arXiv subjects

Andrea Esuli

Publications and source records attributed to Andrea Esuli.

At least 19 recordsLinked to original sources

arXiv2025

Challenging the Abilities of Large Language Models in Italian: a Community Initiative

The rapid progress of Large Language Models (LLMs) has transformed natural language processing and broadened its impact across research and society. Yet, systematic evaluation of these models, especially for languages beyond English, remains limited. "Challenging the Abilities of LAnguage Models in ITAlian" (CALAMITA) is a large-scale collaborative benchmarking initiative for Italian, coordinated under the Italian Association for Computational Linguistics. Unlike existing efforts that focus on leaderboards, CALAMITA foregrounds methodology: it federates more than 80 contributors from academia, industry, and the public sector to design, document, and evaluate a diverse collection of tasks, covering linguistic competence, commonsense reasoning, factual consistency, fairness, summarization, translation, and code generation. Through this process, we not only assembled a benchmark of over 20 tasks and almost 100 subtasks, but also established a centralized evaluation pipeline that supports heterogeneous datasets and metrics. We report results for four open-weight LLMs, highlighting systematic strengths and weaknesses across abilities, as well as challenges in task-specific evaluation. Beyond quantitative results, CALAMITA exposes methodological lessons: the necessity of fine-grained, task-representative metrics, the importance of harmonized pipelines, and the benefits and limitations of broad community engagement. CALAMITA is conceived as a rolling benchmark, enabling continuous integration of new tasks and models. This makes it both a resource -- the most comprehensive and diverse benchmark for Italian to date -- and a framework for sustainable, community-driven evaluation. We argue that this combination offers a blueprint for other languages and communities seeking inclusive and rigorous LLM evaluation practices.

cs.CL↗

arXiv2025

Quantifying Query Fairness Under Unawareness

Traditional ranking algorithms are designed to retrieve the most relevant items for a user's query, but they often inherit biases from data that can unfairly disadvantage vulnerable groups. Fairness in information access systems (IAS) is typically assessed by comparing the distribution of groups in a ranking to a target distribution, such as the overall group distribution in the dataset. These fairness metrics depend on knowing the true group labels for each item. However, when groups are defined by demographic or sensitive attributes, these labels are often unknown, leading to a setting known as "fairness under unawareness". To address this, group membership can be inferred using machine-learned classifiers, and group prevalence is estimated by counting the predicted labels. Unfortunately, such an estimation is known to be unreliable under dataset shift, compromising the accuracy of fairness evaluations. In this paper, we introduce a robust fairness estimator based on quantification that effectively handles multiple sensitive attributes beyond binary classifications. Our method outperforms existing baselines across various sensitive attributes and, to the best of our knowledge, is the first to establish a reliable protocol for measuring fairness under unawareness across multiple queries and groups.

cs.IR↗

arXiv2025

Stress-testing Machine Generated Text Detection: Shifting Language Models Writing Style to Fool Detectors

Recent advancements in Generative AI and Large Language Models (LLMs) have enabled the creation of highly realistic synthetic content, raising concerns about the potential for malicious use, such as misinformation and manipulation. Moreover, detecting Machine-Generated Text (MGT) remains challenging due to the lack of robust benchmarks that assess generalization to real-world scenarios. In this work, we present a pipeline to test the resilience of state-of-the-art MGT detectors (e.g., Mage, Radar, LLM-DetectAIve) to linguistically informed adversarial attacks. To challenge the detectors, we fine-tune language models using Direct Preference Optimization (DPO) to shift the MGT style toward human-written text (HWT). This exploits the detectors' reliance on stylistic clues, making new generations more challenging to detect. Additionally, we analyze the linguistic shifts induced by the alignment and which features are used by detectors to detect MGT texts. Our results show that detectors can be easily fooled with relatively few examples, resulting in a significant drop in detection performance. This highlights the importance of improving detection methods and making them robust to unseen in-domain texts.

cs.CL↗

arXiv2025

Optimizing LLMs for Italian: Reducing Token Fertility and Enhancing Efficiency Through Vocabulary Adaptation

The number of pretrained Large Language Models (LLMs) is increasing steadily, though the majority are designed predominantly for the English language. While state-of-the-art LLMs can handle other languages, due to language contamination or some degree of multilingual pretraining data, they are not optimized for non-English languages, leading to inefficient encoding (high token "fertility") and slower inference speed. In this work, we thoroughly compare a variety of vocabulary adaptation techniques for optimizing English LLMs for the Italian language, and put forward Semantic Alignment Vocabulary Adaptation (SAVA), a novel method that leverages neural mapping for vocabulary substitution. SAVA achieves competitive performance across multiple downstream tasks, enhancing grounded alignment strategies. We adapt two LLMs: Mistral-7b-v0.1, reducing token fertility by 25\%, and Llama-3.1-8B, optimizing the vocabulary and reducing the number of parameters by 1 billion. We show that, following the adaptation of the vocabulary, these models can recover their performance with a relatively limited stage of continual training on the target language. Finally, we test the capabilities of the adapted models on various multi-choice and generative tasks.

cs.CL↗

arXiv2024

The Invalsi Benchmarks: measuring Linguistic and Mathematical understanding of Large Language Models in Italian

While Italian is a high-resource language, there are few Italian-native benchmarks to evaluate generative Large Language Models (LLMs) in this language. This work presents three new benchmarks: Invalsi MATE to evaluate models performance on mathematical understanding in Italian, Invalsi ITA to evaluate language understanding in Italian and Olimpiadi MATE for more complex mathematical understanding. The first two benchmarks are based on the Invalsi tests, which are administered to students of age between 6 and 18 within the Italian school system and have been validated by several experts in teaching and pedagogy, the third one comes from the Italian high school math Olympics. We evaluate 10 powerful language models on these benchmarks and find that they are bound by 71% accuracy on Invasli MATE, achieved by Llama 3.1 70b instruct and by 88% on Invalsi ITA. For both Invalsi MATE and Invalsi ITA we compare LLMs with the average performance of Italian students to show that Llama 3.1 is the only one to outperform them on Invalsi MATE while most models do so on Invalsi ITA, we then show that Olimpiadi MATE is more challenging than Invalsi MATE and the highest accuracy, achieved by Llama 3.1 405b instruct is 45%. We will make data and evaluation code openly available upon acceptance of the paper.

cs.CL↗

arXiv2024

AI "News" Content Farms Are Easy to Make and Hard to Detect: A Case Study in Italian

Large Language Models (LLMs) are increasingly used as "content farm" models (CFMs), to generate synthetic text that could pass for real news articles. This is already happening even for languages that do not have high-quality monolingual LLMs. We show that fine-tuning Llama (v1), mostly trained on English, on as little as 40K Italian news articles, is sufficient for producing news-like texts that native speakers of Italian struggle to identify as synthetic. We investigate three LLMs and three methods of detecting synthetic texts (log-likelihood, DetectGPT, and supervised classification), finding that they all perform better than human raters, but they are all impractical in the real world (requiring either access to token likelihood information or a large dataset of CFM texts). We also explore the possibility of creating a proxy CFM: an LLM fine-tuned on a similar dataset to one used by the real "content farm". We find that even a small amount of fine-tuning data suffices for creating a successful detector, but we need to know which base LLM is used, which is a major challenge. Our results suggest that there are currently no practical methods for detecting synthetic news-like texts 'in the wild', while generating them is too easy. We highlight the urgency of more NLP research on this problem.

cs.CL↗

arXiv2023

Detecting Images Generated by Diffusers

This paper explores the task of detecting images generated by text-to-image diffusion models. To evaluate this, we consider images generated from captions in the MSCOCO and Wikimedia datasets using two state-of-the-art models: Stable Diffusion and GLIDE. Our experiments show that it is possible to detect the generated images using simple Multi-Layer Perceptrons (MLPs), starting from features extracted by CLIP, or traditional Convolutional Neural Networks (CNNs). We also observe that models trained on images generated by Stable Diffusion can detect images generated by GLIDE relatively well, however, the reverse is not true. Lastly, we find that incorporating the associated textual information with the images rarely leads to significant improvement in detection results but that the type of subject depicted in the image can have a significant impact on performance. This work provides insights into the feasibility of detecting generated images, and has implications for security and privacy concerns in real-world applications. The code to reproduce our results is available at: https://github.com/davide-coccomini/Detecting-Images-Generated-by-Diffusers

cs.CV↗

arXiv2023

Measuring Fairness Under Unawareness of Sensitive Attributes: A Quantification-Based Approach

