Papers by Shashank Srivastava

Pragmatic Reasoning Unlocks Quantifier Semantics for Foundation Models (2023.emnlp-main)

Challenge:	Large Language Models (LLMs) offer natural language explanations as an alternative to feature attribution methods for model interpretability, but they may not reflect the model’s truereasoning faithfully.
Approach:	They propose a testbed framework for evaluating faithfulness metrics for natural language explanations using diagnosticity and model-editing methods.
Outcome:	The proposed framework evaluates faithfulness metrics for natural language explanations on four tasks including fact-checking, analogy, object counting, and multi-hop reasoning.

Copied to clipboard

Yiyuan Li, Rakesh Menon, Sayan Ghosh, Shashank Srivastava

Reasoning

Challenge:	Generalized quantifiers are used to indicate the proportions predicates satisfy (e.g., some apples are red).
Approach:	They propose a framework to model quantifier semantics for textbased foundation models by combining natural language inference and the Rational Speech Acts framework.
Outcome:	The proposed framework shows a 20% improvement over a literal listener baseline in predicting percentage scopes for quantifier comprehension even with no training.

Adversarial Scrubbing of Demographic Information for Text Classification (2021.emnlp-main)

Copied to clipboard

Somnath Basu Roy Chowdhury, Sayan Ghosh, Yiyuan Li, Junier Oliva, Shashank Srivastava, Snigdha Chaturvedi

Robustness in NLP

Challenge:	Existing frameworks to debias contextual representations can encode undesirable attributes, like demographic associations of the users, while being trained for an unrelated task.
Approach:	They propose an adversarial learning framework to debias contextual representations by encoding undesirable attributes while being trained for an unrelated task.
Outcome:	The proposed framework debiases representations on 8 datasets while remaining informative on the target task.

Chameleon LLMs: User Personas Influence Chatbot Personality Shifts (2025.emnlp-main)

Copied to clipboard

Jane Xing, Tianyi Niu, Shashank Srivastava

Dialogue Systems & Conversational Agents

Challenge:	Existing studies have examined whether large language models adapt their perceived personalities in response to user interactions.
Approach:	They propose to use a controlled simulation to measure chatbot personality shifts before and after the interaction to determine whether LLMs exhibit conversational adaptations.
Outcome:	The proposed model exhibits personality adaptations over prolonged interactions, while Emotional Stability and Intellect remain relatively stable.

DISCERN: Decoding Systematic Errors in Natural Language for Text Classifiers (2024.emnlp-main)

Copied to clipboard

Rakesh R Menon, Shashank Srivastava

How Helpful is Inverse Reinforcement Learning for Table-to-Text Generation? (2021.acl-short)

Challenge:	Recent work proposes automatic methods for identifying and explaining systematic biases using keywords.
Approach:	They propose automatic methods for identifying and explaining systematic biases using keywords.
Outcome:	The proposed framework improves classifiers by augmenting training sets with synthetically generated instances or annotated examples via active learning.

Copied to clipboard

Sayan Ghosh, Zheng Qi, Snigdha Chaturvedi, Shashank Srivastava

Text Generation

Challenge:	Existing approaches to Table-to-Text generation suffer from issues such as missing information, repetition and repetition.
Approach:	They propose to use Inverse Reinforcement Learning (IRL) to solve the Table-to-Text task . they use multiple interpretable unsupervised reward components that are combined linearly to form a composite reward function.
Outcome:	The proposed task outperforms strong RL baselines marginally in the Table-to-Text task.

Large Language Models Threaten Language’s Epistemic and Communicative Foundations (2025.emnlp-main)

Copied to clipboard

Shashank Srivastava

Semantic Text Processing

Challenge:	Large language models are reshaping the norms of human communication, sometimes decouping words from genuine human thought.
Approach:	They propose to model humans, LLMs, and texts in a provenance network . they propose to use epistemic doppelgängers to generate texts that are indis- tinguishable from human-authored texts .
Outcome:	The proposed models induce semantic drift, erode account-ability, and obfuscate intent and authorship.

What do Large Language Models Learn beyond Language? (2022.findings-emnlp)

Copied to clipboard

Avinash Madasu, Shashank Srivastava

LIA: A Natural Language Programmable Personal Assistant (D18-2)

Challenge:	Pretraining on text confers models with useful ‘inductive biases’ for non-linguistic reasoning.
Approach:	They investigate whether pre-training on text confers these models with helpful ‘inductive biases’ for non-linguistic reasoning.
Outcome:	The proposed models outperform non-pretrained models on 19 non-linguistic tasks and show that they retain inductive biases even when training on multi-lingual text and computer code.

