Large Language models have demonstrated promising performance in research ideation across scientific domains.Hypothesis development, the process of generating a highly specific declarative statement connecting a research idea with empirical validation, has received relatively less attention.Existing approaches trivially deploy retrieval augmentation and focus only on the quality of the final output ignoring the underlying reasoning process behind ideation.We present HypER (Hypothesis Generation with Explanation and Reasoning), a small language model (SLM) trained for literature-guided reasoning and evidence-based hypothesis generation.HypER is trained in a multi-task setting to discriminate between valid and invalid scientific reasoning chains in presence of controlled distractions.We find that HypER outperforms the base model, distinguishing valid from invalid reasoning chains (+22% average absolute F1), generates better evidence-grounded hypotheses (0.327 vs. 0.305 base model) with high feasibility and impact as judged by human experts (>3.5 on 5-point Likert scale).Resource at . Example of a valid reasoning chainTitle: Evidence suggesting that a chronic disease self-management program can improve health status while reducing hospitalization Abstract: This study evaluated the effectiveness (changes in health behaviors, health status, and health service utilization) of a self-management program for chronic disease ...

Can the rapid advances in code generation, function calling, and data analysis using large language models (LLMs) help automate the search and verification of hypotheses purely from a set of provided datasets? To evaluate this question, we present DiscoveryBench, the first comprehensive benchmark that formalizes the multi-step process of data-driven discovery. The benchmark is designed to systematically assess current model capabilities in discovery tasks and provide a useful resource for improving them. Our benchmark contains 264 tasks collected across 6 diverse domains, such as sociology and engineering, by manually deriving discovery workflows from published papers to approximate the real-world challenges faced by researchers, where each task is defined by a dataset, its metadata, and a discovery goal in natural language. We additionally provide 903 synthetic tasks to conduct controlled evaluations on data-driven workflows that are not covered in the manually collected split. Furthermore, our structured formalism of data-driven discovery enables a facet-based evaluation that provides useful insights into different failure modes. We evaluate several popular LLM-based reasoning frameworks using both open and closed LLMs as baselines on DiscoveryBench and find that even the best system scores only 25%. Our benchmark, thus, illustrates the challenges in autonomous data-driven discovery and serves as a valuable resource for the community to make progress.

Recent reasoning methods (e.g., chain-of-thought) help users understand how language models (LMs) answer a single question, but they do little to reveal the LM’s overall understanding, or “theory,” about the question’s topic, making it still hard to trust the model. Our goal is to materialize such theories - here called microtheories (a linguistic analog of logical microtheories) - as a set of sentences encapsulating an LM’s core knowledge about a topic. These statements systematically work together to entail answers to a set of questions to both engender trust and improve performance. Our approach is to first populate a knowledge store with (model-generated) sentences that entail answers to training questions, and then distill those down to a core microtheory which is concise, general, and non-redundant. We show that, when added to a general corpus (e.g., Wikipedia), microtheories can supply critical information not necessarily present in the corpus, improving both a model’s ability to ground its answers to verifiable knowledge (i.e., show how answers are systematically entailed by documents in the corpus, grounding up to +8% more answers), and the accuracy of those grounded answers (up to +8% absolute). We also show that, in a human evaluation in the medical domain, our distilled microtheories contain a significantly higher concentration of topically critical facts than the non-distilled knowledge store. Finally, we show we can quantify the coverage of a microtheory for a topic (characterized by a dataset) using a notion of p-relevance. Together, these suggest that microtheories are an efficient distillation of an LM’s topic-relevant knowledge, that they can usefully augment existing corpora, and can provide both performance gains and an interpretable, verifiable window into the model’s knowledge of a topic.

