Tommaso Di Noia

Explainable Emotion Recognition with EEG Wearable Devices

Distribution-Robust Reranking for Conversational Music Recommendation: PoliBaJukebox at the ACM RecSys Challenge 2026

We describe the PoliBaJukebox submission to the ACM RecSys Challenge 2026 on conversational music recommendation over the TalkPlayData 2 corpus. Our system implements a modular two-stage pipeline: a first stage retrieves candidates from the full track catalog using ten heterogeneous sources, fuses them with weighted Re...

From Side Information and Knowledge Graphs to Large Language Models: Two Decades of Knowledge Integration in Recommender Systems

Over the past two decades, Recommender Systems (RSs) have repeatedly changed how they integrate external knowledge, moving from constraints and side information to Knowledge Graphs (KG) and, more recently, Large Language Models (LLMs). This Past/Present/Future retrospective interprets that evolution as a sequence of re...

Derivative-Free Emergent Optimization via Local Rules and Cellular Automata

We introduce the Cellular Direct Search (CDS), a derivative-free optimization method inspired by the emergent dynamics of Cellular Automata. CDS partitions the parameter space into a discrete grid where a population of agents evolves according to strictly local rules: a Jealous Neighbor Rule for competitive selection a...

Data Processing for Offline Evaluation in Recommender Systems: a Survey

A systematic, cross-domain characterisation of data processing practices for the offline evaluation of recommender systems is provided, identifying substantial heterogeneity in how auxiliary information is prepared and represented, as well as inconsistencies in the specification of data splitting protocols.

PUMA: Post-Hoc Sparsification of Universal Multimodal Embeddings for Efficient Retrieval

Universal multimodal embedders enable retrieval across text, image, and combined queries, but their dense representations incur high memory and inference costs. Post-hoc sparsification could reduce these costs but remains underexplored for multimodal retrieval. We introduce PUMA, a sparse autoencoder recipe that maps u...

Beyond Single-Signal Retrieval: A Graph-Aware Agentic Framework for Conversational Music Recommendation

Experimental results show the pipeline outperforms baselines in ranking, diversity, and explanations, proving that combining graph embeddings with agentic reasoning successfully bridges long-term profiles and immediate intent.

Mind the Metric: Reproducibility and Fair Benchmarking of Spectral Graph Models for Collaborative Filtering

A reproducibility and replicability study that examines three major families of graph models, showing that strong classical methods remain highly competitive and that no single spectral model dominates across domains, and finds that spectral filtering often improves catalog exploration even when accuracy gains are marg...

GradeSQL: Outcome reward models for intelligent Text-to-SQL generation from LLMs

As Large Language Models (LLMs) become foundational to next-generation Intelligent Information Systems, the bridge between natural language interfaces and structured database systems remains a critical bottleneck. While Text-to-SQL generation enables cooperative support for complex query formulation, ensuring the rel...