Copied to clipboard

Igor Labutov, Shashank Srivastava, Tom Mitchell

Multimodality

Challenge:	a prototype of an intelligent personal assistant can be programmed using natural language . a user can instruct her assistants using language similar to how humans teach other humans .
Approach:	They present LIA, an intelligent personal assistant that can be programmed using natural language. LIA resides on a typical mobile Android device.
Outcome:	The proposed system can be programmed using natural language, and it can perceive the external environment through sensors and effectors.

ePiC: Employing Proverbs in Context as a Benchmark for Abstract Language Understanding (2022.acl-long)

Copied to clipboard

Sayan Ghosh, Shashank Srivastava

SocialGaze: Improving the Integration of Human Social Norms in Large Language Models (2024.findings-emnlp)

Challenge:	Large language models have shown exciting progress on several NLP benchmarks . however, evaluating their ability for complex analogical reasoning remains under-explored .
Approach:	They propose a dataset of narratives for employing proverbs in context as a benchmark for abstract language understanding.
Outcome:	The proposed dataset provides fine-grained annotation of aligned spans between proverbs and narratives and contains minimal overlaps between narratives with proverb . the results show that large language models struggle on these tasks compared to humans, and these tasks pose multiple learning challenges.

Copied to clipboard

Anvesh Rao Vijjini, Rakesh R Menon, Jiayi Fu, Shashank Srivastava, Snigdha Chaturvedi

A Spatial Model for Extracting and Visualizing Latent Discourse Structure in Text (P18-1)

Challenge:	Increasingly, large language models (LLMs) are able to understand and rationalize socially acceptable behaviors, but they are often misaligned with human consensus.
Approach:	They propose a multi-step prompting framework that verbalizes a social situation from multiple perspectives before forming a judgment.
Outcome:	The proposed framework improves the alignment with human judgments by up to 11 F1 points with the GPT-3.5 model.

Copied to clipboard

Shashank Srivastava, Nebojsa Jojic

Challenge:	Using sequences of sentences, we show that learning long-range latent discourse structure from large corpora can be useful for machine learning.
Approach:	They propose a probabilistic model of documents as sequences of sentences with a 2- or 3-D spatial grid and embed sentences into a grid.
Outcome:	The proposed model outperforms or is competitive with state-of-the-art generative approaches on tasks such as predicting the outcome of a story, and sentence ordering.

Zero-shot Learning of Classifiers from Natural Language Quantification (P18-1)

Copied to clipboard

Shashank Srivastava, Igor Labutov, Tom Mitchell

Manipulating the Perceived Personality Traits of Language Models (2023.findings-emnlp)

Challenge:	Existing methods to learn concepts from natural language are limited or no labeled examples.
Approach:	They propose a framework through which a set of explanations of a concept can be used to learn a classifier without access to any labeled examples.
Outcome:	The proposed framework outperforms previous approaches for learning with limited data and is comparable with fully supervised classifiers trained from a small number of labeled examples.

Copied to clipboard

Graham Caron, Shashank Srivastava

Beyond Labels: Empowering Human Annotators with Natural Language Explanations through a Novel Active-Learning Architecture (2023.findings-emnlp)

Challenge:	Psychology research has long explored aspects of human personality like extroversion, agreeableness and emotional stability, three of the personality traits that make up the ‘Big Five’.
Approach:	They propose to use text generated from large language models to evaluate perceived personality traits and to frame them as tools for controlling personas in dialog systems.
Outcome:	The proposed models predict personality traits in different contexts and can be manipulated in a predictable way.

Copied to clipboard

Bingsheng Yao, Ishan Jindal, Lucian Popa, Yannis Katsis, Sayan Ghosh, Lihong He, Yuxuan Lu, Shashank Srivastava, Yunyao Li, James Hendler, Dakuo Wang

Predicting Difficulty and Discrimination of Natural Language Questions (2022.acl-short)

Challenge:	Existing low-resource learning techniques focus on label annotation while neglecting the natural language explanation of a data point.
Approach:	They propose a novel architecture that leverages an explanation-generation model to produce explanations guided by human explanations and a prediction model that utilizes generated explanations toward prediction faithfully.
Outcome:	The proposed architecture produces explanations guided by human explanations, a prediction model that utilizes generated explanations toward prediction faithfully, and a data diversity-based AL sampling strategy that benefits from the explanation annotations.

