Yun He, Wenzhe Li, Hejia Zhang, Songlin Li, Karishma Mandyam, Sopan Khosla, Yuanhao Xiong, Nanshu Wang, Xiaoliang Peng, Beibin Li, Shengjie Bi, Shishir G Patil, Qi Qi, Shengyu Feng, Julian Katz-Samuels, Richard Yuanzhe Pang, Sujan Kumar Gonugondla, Hunter Lang, Yue Yu, Yundi Qian, Maryam Fazel-Zarandi, Licheng Yu, Amine Benhalloum, Hany Hassan Awadalla, Manaal Faruqui. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Mintong Kang, Vinayshekhar Bannihatti Kumar, Shamik Roy, Abhishek Kumar, Sopan Khosla, Balakrishnan Murali Narayanaswamy, Rashmi Gangadharaiah. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Scaling test-time compute to search for optimal solutions is an important step towards building generally-capable language models that can reason. Recent work, however, shows that tasks of varying complexity require distinct search strategies to solve optimally, thus making it challenging to design a one-size-fits-all approach. Prior solutions either attempt to predict task difficulty to select the optimal search strategy, often infeasible in practice, or use a static, pre-defined strategy, e.g., repeated parallel sampling or greedy sequential search, which is sub-optimal. In this work, we argue for an alternative view using the probabilistic framework of Bayesian optimization (BO), where the search strategy is adapted dynamically based on the evolving uncertainty estimates of solutions as search progresses. To this end, we introduce Bayesian-OPRO (BOPRO)––a generalization of a recent method for in-context optimization, which iteratively samples from new proposal distributions by modifying the prompt to the LLM with a subset of its previous generations selected to explore or exploit different parts of the search space. We evaluate our method on word search, molecule optimization, and a joint hypothesis+program search task using a 1-D version of the challenging Abstraction and Reasoning Corpus (1D-ARC). Our results show that BOPRO outperforms all baselines in word search (≥10 points) and molecule optimization (higher quality and 17% fewer invalid molecules), but trails a best-k prompting strategy in program search. Our analysis reveals that despite the ability to balance exploration and exploitation using BOPRO, failure is likely due to the inability of code representation models in distinguishing sequences with low edit-distances.

Ritam Dutt, Sopan Khosla, Vinayshekhar Bannihatti Kumar, Rashmi Gangadharaiah. Proceedings of the 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

The aim of the Universal Anaphora initiative is to push forward the state of the art in anaphora and anaphora resolution by expanding the aspects of anaphoric interpretation which are or can be reliably annotated in anaphoric corpora, producing unified standards to annotate and encode these annotations, deliver datasets encoded according to these standards, and developing methods for evaluating models carrying out this type of interpretation. Such expansion of the scope of anaphora resolution requires a comparable expansion of the scope of the scorers used to evaluate this work. In this paper, we introduce an extended version of the Reference Coreference Scorer (Pradhan et al., 2014) that can be used to evaluate the extended range of anaphoric interpretation included in the current Universal Anaphora proposal. The UA scorer supports the evaluation of identity anaphora resolution and of bridging reference resolution, for which scorers already existed but not integrated in a single package. It also supports the evaluation of split antecedent anaphora and discourse deixis, for which no tools existed. The proposed approach to the evaluation of split antecedent anaphora is entirely novel; the proposed approach to the evaluation of discourse deixis leverages the encoding of discourse deixis proposed in Universal Anaphora to enable the use for discourse deixis of the same metrics already used for identity anaphora. The scorer was tested in the recent CODI-CRAC 2021 Shared Task on Anaphora Resolution in Dialogues.

Kundan Krishna, Sopan Khosla, Jeffrey Bigham, Zachary C. Lipton. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Recent work on entity coreference resolution (CR) follows current trends in Deep Learning applied to embeddings and relatively simple task-related features.SOTA models do not make use of hierarchical representations of discourse structure.In this work, we leverage automatically constructed discourse parse trees within a neural approach and demonstrate a significant improvement on two benchmark entity coreference-resolution datasets.We explore how the impact varies depending upon the type of mention.

Information extraction from conversational data is particularly challenging because the task-centric nature of conversation allows for effective communication of implicit information by humans, but is challenging for machines.The challenges may differ between utterances depending on the role of the speaker within the conversation, especially when relevant expertise is distributed asymmetrically across roles.Further, the challenges may also increase over the conversation as more shared context is built up through information communicated implicitly earlier in the dialogue.In this paper, we propose the novel modeling approach MEDFILTER, which addresses these insights in order to increase performance at identifying and categorizing task-relevant utterances, and in so doing, positively impacts performance at a downstream information extraction task.We evaluate this approach on a corpus of nearly 7,000 doctor-patient conversations where MEDFILTER is used to identify medically relevant contributions to the discussion (achieving a 10% improvement over SOTA baselines in terms of area under the PR curve).Identifying task-relevant utterances benefits downstream medical processing, achieving improvements of 15%, 105%, and 23% respectively for the extraction of symptoms, medications, and complaints.

Learning word representations from large corpora relies on the distributional hypothesis that words present in similar contexts tend to have similar meanings. Recent work has shown that word representations learnt in this manner lack sentiment information which, can be introduced using external knowledge. Our work addresses the question: Can affect lexica improve word representations learnt from a corpus ? In this work, we propose techniques to incorporate affect lexica, which capture fine-grained information about a word's psycholinguistic and emotion orientation, into the training for Word2Vec SkipGram, Word2Vec CBOW, and GloVe using a Joint Learning approach. We use affect scores from the Warriner's affect lexicon to regularize the vector representations learnt from an unlabeled corpus. Our proposed method outperforms previously methods on standard tasks for word similarity detection, outlier detection, and sentiment analysis. We also show the usefulness of our approach for the prediction of formality, frustration, and politeness in text.

Human communication includes information, opinions and reactions. Reactions are often captured by the affective-messages in written as well as verbal communications. While there has been work in affect modeling and to some extent affective content generation, the area of affective word distributions is not well studied. Synsets and lexica capture semantic relationships across words. These models, however, lack in encoding affective or emotional word interpretations. Our proposed model, Aff2Vec, provides a method for enriched word embeddings that are representative of affective interpretations of words. Aff2Vec outperforms the state-of-the-art in intrinsic word-similarity tasks. Further, the use of Aff2Vec representations outperforms baseline embeddings in downstream natural language understanding tasks including sentiment analysis, personality detection, and frustration prediction.