arXiv subjects

Nicola Tonellotto

Publications and source records attributed to Nicola Tonellotto.

At least 19 recordsLinked to original sources

arXiv2026

SURF: Subtractive Updates for Recommender Forgetting

The increasing demand for user privacy and compliance with regulations such as GDPR has made machine unlearning a fundamental requirement for modern recommender systems. However, Sequential Recommender Systems (SRS) pose unique challenges for unlearning due to their reliance on temporal interaction patterns. Existing approaches either require computationally prohibitive full retraining or fail to account for the sequential nature of user behavior. We propose SURF (Subtractive Updates for Recommender Forgetting), a lightweight framework for approximate machine unlearning in SRS. SURF operates in three stages: (i) identifying the neighborhood of the item to forget in the embedding space, (ii) training an auxiliary model on this compact local subset, and (iii) subtracting the auxiliary model's scores from the original model at inference time. Experiments against five baselines on 7 datasets show that SURF achieves unlearning effectiveness comparable to full retraining while substantially reducing computational cost, yielding up to a 32% improvement in NDCG@20 while requiring just 2% of the original retraining baseline time budget. We share our code at https://github.com/FilippoBetello/SURF.

cs.IR↗

arXiv2026

Cross-Document Neural Re-Ranking via Query-Induced Subgraphs

Neural re-rankers typically score query-document pairs independently, neglecting cross-document context within the retrieved candidate set. We propose Graph Neural Re-Ranking (GNRR), a framework that extracts a sparse, query-induced subgraph from a pre-computed semantic corpus graph and applies Graph Neural Networks (GNN) to propagate cross-document signals. Unlike self-attention re-rankers, which scale quadratically with the number of candidates ($\mathcal{O}(K^2)$), GNRR achieves $\mathcal{O}(c \cdot K)$ online complexity, where $c$ is the fixed corpus graph degree and $K$ the candidate set size. We evaluate five GNN operators within this framework and find that architecture choice substantially affects generalization to harder queries: the GCN variant is the only one that consistently improves over TCT-ColBERT across all three TREC benchmarks. On TREC-DLHard, the most challenging evaluation benchmark, GNRR achieves $+5.2\%$ relative AP over TCT-ColBERT and $+9.0\%$ AP over a self-attention re-ranker. Notably, self-attention re-ranking degrades AP on DLHard ($-3.5\%$ versus TCT-ColBERT), suggesting that sparse corpus-graph structure provides a complementary re-ranking signal that dense self-attention fails to capture. Efficiency analysis shows that GNN models require fewer parameters and lower per-query latency at $K=1000$ than self-attention, with linear rather than quadratic scaling in candidate set size. Code to reproduce our experiment is available at https://github.com/difra100/Graph-Neural-Re-Ranking-via-Corpus-Graph

cs.IR↗

arXiv2026

Black-Mamba: Biologically-Inspired Leaky Accumulation for Conceptual Knowledge under Distribution Drift

Forecasting under real-world conditions is inherently non-stationary, as the conditional distribution of future observations evolves over time. Recent test-time adaptive sequence models address this challenge by updating internal states during inference, but tie adaptation to instantaneous prediction errors or surprise. This coupling can conflate persistent distribution shift with stochastic innovations, leading to unnecessary updates and inefficient adaptation. We introduce Black-Mamba, a test-time adaptive forecasting architecture that formulates online adaptation as evidence-gated state tracking under distribution drift. The model augments a base predictor with a dynamic memory updated when temporally accumulated surprisal provides sufficient evidence of a regime change. This turns adaptation into a selective, event-driven process rather than a continuous one. Across multiple forecasting benchmarks with non-stationary dynamics, Black-Mamba achieves competitive or improved predictive performance compared to existing test-time adaptation methods while significantly reducing the number of memory updates during inference. Together with mathematical analysis and biological evidence, these results suggest that accumulated surprisal provides a principled signal for distinguishing persistent drift from transient noise, yielding more efficient and robust adaptation.

cs.AI↗

arXiv2026

Does it Really Count? Assessing Semantic Grounding in Text-Guided Class-Agnostic Counting

