The evolution of Visual Large Language Models (VLLMs) has revolutionized the automatic understanding of Visually Rich Documents (VRDs), which contain both textual and visual elements. Although VLLMs excel in Visual Question Answering (VQA) on multi-page VRDs, their ability to detect unanswerable questions is still an open research question. Our research delves into the robustness of the VLLMs to plausible yet unanswerable questions, i.e., questions that appear valid but cannot be answered due to subtle corruptions caused by swaps between related concepts or plausible question formulations. Corruptions are generated by replacing the original natural language entities with other ones of the same type, belonging to different document elements, and in different layout positions or pages of the related document. To this end, we present VRD-UQA (VISUALLY RICH DOCUMENT UNANSWERABLE QUESTION ANSWERING), a benchmark for evaluating VLLMs' resilience to plausible yet unanswerable questions across multiple dimensions. It automatically alters the questions of existing VQA datasets consisting of multi-page VRDs, verifies their unanswerability using a VLLM-as-a-judge approach, and then thoroughly evaluates VLLMs' performance. Experiments, run on 12 models, analyze: (1) The VLLMs' accuracy in detecting unanswerable questions at both page and document levels; (2) The effect of different types of corruption (NLP entity, document element, layout); (3) The effectiveness of different knowledge injection strategies based on in-context learning (OCR, multi-page selection, or the possibility of unanswerability). Our findings reveal VLLMs' limitations and demonstrate that VRD-UQA can serve as an evaluation framework for developing resilient document VQA systems.

Key information extraction (KIE) from visually rich documents, such as receipts and forms, involves a deep understanding of textual, visual, and layout feature information. Transformers fine-tuned for KIE achieve state-of-the-art performance but lack generality and portability across different domains. In contrast, vision large language models (VLLMs) offer higher flexibility and zero-shot capability but fall short with domain-specific layout relations unless performing a resource-demanding supervised fine-tuning. To reach the best compromise solution between lightweight models and VLLMs, we propose KIEPrompter, a cost-effective LLM-based KIE approach that leverages the predictions of lightweight models as external knowledge injected into VLLM prompts. By incorporating these auxiliary predictions, VLLMs are guided to attend relevant multimodal content without ad hoc training. The accuracy results achieved by KIEPrompter in three benchmark document collections are superior to those of VLLMs in both zero-shot and layout-sensitive scenarios. We compare various strategies for incorporating lightweight model predictions, ranging from coarse-grained predictions without explicit confidence scores to fine-grained per-element network logits. We also demonstrate that our approach is robust to the absence of specific classes in trained lightweight models, as the VLLMs' pre-training compensates for the limited generality of lightweight models.

Textual Entailment Recognition (TER) aims to predict whether a pair of premise-hypothesis sentences represents an entailment, a contradiction, or none of the above. Addressing TER in the presence of figurative language is particularly challenging because words are used in a way that deviates from the conventional order and meaning. In this work, we investigate the capabilities of Large Language Models (LLMs) to address TER and generate textual explanations of TER predictions. First, we evaluate LLM performance in Zero- and Few-Shot Learning settings, with and without using Chain-of-Thought prompting. After identifying the best prompts, we highlight the settings in which in-context learning is beneficial. The closed-source models GPT-3.5 Turbo and GPT-4o show unexpected limitations compared to significantly smaller open-source LLMs. Next, we thoroughly analyze the effect of LLM Fine-Tuning, showing substantial improvements in the quality of TER explanations compared to Zero- and Few-Shot Learning. Notably, 9 billion parameter open-source LLMs demonstrate again competitive performance against larger closed-source models. Finally, we compare our LLM-based approach with the state-of-the-art DREAM-FLUTE and Cross-Task architectures. The results show significant performance improvements, particularly in the quality of the generated explanations.

