Papers by Antonio Mallia

2 papers
Statistical Foundations of DIME: Risk Estimation for Practical Index Selection (2026.eacl-short)

Copied to clipboard

Challenge: High-dimensional dense embeddings are noisy or redundant, causing performance degradation and causing errors.
Approach: They propose a method that scores each dimension by fusing the embeddings into a query-dependent matrix.
Outcome: The proposed method improves retrieval effectiveness and reduces embedding size by an average 50% of across different models and datasets at inference time.
Accelerating Learned Sparse Indexes Via Term Impact Decomposition (2022.findings-emnlp)

Copied to clipboard

Challenge: Novel inverted index-based learned sparse ranking models provide more effective, but less efficient, retrieval performance compared to traditional ranking models.
Approach: They propose a technique that allows for automatic pruning of ranking models by storing metadata about index term importance scores.
Outcome: The proposed technique accelerates top-k retrieval by 9.6X without loss in effectiveness.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations