This report describes our contribution to the EVALITA 2023 shared task MULTI-Fake-DetectIVE which involves the classification of news including textual and visual components. To experiment on this task we focus on textual data augmentation, extending the Italian text and the Images available in the training set using machine translation models and image captioning ones. To train using different set of input features, we use different transformer encoders for each variant of text (Italian, English) and modality (Image). For Task 1, among the models we test, we find that using the Italian text together with its translation improves the model performance while the captions don't provide any improvement. We test the same architecture also on Task 2 although in this case we achieve less satisfactory results

In high recall retrieval tasks, human experts review a large pool of documents with the goal of satisfying an information need. Documents are prioritized for review through an active learning policy, and the process is usually referred to as Technology-Assisted Review (TAR). TAR tasks also aim to stop the review process once the target recall is achieved to minimize the annotation cost. In this paper, we introduce a new stopping rule called SALR? (SLD for Active Learning), a modified version of the Saerens-Latinne-Decaestecker algorithm (SLD) that has been adapted for use in active learning. Experiments show that our algorithm stops the review well ahead of the current state-of-the-art methods, while providing the same guarantees of achieving the target recall.

ISTI-CNR released a new web application for the manual and automatic classification of documents. Human annotators collaboratively label documents with machine learning algorithms that learn from annotators' actions and support the activity with classification suggestions. The platform supports the early stages of document labelling, with the ability to change the classification scheme on the go and to reuse and adapt existing classifiers.

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 explainable machine learning (EML) 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; our experiments indicate that the most discriminative features are the lexical ones, followed by the morphological, syntactic, and statistical features, in this order. We also present two case studies, one on Italian and one on Spanish learners of English, in which we analyse individual linguistic traits that the classifiers have singled out as most important for spotting these L1s; we show that the traits identified as most discriminative well align with our intuition, i.e. represent typical patterns of language misuse, underuse, or overuse, by speakers of the given L1. Overall, our study shows that the use of EML can be a valuable tool for the scholar who investigates interlanguage facts and language transfer.

MINECORE is a recently proposed decision-theoretic algorithm for technology-assisted review that attempts to minimise the expected costs of review for responsiveness and privilege in e-discovery. In MINECORE, two probabilistic classifiers that classify documents by responsiveness and by privilege, respectively, generate posterior probabilities. These latter are fed to an algorithm that returns as output, after applying risk minimization, two ranked lists, which indicate exactly which documents the annotators should review for responsiveness and which documents they should review for privilege. In this paper we attempt to find out if the performance of MINECORE can be improved (a) by using, for the purpose of training the two classifiers, active learning (implemented either via relevance sampling, or via uncertainty sampling, or via a combination of them) instead of passive learning, and (b) by using the Saerens-Latinne-Decaestecker algorithm to improve the quality of the posterior probabilities that MINECORE receives as input. We address these two research questions by carrying out extensive experiments on the RCV1-v2 benchmark. We make publicly available the code and data for reproducing all our experiments.

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.

This open access book provides an introduction and an overview of learning to quantify (a.k.a. "quantification"), i.e. the task of training estimators of class proportions in unlabeled data by means of supervised learning. In data science, learning to quantify is a task of its own related to classification yet different from it, since estimating class proportions by simply classifying all data and counting the labels assigned by the classifier is known to often return inaccurate ("biased") class proportion estimates. The book introduces learning to quantify by looking at the supervised learning methods that can be used to perform it, at the evaluation measures and evaluation protocols that should be used for evaluating the quality of the returned predictions, at the numerous fields of human activity in which the use of quantification techniques may provide improved results with respect to the naive use of classification techniques, and at advanced topics in quantification research. The book is suitable to researchers, data scientists, or PhD students, who want to come up to speed with the state of the art in learning to quantify, but also to researchers wishing to apply data science technologies to fields of human activity (e.g., the social sciences, political science, epidemiology, market research) which focus on aggregate ("macro") data rather than on individual ("micro") data.

The Artificial Intelligence for Media and Humanities laboratory (AIMH) has the mission to investigate and advance the state of the art in the Artificial Intelligence field, specifically addressing applications to digital media and digital humanities, and taking also into account issues related to scalability. This report summarize the 2022 activities of the research group.