Open-world text-guided class-agnostic counting (CAC) has emerged as a flexible paradigm for counting arbitrary object classes by using natural language prompts. However, current evaluation protocols primarily focus on standard counting errors within single-category images, overlooking a fundamental requirement: the ability to correctly ground the textual prompt in the visual scene. In this paper, we show that several state-of-the-art CAC models often struggle to determine which object class should be counted based on the given prompt, revealing a misalignment between textual semantics and visual object representations. This limitation leads to spurious counting responses and reduced reliability in real-world scenarios. To systematically address these limitations, we propose a new evaluation framework focused on model robustness and trustworthiness. Our contribution is two-fold: (i) we introduce PrACo++ (Prompt-Aware Counting++), a novel test suite featuring two dedicated evaluation protocols -- the negative-label test and the distractor test -- paired with new specialized metrics; and (ii) we present the MUCCA (MUlti-Category Class-Agnostic counting) evaluation dataset, a new collection of real-world images featuring multiple annotated object categories per scene, unlike existing CAC benchmarks that typically include a single category per image. Our extensive experimental evaluation of 10 state-of-the-art methods shows that, despite strong performance under standard counting metrics, current models exhibit significant weaknesses in understanding and grounding object class descriptions. Finally, we provide a quantitative analysis of how semantic similarity between prompts influences these failures. Overall, our results underscore the need for more semantically grounded architectures and offer a reliable framework for future assessment in open-world text-guided CAC methods.

cs.CV↗

arXiv2026

A Picture of Agentic Search

With automated systems increasingly issuing search queries alongside humans, Information Retrieval (IR) faces a major shift. Yet IR remains human-centred, with systems, evaluation metrics, user models, and datasets designed around human queries and behaviours. Consequently, IR operates under assumptions that no longer hold in practice, with changes to workload volumes, predictability, and querying behaviours. This misalignment affects system performance and optimisation: caching may lose effectiveness, query pre-processing may add overhead without improving results, and standard metrics may mismeasure satisfaction. Without adaptation, retrieval models risk satisfying neither humans, nor the emerging user segment of agents. However, datasets capturing agent search behaviour are lacking, which is a critical gap given IR's historical reliance on data-driven evaluation and optimisation. We develop a methodology for collecting all the data produced and consumed by agentic retrieval-augmented systems when answering queries, and we release the Agentic Search Queryset (ASQ) dataset. ASQ contains reasoning-induced queries, retrieved documents, and thoughts for queries in HotpotQA, Researchy Questions, and MS MARCO, for 3 diverse agents and 2 retrieval pipelines. The accompanying toolkit enables ASQ to be extended to new agents, retrievers, and datasets.

cs.IR↗

arXiv2026

AMAQA: A Metadata-based QA Dataset for RAG Systems

Retrieval-augmented generation (RAG) systems are widely used in question-answering (QA) tasks, but current benchmarks lack metadata integration, limiting their evaluation in scenarios requiring both textual data and external information. To address this, we present AMAQA, a new open-access QA dataset designed to evaluate tasks combining text and metadata. The integration of metadata is especially important in fields that require rapid analysis of large volumes of data, such as cybersecurity and intelligence, where timely access to relevant information is critical. AMAQA includes about 1.1 million English messages collected from 26 public Telegram groups, enriched with metadata such as timestamps and chat names. It also contains 20,000 hotel reviews with metadata. In addition, the dataset provides 2,600 high-quality QA pairs built across both domains, Telegram messages and hotel reviews, making AMAQA a valuable resource for advancing research on metadata-driven QA and RAG systems. Both Telegram messages and Hotel reviews are enriched with emotional tones or toxicity indicators. To the best of our knowledge, AMAQA is the first single-hop QA benchmark to incorporate metadata. We conduct extensive tests on the benchmark, setting a new reference point for future research. We show that leveraging metadata boosts accuracy from 0.5 to 0.86 for GPT-4o and from 0.27 to 0.76 for open source LLMs, highlighting the value of structured context. We conducted experiments on our benchmark to assess the performance of known techniques designed to enhance RAG, highlighting the importance of properly managing metadata throughout the entire RAG pipeline.