Large Language Models (LLMs) have demonstrated robust performance in Semantic Parsing (SP) for well-defined queries with unambiguous intent and answerable responses.However, practical user questions frequently deviate from these ideal conditions, challenging the applicability of existing benchmarks.To address this issue, we introduce SQUAB, an automatic dataset generator of Ambiguous and Unanswerable questions.SQUAB generates complex, annotated SP tests using a blend of SQL and LLM capabilities.Results show that SQUAB reduces test generation costs by up to 99% compared to human-based solutions while aligning with real-world question patterns.Furthermore, these tests challenge LLM performance while revealing disparities between public and proprietary datasets.This highlights the need for a dynamic, automatic dataset generator as SQUAB.The code is designed for user extension to accommodate new ambiguous and unanswerable patterns and is available at https: //github.com/spapicchio/squab.

This paper presents a Retrieval Augmented Generation (RAG) pipeline adapted to simulate the process of resolving legal cases. It introduces two novel approaches to enhance the RAG pipeline for legal reasoning: 1) problem generation, which automatically creates fictitious legal cases to improve retrieval performance, and 2) query rewriting, which reformulates user queries to better align with relevant passages. These methods are evaluated on a legal reasoning dataset derived from real U.S. civil procedure cases. Experiments conducted across different retrieval methods (keyword search, tensor search, and hybrid search with re-ranking) and language models (GPT-4, SAULM, Llama-3 8B) highlight the benefits of the more advanced RAG pipeline. Notably, the query rewriting approach consistently improves performance by enhancing the alignment between queries and passages.

This paper presents a real-world deployment case study on using unsupervised anomaly detection for Anti-Money Laundering (AML).Using more than 2 billion anonymized bank transactions that Intesa Sanpaolo, a primary Italian financial institution, registered over 8 months, we developed, tuned and deployed a machine learning pipeline in production.Experts from Intesa Sanpaolo validated the performance of our approach against the institution's traditional rule-based system and checked new real-world cases the system allowed them to identify.Besides increasing both precision and recall by a factor of 6 in the detection of high-risk cases, our pipeline raises 200+ additional alerts during the 8-month period, manually identified by branch managers, but missed by the rulebased system.More importantly, a manual inspection of 100 new unseen cases revealed 28 significant previously unreported cases.The pipeline, now fully deployed in Intesa Sanpaolo's Transaction Monitoring system, highlights the advantages of machine learning over traditional approaches typically adopted in this traditionally very conservative sector.

ChatGPT potential value in education is broadly recognized and many studies report experiments of its use inside or outside the classroom by students and teachers. On the other hand, the use of ChatGPT rises lots of concerns about well-known problems such as hallucination, plagiarism, overreliance, or misinformation. It is of primary importance to teach students a correct and constructive use of ChatGPT and a critical approach to its returned outputs. The paper presents a classroom experience where students were asked to interact with ChatGPT in the context of a database course. The declared challenge for the students was, given a set of predefined relational database schemata, to invent questions for ChatGPT and try to force wrong SQL solutions. Students had to record the question, the ChatGPT solution, their solution, and the comments about the eventual ChatGPT syntactical and/or semantical errors. This gamification approach was meant to enhance students' motivation, but the main teachers' goal was to make them reflect critically (i) on ChatGPT output, experiencing that it does make mistakes, (ii) on the interpretation of ChatGPT errors, and (iii) on the possible strategies for forcing ChatGPT errors. The experiment involved 166 B.S. students in Engineering and the collected data have been analyzed under different points of view to get an insight into the approach and the critical attitude of the students. The paper reports the results of this analysis and discusses the impact of the activity on learning by analyzing the correlation between students' participation and exam performance.

