Zijian Chen, Xueguang Ma, Shengyao Zhuang, Ping Nie, Kai Zou, Sahel Sharifymoghaddam, Andrew Liu, Joshua Green, Kshama Patel, Ruoxi Meng, Mingyi Su, Yanxi Li, Haoran Hong, Xinyu Shi, Xuye Liu, Hosna Oyarhoseini, Nandan Thakur, Crystina Zhang, Luyu Gao, Wenhu Chen, Jimmy Lin. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Traditional short keyword queries are increasingly being replaced by longer, more detailed queries that reflect complex and nuanced user information needs, especially in conversational assistants equipped with web search capabilities. In this work, we present a methodology for automatically generating such human-style long-form queries (narratives) by clustering raw short queries to form synthetic search sessions, designed to reflect a real user's search behavior. Based on our human interpretation study of a 50-narrative set (comprising both human-written and automated narratives), 44% of the automated narratives are misidentified as human-written, underscoring not only the realism and complexity of the generated content but also its indistinguishability from authentic human narratives. Furthermore, we share a collection of automated narratives as a testbed for evaluating LLMs on long-form question answering (QA), which was used in the TREC 2025 RAG track. Our code is available at https://github.com/castorini/narrative-generation.

Nandan Thakur, Suleman Kazi, Ge Luo, Jimmy Lin, Amin Ahmad. 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.

We introduce FreshStack, a holistic framework for automatically building information retrieval (IR) evaluation benchmarks by incorporating challenging questions and answers. FreshStack conducts the following steps:(1) automatic corpus collection from code and technical documentation,(2) nugget generation from community-asked questions and answers, and(3) nugget-level support, retrieving documents using a fusion of retrieval techniques and hybrid architectures.We use FreshStack to build five datasets on fast-growing, recent, and niche domains to ensure the tasks are sufficiently challenging. On FreshStack, existing retrieval models, when applied out-of-the-box, significantly underperform oracle approaches on all five domains, denoting plenty of headroom to improve IR quality. In addition, we identify cases where rerankers do not improve first-stage retrieval accuracy (two out of five domains) and oracle context helps an LLM generator generate a high-quality RAG answer.We hope FreshStack will facilitate future work toward constructing realistic, scalable, and uncontaminated IR and RAG evaluation benchmarks.

Large Language Models (LLMs) have significantly enhanced the capabilities of information access systems, especially with retrieval-augmented generation (RAG). Nevertheless, the evaluation of RAG systems remains a barrier to continued progress, a challenge we tackle in this work by proposing an automatic evaluation framework that is validated against human annotations. We believe that the nugget evaluation methodology provides a solid foundation for evaluating RAG systems. This approach, originally developed for the TREC Question Answering (QA) Track in 2003, evaluates systems based on atomic facts that should be present in good answers. Our efforts focus on ''refactoring'' this methodology, where we describe the AutoNuggetizer framework that specifically applies LLMs to both automatically create nuggets and automatically assign nuggets to system answers. In the context of the TREC 2024 RAG Track, we calibrate a fully automatic approach against strategies where nuggets are created manually or semi-manually by human assessors and then assigned manually to system answers. Based on results from a community-wide evaluation, we observe strong agreement at the run level between scores derived from fully automatic nugget evaluation and human-based variants. The agreement is stronger when individual framework components such as nugget assignment are automated independently. This suggests that our evaluation framework provides tradeoffs between effort and quality that can be used to guide the development of future RAG systems. However, further research is necessary to refine our approach, particularly in establishing robust per-topic agreement to diagnose system failures effectively.

Retrieval-augmented generation (RAG) enables large language models (LLMs) to generate answers with citations from source documents containing ''ground truth''. A crucial factor in RAG evaluation is ''support'', or whether the information in the cited documents supports the answer. We conducted a comparative study of submissions to the TREC 2024 RAG Track, evaluating an automatic LLM judge (GPT-4o) against human judges for support assessment. We considered two conditions: (1) fully manual assessments from scratch and (2) manual assessments with post-editing of LLM predictions. Our results indicate good agreement between human and GPT-4o predictions. Further analysis of the disagreements shows that an independent human judge correlates better with GPT-4o than a human judge, suggesting that LLM judges can be a reliable alternative for support assessment. We provide a qualitative analysis of human and GPT-4o errors to help guide future evaluations.