Algorithms and models are increasingly deployed to inform decisions about people, inevitably affecting their lives. As a consequence, those in charge of developing these models must carefully evaluate their impact on different groups of people and favour group fairness, that is, ensure that groups determined by sensitive demographic attributes, such as race or sex, are not treated unjustly. To achieve this goal, the availability (awareness) of these demographic attributes to those evaluating the impact of these models is fundamental. Unfortunately, collecting and storing these attributes is often in conflict with industry practices and legislation on data minimisation and privacy. For this reason, it can be hard to measure the group fairness of trained models, even from within the companies developing them. In this work, we tackle the problem of measuring group fairness under unawareness of sensitive attributes, by using techniques from quantification, a supervised learning task concerned with directly providing group-level prevalence estimates (rather than individual-level class labels). We show that quantification approaches are particularly suited to tackle the fairness-under-unawareness problem, as they are robust to inevitable distribution shifts while at the same time decoupling the (desirable) objective of measuring group fairness from the (undesirable) side effect of allowing the inference of sensitive attributes of individuals. More in detail, we show that fairness under unawareness can be cast as a quantification problem and solved with proven methods from the quantification literature. We show that these methods outperform previous approaches to measure demographic parity in five experimental protocols, corresponding to important challenges that complicate the estimation of classifier fairness under unawareness.

cs.CY↗

arXiv2022

Unravelling Interlanguage Facts via Explainable Machine Learning

Native language identification (NLI) is the task of training (via supervised machine learning) a classifier that guesses the native language of the author of a text. This task has been extensively researched in the last decade, and the performance of NLI systems has steadily improved over the years. We focus on a different facet of the NLI task, i.e., that of analysing the internals of an NLI classifier trained by an \emph{explainable} machine learning algorithm, in order to obtain explanations of its classification decisions, with the ultimate goal of gaining insight into which linguistic phenomena ``give a speaker's native language away''. We use this perspective in order to tackle both NLI and a (much less researched) companion task, i.e., guessing whether a text has been written by a native or a non-native speaker. Using three datasets of different provenance (two datasets of English learners' essays and a dataset of social media posts), we investigate which kind of linguistic traits (lexical, morphological, syntactic, and statistical) are most effective for solving our two tasks, namely, are most indicative of a speaker's L1. We also present two case studies, one on Spanish and one on Italian learners of English, in which we analyse individual linguistic traits that the classifiers have singled out as most important for spotting these L1s. Overall, our study shows that the use of explainable machine learning can be a valuable tool for th

cs.CL↗

arXiv2022

Transformer-Based Multi-modal Proposal and Re-Rank for Wikipedia Image-Caption Matching

With the increased accessibility of web and online encyclopedias, the amount of data to manage is constantly increasing. In Wikipedia, for example, there are millions of pages written in multiple languages. These pages contain images that often lack the textual context, remaining conceptually floating and therefore harder to find and manage. In this work, we present the system we designed for participating in the Wikipedia Image-Caption Matching challenge on Kaggle, whose objective is to use data associated with images (URLs and visual data) to find the correct caption among a large pool of available ones. A system able to perform this task would improve the accessibility and completeness of multimedia content on large online encyclopedias. Specifically, we propose a cascade of two models, both powered by the recent Transformer model, able to efficiently and effectively infer a relevance score between the query image data and the captions. We verify through extensive experimentation that the proposed two-model approach is an effective way to handle a large pool of images and captions while maintaining bounded the overall computational complexity at inference time. Our approach achieves remarkable results, obtaining a normalized Discounted Cumulative Gain (nDCG) value of 0.53 on the private leaderboard of the Kaggle challenge.