cs.IR↗

arXiv2026

Early-Exit Graph Neural Networks

Early-exit mechanisms allow deep neural networks to stop inference once prediction confidence is high, reducing latency and energy on easy inputs while retaining full-depth accuracy on harder ones. Similarly, adding early exit mechanisms to Graph Neural Networks (GNNs), the go-to models for graph-structured data, allows for dynamic trading depth for confidence on simple graphs while maintaining full-depth accuracy on harder ones to capture intricate relationships. Yet, their potential in deep GNNs, where over-smoothing, over-squashing or more generally vanishing gradients prevent these model to properly learn, remains largely unexplored. To address this, we introduce Symmetric-Anti-Symmetric GNNs (SAS-GNN), whose symmetry-based inductive biases yield stable intermediate representations that support safe early exits. Building on this backbone, we propose Early-Exit GNNs (EEGNNs), which attach confidence-aware exit neural heads which are trainable end-to-end based on the task objective, enabling on-the-fly termination at node or graph level. Experiments show that EEGNNs learn task-driven exit strategies, while achieving competitive results on heterophilic graphs and long-range tasks. Even when not outperforming the strongest baselines, EEGNNs consistently deliver favorable accuracy-efficiency trade-offs thanks to their adaptive and parameter-efficient design. We plan to release the code to reproduce our experiments.

cs.LG↗

arXiv2026

Statistical Foundations of DIME: Risk Estimation for Practical Index Selection

High-dimensional dense embeddings have become central to modern Information Retrieval, but many dimensions are noisy or redundant. Recently proposed DIME (Dimension IMportance Estimation), provides query-dependent scores to identify informative components of embeddings. DIME relies on a costly grid search to select a priori a dimensionality for all the query corpus's embeddings. Our work provides a statistically grounded criterion that directly identifies the optimal set of dimensions for each query at inference time. Experiments confirm achieving parity of effectiveness and reduces embedding size by an average of $\sim50\%$ across different models and datasets at inference time.

cs.IR↗

arXiv2025

Blending Learning to Rank and Dense Representations for Efficient and Effective Cascades

We investigate the exploitation of both lexical and neural relevance signals for ad-hoc passage retrieval. Our exploration involves a large-scale training dataset in which dense neural representations of MS-MARCO queries and passages are complemented and integrated with 253 hand-crafted lexical features extracted from the same corpus. Blending of the relevance signals from the two different groups of features is learned by a classical Learning-to-Rank (LTR) model based on a forest of decision trees. To evaluate our solution, we employ a pipelined architecture where a dense neural retriever serves as the first stage and performs a nearest-neighbor search over the neural representations of the documents. Our LTR model acts instead as the second stage that re-ranks the set of candidates retrieved by the first stage to enhance effectiveness. The results of reproducible experiments conducted with state-of-the-art dense retrievers on publicly available resources show that the proposed solution significantly enhances the end-to-end ranking performance while relatively minimally impacting efficiency. Specifically, we achieve a boost in nDCG@10 of up to 11% with an increase in average query latency of only 4.3%. This confirms the advantage of seamlessly combining two distinct families of signals that mutually contribute to retrieval effectiveness.

cs.IR↗

arXiv2025

Neural Prioritisation for Web Crawling

Given the vast scale of the Web, crawling prioritisation techniques based on link graph traversal, popularity, link analysis, and textual content are frequently applied to surface documents that are most likely to be valuable. While existing techniques are effective for keyword-based search, both retrieval methods and user search behaviours are shifting from keyword-based matching to natural language semantic matching. The remarkable success of applying semantic matching and quality signals during ranking leads us to hypothesize that crawling could be improved by prioritizing Web pages with high semantic quality. To investigate this, we propose a semantic quality-driven prioritisation technique to enhance the effectiveness of crawling and align the crawler behaviour with recent shift towards natural language search. We embed semantic understanding directly into the crawling process -- leveraging recent neural semantic quality estimators to prioritise the crawling frontier -- with the goal of surfacing content that is semantically rich and valuable for modern search needs. Our experiments on the English subset of ClueWeb22-B and the Researchy Questions query set show that, compared to existing crawling techniques, neural crawling policies significantly improve harvest rate, maxNDCG, and search effectiveness during the early stages of crawling. Meanwhile, crawlers based on our proposed neural policies maintain comparable search performance on keyword queries from the MS MARCO Web Search query set. While this work does not propose a definitive and complete solution, it presents a forward-looking perspective on Web crawling and opens the door to a new line of research on leveraging semantic analysis to effectively align crawlers with the ongoing shift toward natural language search.