Nandan Thakur, Jianmo Ni, Gustavo Hernandez Abrego, John Wieting, Jimmy Lin, Daniel Cer. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

BEIR is a benchmark dataset originally designed for zero-shot evaluation of retrieval models across 18 different domain/task combinations. In recent years, we have witnessed the growing popularity of models based on representation learning, which naturally begs the question: How effective are these models when presented with queries and documents that differ from the training data? While BEIR was designed to answer this question, our work addresses two shortcomings that prevent the benchmark from achieving its full potential: First, the sophistication of modern neural methods and the complexity of current software infrastructure create barriers to entry for newcomers. To this end, we provide reproducible reference implementations that cover learned dense and sparse models. Second, comparisons on BEIR are performed by reducing scores from heterogeneous datasets into a single average that is difficult to interpret. To remedy this, we present meta-analyses focusing on effect sizes across datasets that are able to accurately quantify model differences. By addressing both shortcomings, our work facilitates future explorations in a range of interesting research questions.

The zero-shot effectiveness of neural retrieval models is often evaluated on the BEIR benchmark---a combination of different IR evaluation datasets. Interestingly, previous studies found that particularly on the BEIR~subset Touché 2020, an argument retrieval task, neural retrieval models are considerably less effective than BM25. Still, so far, no further investigation has been conducted on what makes argument retrieval so "special''. To more deeply analyze the respective potential limits of neural retrieval models, we run a reproducibility study on the Touché 2020 data. In our study, we focus on two experiments: (i) a black-box evaluation (i.e., no model retraining), incorporating a theoretical exploration using retrieval axioms, and (ii) a data denoising evaluation involving post-hoc relevance judgments. Our black-box evaluation reveals an inherent bias of neural models towards retrieving short passages from the Touché 2020 data, and we also find that quite a few of the neural models' results are unjudged in the Touché 2020 data. As many of the short Touché passages are not argumentative and thus non-relevant per se, and as the missing judgments complicate fair comparison, we denoise the Touché 2020 data by excluding very short passages (less than 20 words) and by augmenting the unjudged data with post-hoc judgments following the Touché guidelines. On the denoised data, the effectiveness of the neural models improves by up to 0.52 in nDCG@10, but BM25 is still more effective. Our code and the augmented Touché 2020 dataset are available at https://github.com/castorini/touche-error-analysis.

Traditionally, sparse retrieval systems relied on lexical representations to retrieve documents, such as BM25, dominated information retrieval tasks. With the onset of pre-trained transformer models such as BERT, neural sparse retrieval has led to a new paradigm within retrieval. Despite the success, there has been limited software supporting different sparse retrievers running in a unified, common environment. This hinders practitioners from fairly comparing different sparse models and obtaining realistic evaluation results. Another missing piece is, that a majority of prior work evaluates sparse retrieval models on in-domain retrieval, i.e. on a single dataset: MS MARCO. However, a key requirement in practical retrieval systems requires models that can generalize well to unseen out-of-domain, i.e. zero-shot retrieval tasks. In this work, we provide SPRINT, a unified python toolkit based on Pyserini and Lucene, supporting a common interface for evaluating neural sparse retrieval. The toolkit currently includes five built-in models: uniCOIL, DeepImpact, SPARTA, TILDEv2 and SPLADEv2. Users can also easily add customized models by defining their term weighting method. Using our toolkit, we establish strong and reproducible zero-shot sparse retrieval baselines across the well-acknowledged benchmark, BEIR. Our results demonstrate that SPLADEv2 achieves the best average score of 0.470 [email protected] on BEIR amongst all neural sparse retrievers. In this work, we further uncover the reasons behind its performance gain. We show that SPLADEv2 produces sparse representations with a majority of tokens outside of the original query and document which is often crucial for its performance gains, i.e. a limitation among its other sparse counterparts. We provide our SPRINT toolkit, models, and data used in our experiments publicly here: https://github.com/thakur-nandan/sprint.

Abstract MIRACL is a multilingual dataset for ad hoc retrieval across 18 languages that collectively encompass over three billion native speakers around the world. This resource is designed to support monolingual retrieval tasks, where the queries and the corpora are in the same language. In total, we have gathered over 726k high-quality relevance judgments for 78k queries over Wikipedia in these languages, where all annotations have been performed by native speakers hired by our team. MIRACL covers languages that are both typologically close as well as distant from 10 language families and 13 sub-families, associated with varying amounts of publicly available resources. Extensive automatic heuristic verification and manual assessments were performed during the annotation process to control data quality. In total, MIRACL represents an investment of around five person-years of human annotator effort. Our goal is to spur research on improving retrieval across a continuum of languages, thus enhancing information access capabilities for diverse populations around the world, particularly those that have traditionally been underserved. MIRACL is available at http://miracl.ai/.

Kexin Wang, Nandan Thakur, Nils Reimers, Iryna Gurevych. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Nandan Thakur, Nils Reimers, Johannes Daxenberger, Iryna Gurevych. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.