaluation of Knowledge Graph Completion Methods (2020.acl-main)

Copied to clipboard

Zhiqing Sun, Shikhar Vashishth, Soumya Sanyal, Partha Talukdar, Yiming Yang

Structured Data in NLP

Challenge:	Knowledge Graph Completion (KGC) aims at automatically predicting missing links for large-scale knowledge graphs.
Approach:	They propose a protocol to evaluate KGC methods that is robust to handle bias in the model, which can substantially affect the final results.
Outcome:	The proposed evaluation protocol is robust to handle bias in the model, which can substantially affect the final results.

Incorporating Syntactic and Semantic Information in Word Embeddings using Graph Convolutional Networks (P19-1)

Copied to clipboard

Shikhar Vashishth, Manik Bhandari, Prateek Yadav, Piyush Rai, Chiranjib Bhattacharyya, Partha Talukdar

Representation Learning

Challenge:	Existing word embedding methods utilize sequential context of a word to learn its embeddment, but such methods result in an explosion of the vocabulary size.
Approach:	They propose a flexible Graph Convolution based method for learning word embeddings that utilizes the dependency context of a word without increasing the vocabulary size.
Outcome:	The proposed model outperforms existing methods on intrinsic and extrinsic tasks and provides an advantage when used with ELMo.

MergeDistill: Merging Language Models using Pre-trained Distillation (2021.findings-acl)

Copied to clipboard

Simran Khanuja, Melvin Johnson, Partha Talukdar

Question Answering Over Temporal Knowledge Graphs (2021.acl-long)

Challenge:	Existing pre-trained multilingual language models often lack capacity and skewed data . this leads to inequitable representation of languages due to limited capacity and sub-optimal vocabularies.
Approach:	They propose a framework to merge pre-trained multilingual language models to maximize their assets with minimal dependencies.
Outcome:	The proposed framework outperforms teacher-trained models on multiple datasets and with a fixed model capacity.

Copied to clipboard

Apoorv Saxena, Soumen Chakrabarti, Partha Talukdar

HyTE: Hyperplane-based Temporally aware Knowledge Graph Embedding (D18-1)

Challenge:	Temporal Knowledge Graphs (Temporal KGs) provide temporal scopes (start and end times) on each edge in the Knowledge . Lack of broad coverage datasets has been limiting progress in this area .
Approach:	They propose a transformer-based solution that exploits recent advances in Temporal Knowledge Graph embeddings and achieves an increase of 120% in accuracy over the next best performing method.
Outcome:	The proposed solution improves on the only known dataset by 340x . it increases accuracy by 120% over the baseline solution .

Copied to clipboard

Shib Sankar Dasgupta, Swayambhu Nath Ray, Partha Talukdar

Zero-shot Word Sense Disambiguation using Sense Definition Embeddings (P19-1)

Challenge:	Existing KG embedding methods ignore this temporal dimension while learning embedds of the KG elements.
Approach:	They propose a temporally aware KG embedding method which incorporates time in the entity-relation space by associating each timestamp with a corresponding hyperplane.
Outcome:	The proposed method performs KG inference using temporal guidance and predicts scopes for relational facts with missing time annotations.

Copied to clipboard

Sawan Kumar, Sharmistha Jat, Karan Saxena, Partha Talukdar

Improving Multi-hop Question Answering over Knowledge Graphs using Knowledge Base Embeddings (2020.acl-main)

Challenge:	Word Sense Disambiguation (WSD) is an open problem in Natural Language Processing . current methods treat senses as discrete labels and predict the most-frequent-Sense for unseen senses .
Approach:	They propose a supervised model to perform Word Sense Disambiguation (WSD) by predicting over a continuous sense embedding space rather than a discrete label space.
Outcome:	The proposed model generalizes over seen and unseen senses, achieving zero-shot learning.

Copied to clipboard

Apoorv Saxena, Aditay Tripathi, Partha Talukdar

XTREME-UP: A User-Centric Scarce-Data Benchmark for Under-Represented Languages (2023.findings-emnlp)

Challenge:	Existing multi-hop KGQA methods impose heuristic neighborhood limits, which often make it much harder to answer the input NL question.
Approach:	They propose to use knowledge Graphs (KG) to answer natural language queries over the KG.
Outcome:	The proposed method is particularly effective in performing multi-hop KGQA over sparse KGs.