cs.IR↗

arXiv2025

Efficient Recommendation with Millions of Items by Dynamic Pruning of Sub-Item Embeddings

A large item catalogue is a major challenge for deploying modern sequential recommender models, since it makes the memory footprint of the model large and increases inference latency. One promising approach to address this is RecJPQ, which replaces item embeddings with sub-item embeddings. However, slow inference remains problematic because finding the top highest-scored items usually requires scoring all items in the catalogue, which may not be feasible for large catalogues. By adapting dynamic pruning concepts from document retrieval, we propose the RecJPQPrune dynamic pruning algorithm to efficiently find the top highest-scored items without computing the scores of all items in the catalogue. Our RecJPQPrune algorithm is safe-up-to-rank K since it theoretically guarantees that no potentially high-scored item is excluded from the final top K recommendation list, thereby ensuring no impact on effectiveness. Our experiments on two large datasets and three recommendation models demonstrate the efficiency achievable using RecJPQPrune: for instance, on the Tmall dataset with 2.2M items, we can reduce the median model scoring time by 64 times compared to the Transformer Default baseline, and 5.3 times compared to a recent scoring approach called PQTopK. Overall, this paper demonstrates the effective and efficient inference of Transformer-based recommendation models at catalogue scales not previously reported in the literature. Indeed, our RecJPQPrune algorithm can score 2 million items in under 10 milliseconds without GPUs, and without relying on Approximate Nearest Neighbour (ANN) techniques.

cs.IR↗

arXiv2025

A Theoretical Analysis of Recommendation Loss Functions under Negative Sampling

Loss functions like Categorical Cross Entropy (CCE), Binary Cross Entropy (BCE), and Bayesian Personalized Ranking (BPR) are commonly used in training Recommender Systems (RSs) to differentiate positive items - those interacted with by users - and negative items. While prior works empirically showed that CCE outperforms BCE and BPR when using the full set of negative items, we provide a theoretical explanation for this by proving that CCE offers the tightest lower bound on ranking metrics like Normalized Discounted Cumulative Gain (NDCG) and Mean Reciprocal Rank (MRR), followed by BPR and BCE. However, using the full set of negative items is computationally infeasible for large-scale RSs, prompting the use of negative sampling techniques. Under negative sampling, we reveal that BPR and CCE are equivalent when a single negative sample is drawn, and all three losses converge to the same global minimum. We further demonstrate that the sampled losses remain lower bounds for NDCG (MRR), albeit in a probabilistic sense. Our worst-case analysis shows that BCE offers the strongest bound on NDCG (MRR). Experiments on five datasets and four models empirically support these theoretical findings. Our code and supplementary material are available at https://github.com/federicosiciliano/recsys_losses.git.

cs.IR↗

arXiv2025

Document Quality Scoring for Web Crawling

The internet contains large amounts of low-quality content, yet users expect web search engines to deliver high-quality, relevant results. The abundant presence of low-quality pages can negatively impact retrieval and crawling processes by wasting resources on these documents. Therefore, search engines can greatly benefit from techniques that leverage efficient quality estimation methods to mitigate these negative impacts. Quality scoring methods for web pages are useful for many processes typical for web search systems, including static index pruning, index tiering, and crawling. Building on work by Chang et al.~\cite{chang2024neural}, who proposed using neural estimators of semantic quality for static index pruning, we extend their approach and apply their neural quality scorers to assess the semantic quality of web pages in crawling prioritisation tasks. In our experimental analysis, we found that prioritising semantically high-quality pages over low-quality ones can improve downstream search effectiveness. Our software contribution consists of a Docker container that computes an effective quality score for a given web page, allowing the quality scorer to be easily included and used in other components of web search systems.