Zhouhang Xie, Tushar Khot, Bhavana Dalvi Mishra, Harshit Surana, Julian McAuley, Peter Clark, Bodhisattwa Prasad Majumder. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

The promise of autonomous scientific discovery (ASD) hinges not only on answering questions, but also on knowing which questions to ask. Most recent works in ASD explore the use of large language models (LLMs) in goal-driven settings, relying on human-specified research questions to guide hypothesis generation. However, scientific discovery may be accelerated further by allowing the AI system to drive exploration by its own criteria. The few existing approaches in open-ended ASD select hypotheses based on diversity heuristics or subjective proxies for human interestingness, but the former struggles to meaningfully navigate the typically vast hypothesis space, and the latter suffers from imprecise definitions. This paper presents AutoDiscovery—a method for open-ended ASD that instead drives scientific exploration using Bayesian surprise. Here, we quantify the epistemic shift from the LLM’s prior beliefs about a hypothesis to its posterior beliefs after gathering experimental results. To efficiently explore the space of nested hypotheses, our method employs a Monte Carlo tree search (MCTS) strategy with progressive widening using surprisal as the reward function. We evaluate AutoDiscovery in the setting of data-driven discovery across 21 real-world datasets spanning domains such as biology, economics, finance, and behavioral science. Our results demonstrate that under a fixed budget, AutoDiscovery substantially outperforms competitors by producing 5-29% more discoveries deemed surprising by the LLM. Our human evaluation further reveals that two-thirds of discoveries made by our system are surprising to domain experts as well, suggesting this is an important step towards building open-ended ASD systems.

Nathaniel Weir, Kate Sanders, Orion Weller, Shreya Sharma, Dongwei Jiang, Zhengping Jiang, Bhavana Dalvi Mishra, Oyvind Tafjord, Peter Jansen, Peter Clark, Benjamin Van Durme. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Large language models (LLMs) have recently been used for sequential decision making in interactive environments. However, leveraging environment reward signals for continual LLM actor improvement is not straightforward. We propose Skill Set Optimization (SSO) for improving LLM actor performance through constructing and refining sets of transferable skills. SSO constructs skills by extracting common subtrajectories with high rewards and generating subgoals and instructions to represent each skill. These skills are provided to the LLM actor in-context to reinforce behaviors with high rewards. Then, SSO further refines the skill set by pruning skills that do not continue to result in high rewards. We evaluate our method in the classic videogame NetHack and the text environment ScienceWorld to demonstrate SSO's ability to optimize a set of skills and perform in-context policy improvement. SSO outperforms baselines by 40% in our custom NetHack task and outperforms the previous state-of-the-art in ScienceWorld by 35%.

Automated scientific discovery promises to accelerate progress across scientific domains, but evaluating an agent's capacity for end-to-end scientific reasoning is challenging as running real-world experiments is often prohibitively expensive or infeasible. In this work we introduce DiscoveryWorld, a virtual environment that enables benchmarking an agent's ability to perform complete cycles of novel scientific discovery in an inexpensive, simulated, multi-modal, long-horizon, and fictional setting.DiscoveryWorld consists of 24 scientific tasks across three levels of difficulty, each with parametric variations that provide new discoveries for agents to make across runs. Tasks require an agent to form hypotheses, design and run experiments, analyze results, and act on conclusions. Task difficulties are normed to range from straightforward to challenging for human scientists with advanced degrees. DiscoveryWorld further provides three automatic metrics for evaluating performance, including: (1) binary task completion, (2) fine-grained report cards detailing procedural scoring of task-relevant actions, and (3) the accuracy of discovered explanatory knowledge.While simulated environments such as DiscoveryWorld are low-fidelity compared to the real world, we find that strong baseline agents struggle on most DiscoveryWorld tasks, highlighting the utility of using simulated environments as proxy tasks for near-term development of scientific discovery competency in agents.

