activity
20162026
most citedPseudo-Relevance Feedback for Multiple Representation Dense Retrieval

57 citations · 150 across the 37 of their papers we have counts for

collaborators
Showing cs.IRShow all

32 papers · 1 filter

cs.IR2026

Explaining When PRF Fails: Participatory Auditing for Selective Query Expansion

Zeyan Liang, Graham McDonald, Iadh Ounis

Pseudo-Relevance Feedback (PRF) improves retrieval effectiveness on average, but harms a substantial fraction of queries through query drift, an asymmetry hidden by aggregate offli…

cs.IR2026

Interpretable Uncertainty for Adaptive Retrieval and Reasoning in Question Answering

Ritajit Dey, Iadh Ounis, Graham McDonald

Large language models (LLMs) achieve a strong performance in question answering (QA), but remain prone to hallucinations and suffer from limited transparency. Retrieval-augmented g…

cs.IR2026

URecJPQ: Memory-efficient Multimodal Recommendation Models through RecJPQ in Large-Scale Scenarios

Giuseppe Spillo, Zixuan Yi, Aleksandr Petrov +3

Training state-of-the-art recommendation models on large-scale industrial datasets can be a challenging task due to the high number of users and items which are typically represent…

cs.IR2026

Temporal Fact Conflicts in LLMs: Reproducibility Insights from Unifying DYNAMICQA and MULAN

Ritajit Dey, Iadh Ounis, Graham McDonald +1

Large Language Models (LLMs) often struggle with temporal fact conflicts due to outdated or evolving information in their training data. Two recent studies with accompanying datase…

cs.IR2026

LURE-RAG: Lightweight Utility-driven Reranking for Efficient RAG

Manish Chandra, Debasis Ganguly, Iadh Ounis

Most conventional Retrieval-Augmented Generation (RAG) pipelines rely on relevance-based retrieval, which often misaligns with utility -- that is, whether the retrieved passages ac…

cs.IR2025

Document Similarity Enhanced IPS Estimation for Unbiased Learning to Rank

Zeyan Liang, Graham McDonald, Iadh Ounis

Learning to Rank (LTR) models learn from historical user interactions, such as user clicks. However, there is an inherent bias in the clicks of users due to position bias, i.e., us…