cs.IR↗

arXiv2025

MURR: Model Updating with Regularized Replay for Searching a Document Stream

The Internet produces a continuous stream of new documents and user-generated queries. These naturally change over time based on events in the world and the evolution of language. Neural retrieval models that were trained once on a fixed set of query-document pairs will quickly start misrepresenting newly-created content and queries, leading to less effective retrieval. Traditional statistical sparse retrieval can update collection statistics to reflect these changes in the use of language in documents and queries. In contrast, continued fine-tuning of the language model underlying neural retrieval approaches such as DPR and ColBERT creates incompatibility with previously-encoded documents. Re-encoding and re-indexing all previously-processed documents can be costly. In this work, we explore updating a neural dual encoder retrieval model without reprocessing past documents in the stream. We propose MURR, a model updating strategy with regularized replay, to ensure the model can still faithfully search existing documents without reprocessing, while continuing to update the model for the latest topics. In our simulated streaming environments, we show that fine-tuning models using MURR leads to more effective and more consistent retrieval results than other strategies as the stream of documents and queries progresses.

cs.IR↗

arXiv2025

Efficient Constant-Space Multi-Vector Retrieval

Multi-vector retrieval methods, exemplified by the ColBERT architecture, have shown substantial promise for retrieval by providing strong trade-offs in terms of retrieval latency and effectiveness. However, they come at a high cost in terms of storage since a (potentially compressed) vector needs to be stored for every token in the input collection. To overcome this issue, we propose encoding documents to a fixed number of vectors, which are no longer necessarily tied to the input tokens. Beyond reducing the storage costs, our approach has the advantage that document representations become of a fixed size on disk, allowing for better OS paging management. Through experiments using the MSMARCO passage corpus and BEIR with the ColBERT-v2 architecture, a representative multi-vector ranking model architecture, we find that passages can be effectively encoded into a fixed number of vectors while retaining most of the original effectiveness.

cs.IR↗

arXiv2025

Exploring the Effectiveness of Multi-stage Fine-tuning for Cross-encoder Re-rankers

State-of-the-art cross-encoders can be fine-tuned to be highly effective in passage re-ranking. The typical fine-tuning process of cross-encoders as re-rankers requires large amounts of manually labelled data, a contrastive learning objective, and a set of heuristically sampled negatives. An alternative recent approach for fine-tuning instead involves teaching the model to mimic the rankings of a highly effective large language model using a distillation objective. These fine-tuning strategies can be applied either individually, or in sequence. In this work, we systematically investigate the effectiveness of point-wise cross-encoders when fine-tuned independently in a single stage, or sequentially in two stages. Our experiments show that the effectiveness of point-wise cross-encoders fine-tuned using contrastive learning is indeed on par with that of models fine-tuned with multi-stage approaches. Code is available for reproduction at https://github.com/fpezzuti/multistage-finetuning.

cs.IR↗

arXiv2024

ECLIPSE: Contrastive Dimension Importance Estimation with Pseudo-Irrelevance Feedback for Dense Retrieval

Recent advances in Information Retrieval have leveraged high-dimensional embedding spaces to improve the retrieval of relevant documents. Moreover, the Manifold Clustering Hypothesis suggests that despite these high-dimensional representations, documents relevant to a query reside on a lower-dimensional, query-dependent manifold. While this hypothesis has inspired new retrieval methods, existing approaches still face challenges in effectively separating non-relevant information from relevant signals. We propose a novel methodology that addresses these limitations by leveraging information from both relevant and non-relevant documents. Our method, ECLIPSE, computes a centroid based on irrelevant documents as a reference to estimate noisy dimensions present in relevant ones, enhancing retrieval performance. Extensive experiments on three in-domain and one out-of-domain benchmarks demonstrate an average improvement of up to 19.50% (resp. 22.35%) in mAP(AP) and 11.42% (resp. 13.10%) in nDCG@10 w.r.t. the DIME-based baseline (resp. the baseline using all dimensions). Our results pave the way for more robust, pseudo-irrelevance-based retrieval systems in future IR research.

cs.IR↗

arXiv2024