Social media content, such as internet memes or tweets, are nowadays largely or mainly multimodal. Machine learning models often need to jointly process images and text to solve complex tasks such as hate speech detection or sentiment analysis. For example, the misogyny of a meme cannot be accurately predicted while considering the visual and textual modalities separately. Similarly, sentiment annotations for tweets’ images and text can be discordant. Detecting the samples with inconsistent modality contributions is particularly relevant to analyze machine learning model performance and explain classification errors. In this paper, we formalize the types of cross-modal consistency by differentiating between consistent cases and not. Cross-modal consistency denotes whether all modalities agree on the label (i.e., full consistency) or not (i.e., inconsistency). When the visual and textual modalities are discordant, we distinguish the cases in which a joint analysis of multimodal features is sufficient to solve the issue from those requiring a human agreement (i.e., NOR consistency). We also propose a CLIP-based architecture to predict the cross-modal consistency types and identify the modalities causing the inconsistency. The results achieved on benchmark datasets show that cross-modal consistency annotation is cost-effective, i.e., it provides relevant insights into model predictions while requiring a limited extra human effort.

Formal communications such as public calls, announcements, or regulations are supposed to exhibit respect for diversity in terms of gender, race, age, and disability. However, human writers often lack adequate inclusive writing skills. For instance, they tend to overuse the masculine as a neutral form, mainly because they are self-trained on biased text examples. To overcome this issue, we propose to leverage Generative Artificial Intelligence to support inclusive language writing. Focusing on formal Italian communications, we have designed and developed an AI-assisted tool for non-inclusive text detection and reformulation. Thanks to the joint work with a team of linguistic experts, we first define a set of linguistic criteria necessary to model inclusive writing forms in Italian. Based on these criteria, we collect and annotate a dataset of Italian administrative documents enriched with fine-grained inclusive annotations. Finally, we train deep learning models on the collected data for non-inclusive language detection and inclusive language reformulation tasks. We perform quantitative and human-driven evaluations on the trained models. The best detection model correctly classifies 89% of the sentences, whereas the best reformulation model produces 73% fully correct reformulations. Both models have been integrated into a writing assistance tool acting as a text proofreader and self-learning tool for non-expert writers, namely Inclusively . Once a non-inclusive piece of text is detected, the proposed approach suggests inclusive reformulations. The tool also provides explanations of the models’ outputs to increase system transparency. Furthermore, it allows expert end-users to provide further annotations for system fine-tuning. The trained models and the writing assistance tool are publicly available for research purposes.

Tabular Representation Learning (TRL) and Large Language Models (LLMs) have become established for tackling Question Answering (QA) and Semantic Parsing (SP) tasks on tabular data. State-of-the-art models are pre-trained and evaluated on large open-domain datasets. However, the performance on existing QA and SP benchmarks is not necessarily representative of that achieved on proprietary data as the characteristics of the input and the complexity of the posed queries show high variability. To tackle this challenge, our goal is to allow end-users to evaluate TRL and LLM performance on their own proprietary data. We present Query-Aided TRL CHecklist (QATCH), a toolbox to automatically generate a testing checklist tailored to QA and SP. QATCH provides a testing suite highlighting models’ strengths and weaknesses on relational tables unseen at training time. The proposed toolbox relies on a SQL query generator that crafts tests of varying types and complexity including, amongst others, tests on null values, projection, selections, joins, group by, and having clauses. QATCH also supports a set of general cross-task performance metrics providing more insights into SQL-related model capabilities than currently used metrics. The empirical results, achieved by state-of-the-art TRL models and LLMs, show substantial performance differences (1) between existing benchmarks and proprietary data, (2) across queries of different complexity.

The discovery of sequential patterns from spatiotemporal data is known to be a very complex data mining task. The relevance of spatiotemporal patterns to study event correlations in mobility data is established. Prior works addressed either the separate analysis of spatial and temporal dependencies among data, such as the co-location of events, or the study of the joint spatiotemporal properties of the trajectories observed over a region of interest. The aim of this paper is instead to overcome existing approaches by extracting sequences of discrete events showing spatiotemporally invariant properties. For example, ifan arbitrary bike sharing station becomes full (all its docks are used)thenwe will observe an increase in the occupancy level of the bike sharing stations in the surrounding area within ten minutes. We denote such a new pattern as a SpatioTemporally Invariant (STInv) event pattern because we observe several instances in the source data differing just in spatiotemporal shifts. We also propose a new algorithm to mine STInvs based on a prefix-projected sequential pattern growth approach and different quality metrics to quantify the contribution of the spatial invariance. The proposed approach is empirically evaluated on two mobility datasets related to a bike sharing system and traffic data. The results confirm the usability of the proposed solution in real-world scenarios.