cs.CV↗

arXiv2021

LeQua@CLEF2022: Learning to Quantify

LeQua 2022 is a new lab for the evaluation of methods for "learning to quantify" in textual datasets, i.e., for training predictors of the relative frequencies of the classes of interest in sets of unlabelled textual documents. While these predictions could be easily achieved by first classifying all documents via a text classifier and then counting the numbers of documents assigned to the classes, a growing body of literature has shown this approach to be suboptimal, and has proposed better methods. The goal of this lab is to provide a setting for the comparative evaluation of methods for learning to quantify, both in the binary setting and in the single-label multiclass setting. For each such setting we provide data either in ready-made vector form or in raw document form.

cs.LG↗

arXiv2021

QuaPy: A Python-Based Framework for Quantification

QuaPy is an open-source framework for performing quantification (a.k.a. supervised prevalence estimation), written in Python. Quantification is the task of training quantifiers via supervised learning, where a quantifier is a predictor that estimates the relative frequencies (a.k.a. prevalence values) of the classes of interest in a sample of unlabelled data. While quantification can be trivially performed by applying a standard classifier to each unlabelled data item and counting how many data items have been assigned to each class, it has been shown that this "classify and count" method is outperformed by methods specifically designed for quantification. QuaPy provides implementations of a number of baseline methods and advanced quantification methods, of routines for quantification-oriented model selection, of several broadly accepted evaluation measures, and of robust evaluation protocols routinely used in the field. QuaPy also makes available datasets commonly used for testing quantifiers, and offers visualization tools for facilitating the analysis and interpretation of the results. The software is open-source and publicly available under a BSD-3 licence via https://github.com/HLT-ISTI/QuaPy, and can be installed via pip (https://pypi.org/project/QuaPy/)

cs.LG↗

arXiv2021

Fine-grained Visual Textual Alignment for Cross-Modal Retrieval using Transformer Encoders

Despite the evolution of deep-learning-based visual-textual processing systems, precise multi-modal matching remains a challenging task. In this work, we tackle the task of cross-modal retrieval through image-sentence matching based on word-region alignments, using supervision only at the global image-sentence level. Specifically, we present a novel approach called Transformer Encoder Reasoning and Alignment Network (TERAN). TERAN enforces a fine-grained match between the underlying components of images and sentences, i.e., image regions and words, respectively, in order to preserve the informative richness of both modalities. TERAN obtains state-of-the-art results on the image retrieval task on both MS-COCO and Flickr30k datasets. Moreover, on MS-COCO, it also outperforms current approaches on the sentence retrieval task. Focusing on scalable cross-modal information retrieval, TERAN is designed to keep the visual and textual data pipelines well separated. Cross-attention links invalidate any chance to separately extract visual and textual features needed for the online search and the offline indexing steps in large-scale retrieval systems. In this respect, TERAN merges the information from the two domains only during the final alignment phase, immediately before the loss computation. We argue that the fine-grained alignments produced by TERAN pave the way towards the research for effective and efficient methods for large-scale cross-modal information retrieval. We compare the effectiveness of our approach against relevant state-of-the-art methods. On the MS-COCO 1K test set, we obtain an improvement of 5.7% and 3.5% respectively on the image and the sentence retrieval tasks on the Recall@1 metric. The code used for the experiments is publicly available on GitHub at https://github.com/mesnico/TERAN.

cs.CV↗

arXiv2021

Transformer Reasoning Network for Image-Text Matching and Retrieval

Image-text matching is an interesting and fascinating task in modern AI research. Despite the evolution of deep-learning-based image and text processing systems, multi-modal matching remains a challenging problem. In this work, we consider the problem of accurate image-text matching for the task of multi-modal large-scale information retrieval. State-of-the-art results in image-text matching are achieved by inter-playing image and text features from the two different processing pipelines, usually using mutual attention mechanisms. However, this invalidates any chance to extract separate visual and textual features needed for later indexing steps in large-scale retrieval systems. In this regard, we introduce the Transformer Encoder Reasoning Network (TERN), an architecture built upon one of the modern relationship-aware self-attentive architectures, the Transformer Encoder (TE). This architecture is able to separately reason on the two different modalities and to enforce a final common abstract concept space by sharing the weights of the deeper transformer layers. Thanks to this design, the implemented network is able to produce compact and very rich visual and textual features available for the successive indexing step. Experiments are conducted on the MS-COCO dataset, and we evaluate the results using a discounted cumulative gain metric with relevance computed exploiting caption similarities, in order to assess possibly non-exact but relevant search results. We demonstrate that on this metric we are able to achieve state-of-the-art results in the image retrieval task. Our code is freely available at https://github.com/mesnico/TERN.