The Saerens-Latinne-Decaestecker (SLD) algorithm is a method whose goal is improving the quality of the posterior probabilities (or simply "posteriors") returned by a probabilistic classifier in scenarios characterized by prior probability shift (PPS) between the training set and the unlabelled ("test") set. This is an important task, (a) because posteriors are of the utmost importance in downstream tasks such as, e.g., multiclass classification and cost-sensitive classification, and (b) because PPS is ubiquitous in many applications. In this paper we explore whether using SLD can indeed improve the quality of posteriors returned by a classifier trained via active learning (AL), a class of machine learning (ML) techniques that indeed tend to generate substantial PPS. Specifically, we target AL via relevance sampling (ALvRS) and AL via uncertainty sampling (ALvUS), two AL techniques that are very well-known especially because, due to their low computational cost, are suitable to being applied in scenarios characterized by large datasets. We present experimental results obtained on the RCV1-v2 dataset, showing that SLD fails to deliver better-quality posteriors with both ALvRS and ALvUS, thus contradicting previous findings in the literature, and that this is due not to the amount of PPS that these techniques generate, but to how the examples they prioritize for annotation are distributed.

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 Y={y1,...,yn} 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; this is the first time that an evaluation exercise solely dedicated to quantification is organized. For both the binary setting and the single-label multiclass setting, data were provided to participants both in ready-made vector form and in raw document form. In this overview article we describe the structure of the lab, we report the results obtained by the participants in the four proposed tasks and subtasks, and we comment on the lessons that can be learned from these results.

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 Y = {y1 , ..., yn } 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; this is the first time that an evaluation exercise solely dedicated to quantification is organized. For both the binary setting and the single-label multiclass setting, data were provided to participants both in ready-made vector form and in raw document form. In this overview article we describe the structure of the lab, we report the results obtained by the participants in the four proposed tasks and subtasks, and we comment on the lessons that can be learned from these results.

We present the Interactive Classification System (ICS), a web-based application that supports the activity of manual text classification. The application uses machine learning to continuously fit automatic classification models that are in turn used to actively support its users with classification suggestions. The key requirement we have established for the development of ICS is to give its users total freedom of action: they can at any time modify any classification schema and any label assignment, possibly reusing any relevant information from previous activities. We investigate how this requirement challenges the typical scenarios faced in machine learning research, which instead give no active role to humans or place them into very constrained roles, e.g., on-demand labeling in active learning processes, and always assume some degree of batch processing of data. We satisfy the "total freedom" requirement by designing an unobtrusive machine learning model, i.e., the machine learning component of ICS as an unobtrusive observer of the users, that never interrupts them, continuously adapts and updates its models in response to their actions, and it is always available to perform automatic classifications. Our efficient implementation of the unobtrusive machine learning model combines various machine learning methods and technologies, such as hash-based feature mapping, random indexing, online learning, active learning, and asynchronous processing.

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 litera- ture 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 set- ting 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.

The Artificial Intelligence for Media and Humanities laboratory (AIMH) has the mission to investigate and advance the state of the art in the Artificial Intelligence field, specifically addressing applications to digital media and digital humanities, and taking also into account issues related to scalability. This report summarize the 2021 activities of the research group.

QuaPy is an open source framework for Quantification (a.k.a. Supervised Prevalence Estimation) written in Python. QuaPy roots on the concept of data sample, and provides implementations of most important concepts in quantification literature, such as the most important quantification baselines, many advanced quantification methods, quantification-oriented model selection, many evaluation measures and protocols used for evaluating quantification methods. QuaPy also integrates commonly used datasets and offers visualization tools for facilitating the analysis and interpretation of results.

TwiGet is a python package for the management of the queries on filtered stream of the Twitter API, and the collection of tweets from it. It can be used as a command line tool (twiget-cli) or as a python class (TwiGet).

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 outpermsngformed 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 GitHub, and can be installed via pip.

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.

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.

Obtaining high-quality labelled data for training a classifier in a new application domain is often costly. Transfer Learning(a.k.a. "Inductive Transfer") tries to alleviate these costs by transferring, to the "target"domain of interest, knowledge available from a different "source"domain. In transfer learning the lack of labelled information from the target domain is compensated by the availability at training time of a set of unlabelled examples from the target distribution. Transductive Transfer Learning denotes the transfer learning setting in which the only set of target documents that we are interested in classifying is known and available at training time. Although this definition is indeed in line with Vapnik's original definition of "transduction", current terminology in the field is confused. In this article, we discuss how the term "transduction"has been misused in the transfer learning literature, and propose a clarification consistent with the original characterization of this term given by Vapnik. We go on to observe that the above terminology misuse has brought about misleading experimental comparisons, with inductive transfer learning methods that have been incorrectly compared with transductive transfer learning methods. We then, give empirical evidence that the difference in performance between the inductive version and the transductive version of a transfer learning method can indeed be statistically significant (i.e., that knowing at training time the only data one needs to classify indeed gives an advantage). Our clarification allows a reassessment of the field, and of the relative merits of the major, state-of-The-Art algorithms for transfer learning in text classification.