In this paper, we present the design, development, and testing of AI chatbots to support the help desk service provided by the Recruitment and Admissions Unit at Politecnico di Torino, a technical university located in the north-west of Italy. We explore the use of data-driven, AI-based conversational agents providing targeted responses to applicants’ queries based on both the past student-office interactions through a ticketing system and a collection of Frequently Asked Questions (FAQs). With an ever-increasing number of requests from international applicants (20k+ requests from 100+ countries since January 2024), the adoption of AI-based solutions allows a significant reduction of the average waiting time per request, fostering the application, enrollment, and integration of foreign students coming from a variety of different countries. We develop separate chatbot systems to handle FAQs and manage inquiries submitted through the university’s ticketing system. We explore the use of intent-based and generative approaches as well as a combination of the two. Our findings indicate that the intent-based approach excels in handling FAQs and well-defined tickets, whereas the generative-only strategy is more suitable for open-ended requests.

The use of Large Language Models (LLMs) such as OpenAI ChatGPT to enhance teachers' and learners' experience has become established. The impressive capabilities of ChatGPT in solving Text2SQL problems prompts their use in database courses to solve SQL exercises. In this paper, we dig deep into ChatGPT abilities applied to SQL exercises. We quantitatively and qualitatively evaluate the performance of a ChatGPT-as-a-SQL-assistant on benchmark data, with particular attention paid to its ability to correctly detect syntactic and semantic errors, provide insightful judgment explanations, and assign grades comparable to those of human teachers. Furthermore, we also analyze the benefits of leveraging few-shot learning to adapt LLM responses to the expectation.

Thanks to the rapid progress of the digitalization process, Visually-Rich Documents (VRDs) such as PDF files or scanned documents have become among the most widespread sources of knowledge. However, Question Answering on VRDs is challenged by the presence of multi-page relationships between document elements such as tables, figures, sections. This paper addresses a specific Visual Question Answering subtask from VDRs where answer generation leverages pairwise element relations in multi-page documents. We explore the performance of text-only and multimodal Transformer-based architectures as well as open-source Large Language Models. The results show that multimodal Transformers outperform the other tested methods, particularly when training samples contain explicit textual references to the elements in the document layout.

The evaluation of spoken language understanding (SLU) systems is often restricted to assessing their global performance or examining predefined subgroups of interest. However, a more detailed analysis at the subgroup level has the potential to uncover valuable insights into how speech system performance differs across various subgroups. In this work, we identify biased data subgroups and describe them at the level of user demographics, recording conditions, and speech targets. We propose a new task-, model- and dataset-agnostic approach to detect significant intra- and cross-model performance gaps. We detect problematic data subgroups in SLU models by leveraging the notion of subgroup divergence. We also compare the outcome of different SLU models on the same dataset and task at the subgroup level. We identify significant gaps in subgroup performance between models different in size, architecture, or pre-training objectives, including multi-lingual and mono-lingual models, yet comparable to each other in overall performance. The results, obtained on two SLU models, four datasets, and three different tasks–intent classification, automatic speech recognition, and emotion recognition–confirm the effectiveness of the proposed approach in providing a nuanced SLU model assessment.