Copied to clipboard

Sebastian Ruder, Jonathan Clark, Alexander Gutkin, Mihir Kale, Min Ma, Massimo Nicosia, Shruti Rijhwani, Parker Riley, Jean-Michel Sarr, Xinyi Wang, John Wieting, Nitish Gupta, Anna Katanova, Christo Kirov, Dana Dickinson, Brian Roark, Bidisha Samanta, Connie Tao, David Adelani, Vera Axelrod, Isaac Caswell, Colin Cherry, Dan Garrette, Reeve Ingle, Melvin Johnson, Dmitry Panteleev, Partha Talukdar

UGIF-DataSet: A New Dataset for Cross-lingual, Cross-modal Sequential actions on the UI (2024.findings-naacl)

Challenge:	Existing datasets are often informed by established research directions in the NLP community.
Approach:	They propose a benchmark to evaluate the capabilities of language models across 88 under-represented languages over 9 key user-centric technologies including ASR, OCR, MT, and information access tasks.
Outcome:	The proposed benchmark evaluates the capabilities of language models across 88 under-represented languages over 9 key user-centric technologies including ASR, OCR, MT, and information access tasks.

Copied to clipboard

Sagar Gubbi Venkatesh, Partha Talukdar, Srini Narayanan

Multilinguality

Challenge:	Identifying the right help document, understanding instructions from the document, and using them to resolve the issue at hand is challenging.
Approach:	They propose to use help documents to create step-by-step tutorials overlaid on the phone UI to overcome challenges in retrieval, parsing, and grounding in multilingual-multimodal setting.
Outcome:	The proposed dataset contains 4,184 tasks across 8 languages and shows that the end-to-end completion rate drops from 48% in English to 32% for other languages.

Reordering Examples Helps during Priming-based Few-Shot Learning (2021.findings-acl)

Copied to clipboard

Sawan Kumar, Partha Talukdar

Parameter-Efficient Finetuning for Robust Continual Multilingual Learning (2023.findings-acl)

Challenge:	Existing methods for learning from limited data are not efficient . we show that presenting examples in the right order is key for generalization .
Approach:	They propose a method to learn from limited data using examples as prompts . they propose PERO, which uses examples as search over set of permutations .
Outcome:	The proposed method can generalize using as few as 10 examples, the authors show . it can be used on sentiment classification, natural language inference and fact retrieval tasks .

Copied to clipboard

Kartikeya Badola, Shachi Dave, Partha Talukdar

Relating Simple Sentence Representations in Deep Neural Networks and the Brain (P19-1)

Challenge:	Existing approaches to Continual Multilingual Learning (CML) are based on updating models using new data in stages.
Approach:	They propose a parameter-efficient finetuning strategy to increase the number of languages on which the model improves after an update while reducing the magnitude of loss for the remaining languages.
Outcome:	The proposed model improves on the languages included in the latest update while reducing the loss of performance on the remaining languages.

Copied to clipboard

Sharmistha Jat, Hao Tang, Partha Talukdar, Tom Mitchell

Re-contextualizing Fairness in NLP: The Case of India (2022.aacl-main)

Challenge:	Existing deep learning models for natural language processing are not fully studied.
Approach:	They investigate whether deep recurrent models learn sentences against those encoded by the brain and whether there is any correspondence between hidden layers of these models and brain regions when processing sentences.
Outcome:	The proposed models can be used to synthesize brain data and improve subsequent stimuli decoding accuracy.

Copied to clipboard

Shaily Bhatt, Sunipa Dev, Partha Talukdar, Shachi Dave, Vinodkumar Prabhakaran

Few-shot Controllable Style Transfer for Low-Resource Multilingual Settings (2022.acl-long)

Challenge:	Recent research has revealed undesirable biases in NLP data and models . however, these efforts focus of social disparities in the West and are not directly portable to other geo-cultural contexts.
Approach:	They propose a framework to re-contextualize NLP fairness research for the Indian context . they build resources for fairness evaluation in the Indian and delve deeper into social stereotypes for Region and Religion .
Outcome:	The proposed framework can be generalized to other geo-cultural contexts.

Copied to clipboard

Kalpesh Krishna, Deepak Nathani, Xavier Garcia, Bidisha Samanta, Partha Talukdar