cs.CV↗

arXiv2020

Cross-Lingual Sentiment Quantification

\emph{Sentiment Quantification} (i.e., the task of estimating the relative frequency of sentiment-related classes -- such as \textsf{Positive} and \textsf{Negative} -- in a set of unlabelled documents) is an important topic in sentiment analysis, as the study of sentiment-related quantities and trends across a population is often of higher interest than the analysis of individual instances. In this work we propose a method for \emph{Cross-Lingual Sentiment Quantification}, the task of performing sentiment quantification when training documents are available for a source language $\mathcal{S}$ but not for the target language $\mathcal{T}$ for which sentiment quantification needs to be performed. Cross-lingual sentiment quantification (and cross-lingual \emph{text} quantification in general) has never been discussed before in the literature; we establish baseline results for the binary case by combining state-of-the-art quantification methods with methods capable of generating cross-lingual vectorial representations of the source and target documents involved. We present experimental results obtained on publicly available datasets for cross-lingual sentiment classification; the results show that the presented methods can perform cross-lingual sentiment quantification with a surprising level of accuracy.

cs.LG↗

arXiv2019

Word-Class Embeddings for Multiclass Text Classification

Pre-trained word embeddings encode general word semantics and lexical regularities of natural language, and have proven useful across many NLP tasks, including word sense disambiguation, machine translation, and sentiment analysis, to name a few. In supervised tasks such as multiclass text classification (the focus of this article) it seems appealing to enhance word representations with ad-hoc embeddings that encode task-specific information. We propose (supervised) word-class embeddings (WCEs), and show that, when concatenated to (unsupervised) pre-trained word embeddings, they substantially facilitate the training of deep-learning models in multiclass classification by topic. We show empirical evidence that WCEs yield a consistent improvement in multiclass classification accuracy, using four popular neural architectures and six widely used and publicly available datasets for multiclass text classification. Our code that implements WCEs is publicly available at https://github.com/AlexMoreo/word-class-embeddings

cs.LG↗

arXiv2019

Funnelling: A New Ensemble Method for Heterogeneous Transfer Learning and its Application to Cross-Lingual Text Classification

Cross-lingual Text Classification (CLC) consists of automatically classifying, according to a common set C of classes, documents each written in one of a set of languages L, and doing so more accurately than when naively classifying each document via its corresponding language-specific classifier. In order to obtain an increase in the classification accuracy for a given language, the system thus needs to also leverage the training examples written in the other languages. We tackle multilabel CLC via funnelling, a new ensemble learning method that we propose here. Funnelling consists of generating a two-tier classification system where all documents, irrespectively of language, are classified by the same (2nd-tier) classifier. For this classifier all documents are represented in a common, language-independent feature space consisting of the posterior probabilities generated by 1st-tier, language-dependent classifiers. This allows the classification of all test documents, of any language, to benefit from the information present in all training documents, of any language. We present substantial experiments, run on publicly available multilingual text collections, in which funnelling is shown to significantly outperform a number of state-of-the-art baselines. All code and datasets (in vector form) are made publicly available.

cs.LG↗

arXiv2019

Learning to Weight for Text Classification

In information retrieval (IR) and related tasks, term weighting approaches typically consider the frequency of the term in the document and in the collection in order to compute a score reflecting the importance of the term for the document. In tasks characterized by the presence of training data (such as text classification) it seems logical that the term weighting function should take into account the distribution (as estimated from training data) of the term across the classes of interest. Although `supervised term weighting' approaches that use this intuition have been described before, they have failed to show consistent improvements. In this article we analyse the possible reasons for this failure, and call consolidated assumptions into question. Following this criticism we propose a novel supervised term weighting approach that, instead of relying on any predefined formula, learns a term weighting function optimised on the training set of interest; we dub this approach \emph{Learning to Weight} (LTW). The experiments that we run on several well-known benchmarks, and using different learning methods, show that our method outperforms previous term weighting approaches in text classification.

cs.LG↗