Text Style Transfer (TST) is a relevant branch of natural language processing that aims to control the style attributes of a piece of text while preserving its original content. To address TST in the absence of parallel data, Cycle-consistent Generative Adversarial Networks (CycleGANs) have recently emerged as promising solutions. Existing CycleGAN-based TST approaches suffer from the following limitations: (1) They apply self-supervision, based on the cycle-consistency principle, in the latent space. This approach turns out to be less robust to mixed-style inputs, i.e., when the source text is partly in the original and partly in the target style; (2) Generators and discriminators rely on recurrent networks, which are exposed to known issues with long-term text dependencies; (3) The target style is weakly enforced, as the discriminator distinguishes real from fake sentences without explicitly accounting for the generated text's style. We propose a new CycleGAN-based TST approach that applies self-supervision directly at the sequence level to effectively handle mixed-style inputs and employs Transformers to leverage the attention mechanism for both text encoding and decoding. We also employ a pre-trained style classifier to guide the generation of text in the target style while maintaining the original content's meaning. The experimental results achieved on the formality and sentiment transfer tasks show that our approach outperforms existing ones, both CycleGAN-based and not (including an open-source Large Language Model), on benchmark data and shows better robustness to mixed-style inputs.

The use of Shapley Values (SVs) to explain machine learning model predictions is established. Recent research efforts have been devoted to generating efficient Neural Network-based SVs estimates. However, the variability of the generated estimates, which depend on the selected data sampling, model, and training parameters, brings the reliability of such estimates into question. By leveraging the concept of Interval SVs, we propose to incorporate SVs uncertainty directly into the learning process. Specifically, we explain ensemble models composed of multiple predictors, each one generating potentially different outcomes. Unlike all existing approaches, the explainer design is tailored to Interval SVs learning instead of SVs only. We present three new Network-based explainers relying on different ISV paradigms, i.e., a Multi-Task Learning network inspired by the Shapley value's weighted least squares characterization and two Interval Shapley-Like Value Neural estimators. The experiments thoroughly evaluate the new approaches on ten benchmark datasets, looking for the best compromise between intervals’ accuracy and explainers’ efficiency.

Industrial vehicles equipped with CAN bus devices transmit large volumes of IoT signals. The analysis of CAN bus data can be helpful to monitor the current vehicles’ workload, namely the vehicle duty levels, in an automated fashion. Despite the use of machine learning techniques to automatically detect vehicle duty levels is particularly appealing, existing approaches are challenged by the high cost of human data annotation and by the high heterogeneity of the analyzed vehicles types and models. In this paper, we present a self-supervised approach to automatically detect vehicles’ duty levels based on contrastive learning. The multivariate CAN Bus signals are first divided into fixed-sized segments and then embedded into a vector space shared by all vehicles of the same model by leveraging a constrastive approach with a mixup augmentation strategy. The key idea is to embed similar segments in close proximity by self-learning a model-specific clustering, which allows automatic duty level assignment with minimal human supervision. We validate the proposed approach in a real industrial use case, analyzing CAN Bus data acquired from test heavy-duty vehicles. Data were provided by a multinational Internet-of-Things company specialized in telematics solution. The experiments show clustering performance superior to state-of-the-art models and as well as an higher ability to differentiate between Moving and Working duty levels.

End-to-End Spoken Language Understanding models are generally evaluated according to their overall accuracy, or separately on (a priori defined) data subgroups of interest. We propose a technique for analyzing model performance at the subgroup level, which considers all subgroups that can be defined via a given set of metadata and are above a specified minimum size. The metadata can represent user characteristics, recording conditions, and speech targets. Our technique is based on advances in model bias analysis, enabling efficient exploration of resulting subgroups. A fine-grained analysis reveals how model performance varies across sub-groups, identifying modeling issues or bias towards specific subgroups.We compare the subgroup-level performance of models based on wav2vec 2.0 and HuBERT on the Fluent Speech Commands dataset. The experimental results illustrate how subgroup-level analysis reveals a finer and more complete picture of performance changes when models are replaced, automatically identifying the subgroups that most benefit or fail to benefit from the change.