Copied to clipboard

Matthew Byrd, Shashank Srivastava

Challenge:	Item Response Theory (IRT) has been used to numerically characterize question difficulty and discrimination for human subjects in domains including cognitive psychology and education.
Approach:	They explore the relationship between difficulty and discrimination in question-answering contexts by using IRT to characterize item difficulty and item discrimination.
Outcome:	The proposed models can predict difficulty and discrimination parameters for new questions and explain them with features of questions, answers, and associated contexts.

Is Chain-of-Thought Really Not Explainability? Chain-of-Thought Can Be Faithful without Hint Verbalization (2026.acl-long)

Copied to clipboard

Kerem Zaman, Shashank Srivastava

Learning Web-based Procedures by Reasoning over Explanations and Demonstrations in Context (2020.acl-main)

Challenge:	Recent work labels a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction.
Approach:	They propose to use the Biasing Features metric to label a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction.
Outcome:	The proposed metric confuses unfaithfulness with incompleteness, the authors argue . larger inference-time budgets greatly increase hint verbalization, they show .

Copied to clipboard

Shashank Srivastava, Oleksandr Polozov, Nebojsa Jojic, Christopher Meek

Where Have I Heard This Story Before? Identifying Narrative Similarity in Movie Remakes (N18-2)

Challenge:	a new direction for semantic parsing that models explanations to demonstrations is proposed . bottom-up approach to generating logical forms is complicated in domains with rich composition .
Approach:	They propose a new direction for semantic parsing that models explanations in a context . they use inverse semantics to reason backwards from observed demonstrations .
Outcome:	The proposed approach shows better task completion rates than a baseline method . it is competitive with exploration-and-demonstration based methods, but requires no exploration of environment .

Copied to clipboard

Snigdha Chaturvedi, Shashank Srivastava, Dan Roth

Challenge:	Existing methods to identify instances of similar narratives are limited by annotated data.
Approach:	They propose a task for identifying instances of similar narratives from a collection of narrative texts.
Outcome:	The proposed approach yields an 8% absolute improvement over a baseline on a novel dataset of plot summaries of 577 movie remakes from Wikipedia.

Leveraging Multiple Teachers for Test-Time Adaptation of Language-Guided Classifiers (2023.findings-emnlp)

Copied to clipboard

Kangda Wei, Sayan Ghosh, Rakesh Menon, Shashank Srivastava

Classifying Unreliable Narrators with Large Language Models (2025.acl-long)

Challenge:	Recent approaches focus on language-guided classifiers that can generalize in zero-shot settings, but their performance varies significantly between different language explanations in unpredictable ways.
Approach:	They propose a framework that uses data programming to adapt a language-guided classifier for a new task when provided with multiple teachers and unlabeled test examples.
Outcome:	The proposed framework outperforms a baseline from previous work by 9.3%.

Copied to clipboard

Anneliese Brei, Katharine Henry, Abhisheik Sharma, Shashank Srivastava, Snigdha Chaturvedi

Information Retrieval

Challenge:	a recent study identifies unreliable narrators, i.e. those who unintentionally misrepresent information . authors propose using computational methods to identify unredependable narrators . adbrei: readers implicitly question the reliability of the nrator .
Approach:	They propose using computational methods to identify unreliable narrators . they use literary theory to define different types of unredependable narrators .
Outcome:	The proposed method can identify unreliable narrators on real-world text data.

Does Social Pressure Drive Persuasion in Online Fora? (2021.emnlp-main)

Copied to clipboard

Ayush Jain, Shashank Srivastava

Sentiment Analysis

Challenge:	Using social features, we hypothesize that comments from the ambient community can either affirm the original view or implicitly exert pressure to change it.
Approach:	They propose a structured model to capture the ambient community’s sentiment towards the discussion and its effect on persuasion.
Outcome:	The proposed model captures the ambient community’s sentiment towards the discussion and its effect on persuasion.

INTERACT: Enabling Interactive, Question-Driven Learning in Large Language Models (2025.acl-long)

Copied to clipboard

Aum Kendapadi, Kerem Zaman, Rakesh R Menon, Shashank Srivastava

CLUES: A Benchmark for Learning Classifiers using Natural Language Explanations (2022.acl-long)

Challenge:	Large language models (LLMs) absorb static data without the ability to question and refine knowledge.
Approach:	They propose a framework in which a “student” LLM engages a ‘teacher’ LLM through iterative inquiries to acquire knowledge across 1,347 contexts.
Outcome:	The proposed framework achieves up to 25% improvement in 1,347 contexts across a wide range of scenarios and LLM architectures, with ‘cold-start’ student models matching static learning baselines in as few as five dialogue turns.

