Rossano Venturini

Franco Maria Nardini, Cosimo Rulli, Salvatore Trani et al.

Quantization and pruning are two effective Deep Neural Networks model compression methods. In this paper, we propose Automatic Prune Binarization (APB), a novel compression technique combining quantization with pruning. APB enhances the representational capability of binary networks using a few full-precision weights. Our technique jointly maximizes the accuracy of the network while minimizing its memory impact by deciding whether each weight should be binarized or kept in full precision. We show how to efficiently perform a forward pass through layers compressed using APB by decomposing it into a binary and a sparse-dense matrix multiplication. Moreover, we design two novel efficient algorithms for extremely quantized matrix multiplication on CPU, leveraging highly efficient bitwise operations. The proposed algorithms are 6.9x and 1.5x faster than available state-of-the-art solutions. We extensively evaluate APB on two widely adopted model compression datasets, namely CIFAR10 and ImageNet. APB delivers better accuracy/memory trade-off compared to state-of-the-art methods based on i) quantization, ii) pruning, and iii) combination of pruning and quantization. APB outperforms quantization in the accuracy/efficiency trade-off, being up to 2x faster than the 2-bit quantized model with no loss in accuracy.

Thong Nguyen, Cosimo Rulli, Franco Maria Nardini et al.

State-of-the-art Learned Sparse Retrieval (LSR) models, such as Splade, typically employ a Language Modeling (LM) head to project latent hidden states into a lexically-anchored logit matrix. This intermediate matrix is subsequently transformed into a sparse lexical representation through element-wise operations (ReLU, Log1P) and max-pooling over the sequence dimension. Despite its effectiveness, the LM head creates a massive memory bottleneck due to the sheer size of the vocabulary (V), which can range from 30,000 to over 250,000 tokens in recent models. Materializing this matrix creates a significant memory bottleneck, limiting model scaling. The resulting I/O overhead between operators further throttles throughput and runtime performance. In this paper, we propose Sparton, a fast memory-efficient Triton kernel tailored for the LM head in LSR models. Sparton utilizes a fused approach that integrates the tiled matrix multiplication, ReLU, Log1P, and max-reduction into a single GPU kernel. By performing an early online reduction directly on raw logit tiles, Sparton avoids materializing the full logit matrix in memory. Our experiments demonstrate that the Sparton kernel, in isolation, achieves up to a 4.8x speedup and an order-of-magnitude reduction in peak memory usage compared to PyTorch baselines. Integrated into Splade (|V| ~ 30k), Sparton enables a 33% larger batch size and 14% faster training with no effectiveness loss. On a multilingual backbone (|V| ~ 250k), these gains jump to a 26x larger batch size and 2.5x faster training.

Silvio Martinico, Franco Maria Nardini, Cosimo Rulli et al.

Multivector retrieval models achieve state-of-the-art effectiveness through fine-grained token-level representations, but their deployment incurs substantial computational and memory costs. Current solutions, based on the well-known k-means clustering algorithm, group similar vectors together to enable both effective compression and efficient retrieval. However, standard k-means scales poorly with the number of clusters and dataset size, and favours frequent tokens during training while underrepresenting rare, discriminative ones. In this work, we introduce TACHIOM, a multivector retrieval system that exploits token-level structure to significantly accelerate both clustering and retrieval. By accounting for tokens' distribution during centroid allocation, TACHIOM easily scales to millions of centroids, enabling highly accurate document scoring using only centroids, avoiding expensive token-level computation. TACHIOM combines a graph-based index over centroids with an optimized Product Quantization layout for efficient final scoring. Experiments on MS-MARCOv1 and LoTTE show that TACHIOM achieves up to $247\times$ faster clustering than k-means and up to $9.8\times$ retrieval speedup over state-of-the-art systems while maintaining comparable or superior effectiveness.

Sebastian Bruch, Franco Maria Nardini, Cosimo Rulli et al.

Sparse embeddings of data form an attractive class due to their inherent interpretability: Every dimension is tied to a term in some vocabulary, making it easy to visually decipher the latent space. Sparsity, however, poses unique challenges for Approximate Nearest Neighbor Search (ANNS) which finds, from a collection of vectors, the k vectors closest to a query. To encourage research on this underexplored topic, sparse ANNS featured prominently in a BigANN Challenge at NeurIPS 2023, where approximate algorithms were evaluated on large benchmark datasets by throughput and accuracy. In this work, we introduce a set of novel data structures and algorithmic methods, a combination of which leads to an elegant, effective, and highly efficient solution to sparse ANNS. Our contributions range from a theoretically-grounded sketching algorithm for sparse vectors to reduce their effective dimensionality while preserving inner product-induced ranks; a geometric organization of the inverted index; and the blending of local and global information to improve the efficiency and efficacy of ANNS. Empirically, our final algorithm, dubbed Seismic, reaches sub-millisecond per-query latency with high accuracy on a large-scale benchmark dataset using a single CPU.

Lorenzo Beretta, Franco Maria Nardini, Roberto Trani et al.