Recent large-scale Spoken Language Understanding datasets focus predominantly on English and do not account for language-specific phenomena such as particular phonemes or words in different lects.We introduce ITALIC, the first largescale speech dataset designed for intent classification in Italian.The dataset comprises 16,521 crowdsourced audio samples recorded by 70 speakers from various Italian regions and annotated with intent labels and additional metadata.We explore the versatility of ITALIC by evaluating current state-of-the-art speech and text models.Results on intent classification suggest that increasing scale and running language adaptation yield better speech models, monolingual text models outscore multilingual ones, and that speech recognition on ITALIC is more challenging than on existing Italian benchmarks.We release both the dataset and the annotation scheme to streamline the development of new Italian SLU models and language-specific datasets.

Automated text summarization is particularly important in the legal domain due to the length and inherent complexity of the analyzed documents. The Legal AI community has already started to address the text summarization problem. However, most existing approaches focus on English-written documents. Up to now, limited efforts have been devoted to summarizing Italian legal documents. Existing approaches extract portions of existing content without rephrasing them. To bridge this gap, in this work we aim at generating abstractive summaries of Italian legal news. We propose to condense the original news content into different summary types, i.e., an abstract, a title, or a subheader. We benchmark different state-of-the-art summarization models to generate abstractive summaries of Italian legal news. We also investigate the suitability of augmented models capable of handling long Italian documents. The experimental results achieved on a proprietary Italian dataset show the effectiveness of abstractive models in generating fairly accurate summaries and the importance of using larger contextual windows to generate news abstracts.

Document classification is helpful for law professionals to improve content browsing and retrieval. Pretrained Language Models, such as BERT, have become established for legal document classification. However, legal content is quite diversified. For example, documents vary in length from very short maxims to relatively long judgements; certain document types are rich of domain-specific expressions and can be annotated with multiple labels from domain-specific taxonomies. This paper studies to what extent existing pretrained models are suited to the legal domain. Specifically, we examine a real business case focused on Italian legal document classification. On a proprietary dataset with thousands of diversified categories (e.g., legal judgements, maxims, and legal news) we explore the use of Pretrained Language Models adapted to handle various content types. We collect both quantitative and qualitative results, highlighting best and worst cases, anomalous categories, and limitations of currently available models.

Table Representation Learning (TRL) models are commonly pre-trained on large open-domain datasets comprising millions of tables and then used to address downstream tasks. Choosing the right TRL model to use on proprietary data can be challenging, as the best results depend on the content domain, schema, and data quality. Our purpose is to support end-users in testing TRL models on proprietary data in two established SQL-centric tasks, i.e., Question Answering (QA) and Semantic Parsing (SP). We present QATCH (Query-Aided TRL Checklist), a toolbox to highlight TRL models’ strengths and weaknesses on relational tables unseen at training time. For an input table, QATCH automatically generates a testing checklist tailored to QA and SP. Checklist generation is driven by a SQL query engine that crafts tests of different complexity. This design facilitates inherent portability, allowing the checks to be used by alternative models. We also introduce a set of cross-task performance metrics evaluating the TRL model’s performance over its output. Finally, we show how QATCH automatically generates tests for proprietary datasets to evaluate various state-of-the-art models including TAPAS, TAPEX, and CHATGPT.

Abstract Driving the decisions of stock market investors is among the most challenging financial research problems. Markowitz’s approach to portfolio selection models stock profitability and risk level through a mean–variance model, which involves estimating a very large number of parameters. In addition to requiring considerable computational effort, this raises serious concerns about the reliability of the model in real-world scenarios. This paper presents a hybrid approach that combines itemset extraction with portfolio selection. We propose to adapt Markowitz’s model logic to deal with sets of candidate portfolios rather than with single stocks. We overcome some of the known issues of the Markovitz model as follows: (i) Complexity: we reduce the model complexity, in terms of parameter estimation, by studying the interactions among stocks within a shortlist of candidate stock portfolios previously selected by an itemset mining algorithm. (ii) Portfolio-level constraints: we not only perform stock-level selection, but also support the enforcement of arbitrary constraints at the portfolio level, including the properties of diversification and the fundamental indicators. (iii) Usability: we simplify the decision-maker’s work by proposing a decision support system that enables flexible use of domain knowledge and human-in-the-loop feedback. The experimental results, achieved on the US stock market, confirm the proposed approach’s flexibility, effectiveness, and scalability.