When people think of everyday things like an egg, they typically have a mental image associated with it.This allows them to correctly judge, for example, that "the yolk surrounds the shell" is a false statement.Do language models similarly have a coherent picture of such everyday things?To investigate this, we propose a benchmark dataset consisting of 100 everyday things, their parts, and the relationships between these parts, expressed as 11,720 "X relation Y?" true/false questions.Using these questions as probes, we observe that state-ofthe-art pre-trained language models (LMs) like GPT-3 and Macaw have fragments of knowledge about these everyday things, but do not have fully coherent "parts mental models" (54-59% accurate, 19-43% conditional constraint violation).We propose an extension where we add a constraint satisfaction layer on top of the LM's raw predictions to apply commonsense constraints.As well as removing inconsistencies, we find that this also significantly improves accuracy (by 16-20%), suggesting how the incoherence of the LM's pictures of everyday things can be significantly reduced. 1

Our goal is a teachable reasoning system for question-answering (QA), where a user can interact with faithful answer explanations, and correct its errors so that the system improves over time.Our approach is to augment a QA model with a dynamic memory of user feedback, containing user-supplied corrections to erroneous model beliefs that users identify during interaction.Retrievals from memory are used as additional context for QA, to help avoid previous mistakes in similar new situationsa novel application of memory-based continuous learning.With simulated feedback, we find that our system (called TeachMe 1 ) continually improves with time, and without model retraining, requiring feedback on only 25% of training examples to reach within 1% of the upper-bound (feedback on all examples).Similarly, in experiments with real users, we observe a similar trend, with performance improving by over 15% on a hidden test set after teaching.This suggests new opportunities for using frozen language models in an interactive setting where users can inspect, debug, and correct the model's beliefs, leading to improved system's performance over time.

Our goal is a question-answering (QA) system that can show how its answers are implied by its own internal beliefs via a systematic chain of reasoning.Such a capability would allow better understanding of why a model produced the answer it did.Our approach is to recursively combine a trained backward-chaining model, capable of generating a set of premises entailing an answer hypothesis, with a verifier that checks that the model itself believes those premises (and the entailment itself) through self-querying.To our knowledge, this is the first system to generate multistep chains that are both faithful (the answer follows from the reasoning) and truthful (the chain reflects the system's own internal beliefs).In evaluation using two different datasets, users judge that a majority (70%+) of generated chains clearly show how an answer follows from a set of facts -substantially better than a high-performance baseline -while preserving answer accuracy.By materializing model beliefs that systematically support an answer, new opportunities arise for understanding the model's system of belief, and diagnosing and correcting its misunderstandings when an answer is wrong.

When people answer questions about a specific situation, e.g., "I cheated on my mid-term exam last week.Was that wrong?", cognitive science suggests that they form a mental picture of that situation before answering.While we do not know how language models (LMs) answer such questions, we conjecture that they may answer more accurately if they are also provided with additional details about the question situation, elaborating the "scene".To test this conjecture, we train a new model, DREAM, to answer questions that elaborate the scenes that situated questions are about, and then provide those elaborations as additional context to a question-answering (QA) model.We find that DREAM is able to create better scene elaborations (more accurate, useful, and consistent) than a representative state-of-the-art, zero-shot model (Macaw).We also find that using the scene elaborations as additional context improves the answer accuracy of a downstream QA system, including beyond that obtainable by simply further fine-tuning the QA system on DREAM's training data.These results suggest that adding focused elaborations about a situation can improve a system's reasoning about it, and may serve as an effective way of injecting new scenario-based knowledge into QA models.Finally, our approach is dataset-neutral; we observe improved QA performance across different models, with even bigger gains on models with fewer parameters. 1

Bhavana Dalvi, Peter Jansen, Oyvind Tafjord, Zhengnan Xie, Hannah Smith, Leighanna Pipatanangkura, Peter Clark. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. 2021.

Niket Tandon, Keisuke Sakaguchi, Bhavana Dalvi, Dheeraj Rajagopal, Peter Clark, Michal Guerquin, Kyle Richardson, Eduard Hovy. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2020.