Copied to clipboard

Rakesh R. Menon, Sayan Ghosh, Shashank Srivastava

Mapping Language to Programs using Multiple Reward Components with Inverse Reinforcement Learning (2021.findings-emnlp)

Challenge:	Supervised learning has traditionally focused on inductive learning by looking at labeled examples of a task.
Approach:	They propose a benchmark for Classifier Learning Using natural language ExplanationS that provides natural language supervision over structured data and entailment-based models that learn from explanations.
Outcome:	The proposed model generalizes 18% better (relative) on novel tasks than a baseline that does not use explanations.

Copied to clipboard

Sayan Ghosh, Shashank Srivastava

Programming Languages in NLP

Challenge:	Existing approaches focus on likelihood-based training or using reinforcement learning to fine-tune models based on a single reward.
Approach:	They propose an approach to fine-tune programs from natural language instruction . they propose a reward function that linearly combines them and a policy for program generation .
Outcome:	The proposed approach achieves better performance than competing methods using Reinforcement Learning.

Explaining Differences Between Model Pairs in Natural Language through Sample Learning (2025.emnlp-main)

Copied to clipboard

Advaith Malladi, Rakesh R Menon, Yuvraj Jain, Shashank Srivastava

MaNtLE: Model-agnostic Natural Language Explainer (2023.emnlp-main)

Challenge:	a framework that generates faithful natural language explanations of when and how two ML models converge or diverge in their predictions requires access to training data.
Approach:	They propose a framework that generates faithful natural language explanations of when and how two ML models converge or diverge in their predictions.
Outcome:	The proposed framework generates faithful natural language explanations of when and how two models diverge in their predictions.

Copied to clipboard

Rakesh Menon, Kerem Zaman, Shashank Srivastava

Improving and Simplifying Pattern Exploiting Training (2021.emnlp-main)

Challenge:	Recent research suggests that practitioners prefer examining language explanations that explain sub-groups of examples.
Approach:	They propose a model-agnostic natural language explainer that generates faithful explanations of classifier rationale for structured classification tasks.
Outcome:	The proposed model-agnostic natural language explainer generates faithful explanations of classifier rationale for structured classification tasks.

Copied to clipboard

Derek Tam, Rakesh R. Menon, Mohit Bansal, Shashank Srivastava, Colin Raffel

Fuse to Forget: Bias Reduction and Selective Memorization through Model Fusion (2024.emnlp-main)

Challenge:	Recent studies have shown that pre-trained language models can learn well when primed with only a few labeled examples.
Approach:	They propose a method that uses task-specific unlabeled data to provide denser supervision during fine-tuning.
Outcome:	The proposed approach outperforms GPT-3 on SuperGLUE without any unlabeled data.

Copied to clipboard

Kerem Zaman, Leshem Choshen, Shashank Srivastava

Learning to Ask for Conversational Machine Learning (D19-1)

Challenge:	a recent study examined the effects of model fusion on learning of shortcuts and social biases in fine-tuned language models.
Approach:	They investigate whether model fusion can be used to reduce unwanted knowledge . they examine classification tasks with artificially fusioned models .
Outcome:	The proposed model fusion can reduce unshared knowledge, the authors show . their results show that the model merged models can improve performance and generalize .

Copied to clipboard

Shashank Srivastava, Igor Labutov, Tom Mitchell

Information Retrieval

Challenge:	Empirical analysis across three domains shows that learned question-asking strategies expedite classifier training by asking appropriate questions at different points in the learning process.
Approach:	They propose a reinforcement learning framework where the learner’s actions correspond to question types and the reward for asking a question is based on how the teacher’s response changes performance of the resulting machine learning model.
Outcome:	The proposed framework outperforms a random policy on learning classification tasks, but the dialog looks contrived from a human perspective.

LaSQuE: Improved Zero-Shot Classification from Explanations Through Quantifier Modeling and Curriculum Learning (2023.findings-acl)

Copied to clipboard

Sayan Ghosh, Rakesh R. Menon, Shashank Srivastava

PRover: Proof Generation for Interpretable Reasoning over Rules (2020.emnlp-main)

Challenge:	Several recent approaches have explored training machine learning models via natural language supervision, but they fail to leverage linguistic quantifiers and mimic humans in compositionally learning complex tasks.
Approach:	They propose a method that can learn zero-shot classifiers from language explanations by using three new strategies: (1) modeling the semantics of linguistic quantifiers in explanations; (2) aggregating information from multiple explanations using an attention-based mechanism; (3) model training via curriculum learning.
Outcome:	The proposed method outperforms previous work showing an absolute gain of up to 7% in generalizing to unseen real-world classification tasks.

Copied to clipboard

Swarnadeep Saha, Sayan Ghosh, Shashank Srivastava, Mohit Bansal