Density-based clustering focuses on defining clusters consisting of contiguous regions characterized by similar densities of points. Traditional approaches identify core points first, whereas more recent ones initially identify the cluster borders and then propagate cluster labels within the delimited regions. Both strategies encounter issues in presence of multi-density regions or when clusters are characterized by noisy borders. To overcome the above issues, we present a new clustering algorithm that relies on the concept of bridge point. A bridge point is a point whose neighborhood includes points of different clusters. The key idea is to use bridge points, rather than border points, to partition points into clusters. We have proved that a correct bridge point identification yields a cluster separation consistent with the expectation. To correctly identify bridge points in absence of a priori cluster information we leverage an established unsupervised outlier detection algorithm. Specifically, we empirically show that, in most cases, the detected outliers are actually a superset of the bridge point set. Therefore, to define clusters we spread cluster labels like a wildfire until an outlier, acting as a candidate bridge point, is reached. The proposed algorithm performs statistically better than state-of-the-art methods on a large set of benchmark datasets and is particularly robust to the presence of intra-cluster multiple densities and noisy borders.

Transaction Monitoring is one of the main labor-intensive tasks of anti-financial crime and it requires to scrutinise billions of transactions per month against possible crimes. The first step in the process is the correct identification of the involved parties. This foundational step defines the focal entities on which transaction monitoring algorithms rely to spot suspicious events. Unfortunately, the loose syntax of protocols and the free text fields of inter-banking communications make party disambiguation particularly challenging. The first step of a fully automated data-driven strategy is thus the detection of the actual entity owning or using a given account.In this paper, we leverage data-driven techniques to identify and disambiguate the owners of accounts involved in cross-border international transactions when a Financial Institution only knows a minority fraction of such parties as its own customers. For this, we propose a data science pipeline relying on hierarchical clustering to capture similarities among names of parties involved in actual transactions. We test and tune the proposed approach using a large, real-world, multi-language, proprietary dataset of actual international transactions. Our highly parallel implementation completes the identification of parties that share an account and identifies all accounts owned by a party with f-score higher than 0.8.

Learning environments foster the exchange of large amounts of data among learners and teachers. Summarization techniques leverage information retrieval and machine learning techniques to condense the key information hidden in large data collections into actionable summaries. Their integration into existing education technology systems is particularly appealing as it enables smart, automated solutions to challenging learning tasks such as content curation, accessibility, and personalization. This paper presents a general-purpose summarization-based methodology to learn. It aims at extending the current ed-ucational learning systems by envisaging the integration of summarization methods at different learning stages. Specifically, it tailors the output summaries to different end-users (either teachers or learners), content types (e.g., text, audio, video), and learning goals. With the goal of making the devised methodology actionable, the paper also examines the current role of summa-rization in the learning process and highlights the open directions and perspectives.

The Matrix Profile is an established data structure used to efficiently compute similarities among subseries in large time series datasets. State-of-the-art anomaly detection methods rely on the Matrix Profile to identify abnormal behaviors in an unsupervised fashion. Their time complexity is quadratic with the series length. This paper aims at further enhancing the scalability of Matrix Profile-based anomaly detection approaches. The key idea is to restrict the search to a particular kind of anomalies, namely the Contrasting Consecutive Patterns. They consist of two consecutive, highly dissimilar subsequences observed in the input time series data. To efficiently extract such anomalous patterns, we present a new algorithm that scales linearly with the time series length. The experimental results show the efficiency and usability of the proposed approach on real-world time series datasets.