Arman Cohan, Iz Beltagy, Daniel King, Bhavana Dalvi, Dan Weld. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Bhavana Dalvi, Niket Tandon, Antoine Bosselut, Wen-tau Yih, Peter Clark. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Niket Tandon, Bhavana Dalvi, Keisuke Sakaguchi, Peter Clark, Antoine Bosselut. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Comprehending procedural text, e.g., a paragraph describing photosynthesis, requires modeling actions and the state changes they produce, so that questions about entities at different timepoints can be answered.Although several recent systems have shown impressive progress in this task, their predictions can be globally inconsistent or highly improbable.In this paper, we show how the predicted effects of actions in the context of a paragraph can be improved in two ways: (1) by incorporating global, commonsense constraints (e.g., a non-existent entity cannot be destroyed), and (2) by biasing reading with preferences from large-scale corpora (e.g., trees rarely move).Unlike earlier methods, we treat the problem as a neural structured prediction task, allowing hard and soft constraints to steer the model away from unlikely predictions.We show that the new model significantly outperforms earlier systems on a benchmark dataset for procedural text comprehension (+8% relative gain), and that it also avoids some of the nonsensical predictions that earlier systems make.

Bhavana Dalvi, Lifu Huang, Niket Tandon, Wen-tau Yih, Peter Clark. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

Dongyeop Kang, Waleed Ammar, Bhavana Dalvi, Madeleine van Zuylen, Sebastian Kohlmeier, Eduard Hovy, Roy Schwartz. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

Our goal is to construct a domain-targeted, high precision knowledge base (KB), containing general (subject,predicate,object) statements about the world, in support of a downstream question-answering (QA) application. Despite recent advances in information extraction (IE) techniques, no suitable resource for our task already exists; existing resources are either too noisy, too named-entity centric, or too incomplete, and typically have not been constructed with a clear scope or purpose. To address these, we have created a domain-targeted, high precision knowledge extraction pipeline, leveraging Open IE, crowdsourcing, and a novel canonical schema learning algorithm (called CASI), that produces high precision knowledge targeted to a particular domain - in our case, elementary science. To measure the KB’s coverage of the target domain’s knowledge (its “comprehensiveness” with respect to science) we measure recall with respect to an independent corpus of domain text, and show that our pipeline produces output with over 80% precision and 23% recall with respect to that target, a substantially higher coverage of tuple-expressible science knowledge than other comparable resources. We have made the KB publicly available.

Traffic engineering (TE) has long been used by network providers to reduce network congestion and improve resource utilization. Due to its significance, several traffic engineering algorithms have been proposed in literature. However, most of these algorithms optimize maximum link utilization (MLU) in network, and/or assume that network has the capability to route demands on arbitrary paths. Optimizing only for MLU can result in longer route computations to save bandwidth along shorter paths, thereby hurting application performance (as shown by recent research). Further, minimizing MLU can lead to solutions where several links have utilization close to MLU, while many others are under- utilized. Besides, as large fraction of today's Internet uses OSPF routing protocol, it cannot benefit from TE algorithms assuming arbitrary routing capabilities. To address these problems, we present Wise-OSPF (WOSPF), a traffic engineering solution for OSPF networks. WOSPF formulates TE as an optimization problem. The objective of WOSPF is to minimize the difference between the maximum and minimum link utilizations across the network, which leads to more uniform traffic distribution compared to optimizing MLU. As WOSPF uses OSPF for routing demands, it does not compute unnecessarily long routes and can be employed in legacy OSPF networks with minimal changes. Our results show that WOSPF reduces standard deviation of link utilizations in network by 31.35% compared to an optimal MLU based TE approach, while achieving an MLU within 1.9% of the optimal.

In an entity classification task, topic or concept hierarchies are often incomplete. Previous work by Dalvi et al. [12] has showed that in non-hierarchical semi-supervised classification tasks, the presence of such unanticipated classes can cause semantic drift for seeded classes. The Exploratory learning [12] method was proposed to solve this problem; however it is limited to the flat classification task. This paper builds such exploratory learning methods for hierarchical classification tasks.