A tournament graph is a complete directed graph, which can be used to model a round-robin tournament between $n$ players. In this paper, we address the problem of finding a champion of the tournament, also known as Copeland winner, which is a player that wins the highest number of matches. In detail, we aim to investigate algorithms that find the champion by playing a low number of matches. Solving this problem allows us to speed up several Information Retrieval and Recommender System applications, including question answering, conversational search, etc. Indeed, these applications often search for the champion inducing a round-robin tournament among the players by employing a machine learning model to estimate who wins each pairwise comparison. Our contribution, thus, allows finding the champion by performing a low number of model inferences. We prove that any deterministic or randomized algorithm finding a champion with constant success probability requires $Ω(\ell n)$ comparisons, where $\ell$ is the number of matches lost by the champion. We then present an asymptotically-optimal deterministic algorithm matching this lower bound without knowing $\ell$, and we extend our analysis to three variants of the problem. Lastly, we conduct a comprehensive experimental assessment of the proposed algorithms on a question answering task on public data. Results show that our proposed algorithms speed up the retrieval of the champion up to $13\times$ with respect to the state-of-the-art algorithm that perform the full tournament.

Simon Gog, Giulio Ermanno Pibiri, Rossano Venturini

Query Auto-Completion (QAC) is an ubiquitous feature of modern textual search systems, suggesting possible ways of completing the query being typed by the user. Efficiency is crucial to make the system have a real-time responsiveness when operating in the million-scale search space. Prior work has extensively advocated the use of a trie data structure for fast prefix-search operations in compact space. However, searching by prefix has little discovery power in that only completions that are prefixed by the query are returned. This may impact negatively the effectiveness of the QAC system, with a consequent monetary loss for real applications like Web Search Engines and eCommerce. In this work we describe the implementation that empowers a new QAC system at eBay, and discuss its efficiency/effectiveness in relation to other approaches at the state-of-the-art. The solution is based on the combination of an inverted index with succinct data structures, a much less explored direction in the literature. This system is replacing the previous implementation based on Apache SOLR that was not always able to meet the required service-level-agreement.

Giulio Ermanno Pibiri, Rossano Venturini

The data structure at the core of large-scale search engines is the inverted index, which is essentially a collection of sorted integer sequences called inverted lists. Because of the many documents indexed by such engines and stringent performance requirements imposed by the heavy load of queries, the inverted index stores billions of integers that must be searched efficiently. In this scenario, index compression is essential because it leads to a better exploitation of the computer memory hierarchy for faster query processing and, at the same time, allows reducing the number of storage machines. The aim of this article is twofold: first, surveying the encoding algorithms suitable for inverted index compression and, second, characterizing the performance of the inverted index through experimentation.

Raffaele Perego, Giulio Ermanno Pibiri, Rossano Venturini

The sheer increase in volume of RDF data demands efficient solutions for the triple indexing problem, that is devising a compressed data structure to compactly represent RDF triples by guaranteeing, at the same time, fast pattern matching operations. This problem lies at the heart of delivering good practical performance for the resolution of complex SPARQL queries on large RDF datasets. In this work, we propose a trie-based index layout to solve the problem and introduce two novel techniques to reduce its space of representation for improved effectiveness. The extensive experimental analysis conducted over a wide range of publicly available real-world datasets, reveals that our best space/time trade-off configuration substantially outperforms existing solutions at the state-of-the-art, by taking 30-60% less space and speeding up query execution by a factor of 2-81x.

Giulio Ermanno Pibiri, Rossano Venturini

This paper deals with the two fundamental problems concerning the handling of large n-gram language models: indexing, that is compressing the n-gram strings and associated satellite data without compromising their retrieval speed; and estimation, that is computing the probability distribution of the strings from a large textual source. Regarding the problem of indexing, we describe compressed, exact and lossless data structures that achieve, at the same time, high space reductions and no time degradation with respect to state-of-the-art solutions and related software packages. In particular, we present a compressed trie data structure in which each word following a context of fixed length k, i.e., its preceding k words, is encoded as an integer whose value is proportional to the number of words that follow such context. Since the number of words following a given context is typically very small in natural languages, we lower the space of representation to compression levels that were never achieved before. Despite the significant savings in space, our technique introduces a negligible penalty at query time. Regarding the problem of estimation, we present a novel algorithm for estimating modified Kneser-Ney language models, that have emerged as the de-facto choice for language modeling in both academia and industry, thanks to their relatively low perplexity performance. Estimating such models from large textual sources poses the challenge of devising algorithms that make a parsimonious use of the disk. The state-of-the-art algorithm uses three sorting steps in external memory: we show an improved construction that requires only one sorting step thanks to exploiting the properties of the extracted n-gram strings. With an extensive experimental analysis performed on billions of n-grams, we show an average improvement of 4.5X on the total running time of the state-of-the-art approach.

Giulio Ermanno Pibiri, Rossano Venturini

The ubiquitous Variable-Byte encoding is one of the fastest compressed representation for integer sequences. However, its compression ratio is usually not competitive with other more sophisticated encoders, especially when the integers to be compressed are small that is the typical case for inverted indexes. This paper shows that the compression ratio of Variable-Byte can be improved by 2x by adopting a partitioned representation of the inverted lists. This makes Variable-Byte surprisingly competitive in space with the best bit-aligned encoders, hence disproving the folklore belief that Variable-Byte is space-inefficient for inverted index compression. Despite the significant space savings, we show that our optimization almost comes for free, given that: we introduce an optimal partitioning algorithm that does not affect indexing time because of its linear-time complexity; we show that the query processing speed of Variable-Byte is preserved, with an extensive experimental analysis and comparison with several other state-of-the-art encoders.