Whereas people learn many different types of knowledge from diverse experiences over many years, most current machine learning systems acquire just a single function or data model from just a single data set. We propose a never-ending learning paradigm for machine learning, to better reflect the more ambitious and encompassing type of learning performed by humans. As a case study, we describe the Never-Ending Language Learner (NELL), which achieves some of the desired properties of a never-ending learner, and we discuss lessons learned. NELL has been learning to read the web 24 hours/day since January 2010, and so far has acquired a knowledge base with over 80 million confidence-weighted beliefs (e.g., servedWith(tea, biscuits)). NELL has also learned millions of features and parameters that enable it to read these beliefs from the web. Additionally, it has learned to reason over these beliefs to infer new beliefs, and is able to extend its ontology by synthesizing new relational predicates. NELL can be tracked online at http://rtw.ml.cmu.edu, and followed on Twitter at @CMUNELL.

While there has been much research on automatically constructing structured Knowledge Bases (KBs), most of it has focused on generating facts to populate a KB. However, a useful KB must go beyond facts. For example, glosses (short natural language definitions) have been found to be very useful in tasks such as Word Sense Disambiguation. However, the important problem of Automatic Gloss Finding, i.e., assigning glosses to entities in an initially gloss-free KB, is relatively unexplored. We address that gap in this paper. In particular, we propose GLOFIN, a hierarchical semi-supervised learning algorithm for this problem which makes effective use of limited amounts of supervision and available ontological constraints. To the best of our knowledge, GLOFIN is the first system for this task. Through extensive experiments on real-world datasets, we demonstrate GLOFIN's effectiveness. It is encouraging to see that GLOFIN outperforms other state-of-the-art SSL algorithms, especially in low supervision settings. We also demonstrate GLOFIN's robustness to noise through experiments on a wide variety of KBs, ranging from user contributed (e.g., Freebase) to automatically constructed (e.g., NELL). To facilitate further research in this area, we have made the datasets and code used in this paper publicly available.

In this paper, we propose a single low-dimensional representation for entities found in different datasets on the web. Our proposed PIC-D embeddings can represent large D-partite graphs using small number of dimensions enabling fast similarity queries. Our experiments show that this representation can be constructed in small amount of time (linear in number of dimensions). We demonstrate how it can be used for variety of similarity queries like set expansion, automatic set instance acquisition, and column classification. Our approach results in comparable precision with respect to task specific baselines and up to two orders of magnitude improvement in terms of query response time.

We describe a open-domain information extraction method for extracting concept-instance pairs from an HTML corpus. Most earlier approaches to this problem rely on combining clusters of distributionally similar terms and concept-instance pairs obtained with Hearst patterns. In contrast, our method relies on a novel approach for clustering terms found in HTML tables, and then assigning concept names to these clusters using Hearst patterns. The method can be efficiently applied to a large corpus, and experimental results on several datasets show that our method can accurately extract large numbers of concept-instance pairs.

Keyword search on graph structured data has attracted a lot of attention in recent years. Graphs are a natural "lowest common denominator" representation which can combine relational, XML and HTML data. Responses to keyword queries are usually modeled as trees that connect nodes matching the keywords. In this paper we address the problem of keyword search on graphs that may be significantly larger than memory. We propose a graph representation technique that combines a condensed version of the graph (the "supernode graph") which is always memory resident, along with whatever parts of the detailed graph are in a cache, to form a multi-granular graph representation. We propose two alternative approaches which extend existing search algorithms to exploit multigranular graphs; both approaches attempt to minimize IO by directing search towards areas of the graph that are likely to give good results. We compare our algorithms with a virtual memory approach on several real data sets. Our experimental results show significant benefits in terms of reduction in IO due to our algorithms.