Discovering patterns that represent key spatial or temporal dependencies among data is a well-known exploratory data mining task. However, prior works either separately analyze spatial and temporal dependencies or discover joint spatiotemporal properties of specific trajectories observed over a region of interest. With the goal of generalizing the information provided by spatiotemporal patterns, in this paper we extract sequences of discrete events showing spatiotemporally invariant properties. We seek patterns whose corresponding instances in the source data differ only due to an invariant spatiotemporal transformation. We denote such a new type of patterns as SpatioTemporally Invariant. We also propose an efficient algorithm to mine STInvs and validate its efficiency and effectiveness on real data.

Mosquitoes are a major global health problem. They are responsible for the transmission of diseases and can have a large impact on local economies. Monitoring mosquitoes is therefore helpful in preventing the outbreak of mosquito-borne diseases. In this paper, we propose a novel data-driven approach that leverages Transformer-based models for the identification of mosquitoes in audio recordings. The task aims at detecting the time intervals corresponding to the acoustic mosquito events in an audio signal. We formulate the problem as a sequence tagging task and train a Transformer-based model using a real-world dataset collecting mosquito recordings. By leveraging the sequential nature of mosquito recordings, we formulate the training objective so that the input recordings do not require fine-grained annotations. We show that our approach is able to outperform baseline methods using standard evaluation metrics, albeit suffering from unexpectedly high false negatives detection rates. In view of the achieved results, we propose future directions for the design of more effective mosquito detection models.

Location-Based Social Networks can be profitably exploited to characterize citizens’ activities in urban environments. However, collecting LBSN is potentially challenging due to privacy concerns, connectivity issues, and potential imbalances in LBSN service usage. We propose to complement LBSN data with mobility data in the analysis of citizens’ activities in urban areas. Unlike the explicit insights provided by LBSN users, mobility data give implicit feedback on citizens’ habits. This paper explores the spatial and temporal conditions under which user habits are coherent according to both sources and reports the most reliable common sequences of visited categories of Points-Of-Interests. To this aim, it relies on a multidimensional model in which recurrent citizens’ activities are described by a new pattern type, namely the generalized activity pattern. It also detects the eventual presence of bias between LBSN and mobility user activities by customizing the established Statistical Parity metric. The motivations behind the detected bias are explained in terms of combinations of POI categories that are most likely to be the main causes. We evaluate the proposed approach on real-world data achieved from Foursquare check-ins, taxi service, and free-floating car sharing. The results highlight not only the complementarity of the data sources regarding specific POI categories, but also their interchangeability in many spatio-temporal conditions.

Preserving diversity and inclusion is becoming a compelling need in both industry and academia. The ability to use appropriate forms of writing, speaking, and gestures is not widespread even in formal communications such as public calls, public announcements, official reports, and legal documents. The improper use of linguistic expressions can foment unacceptable forms of exclusion, stereotypes as well as forms of verbal violence against minorities, including women. Furthermore, existing machine translation tools are not designed to generate inclusive content.The present paper investigates a joint effort of the research communities of linguistics and Deep Learning Natural Language Understanding in fighting against non-inclusive, prejudiced language forms. It presents a methodology aimed at tackling the improper use of language in formal communication, with a particular attention paid to Romanic languages (Italian, in particular). State-of-the-art Deep Language Modeling architectures are exploited to automatically identify non-inclusive text snippets, suggest alternative forms, and produce inclusive text rephrasing. A preliminary evaluation conducted on a benchmark dataset shows promising results, i.e., 85% accuracy in predicting inclusive/non-inclusive communications.

Disseminating the main research findings is one of the main requirements to become a successful researcher. Presentation slides are the most common way to present paper content. To support researchers in slide preparation, the NLP research community has explored the use of summarization techniques to automatically generate a draft of the slides consisting of the most salient sentences or phrases. State-of-the-art methods adopt a supervised approach, which first estimates global content relevance using a set of training papers and slides, then performs content selection by optimizing also section-level coverage. However, in several domains and contexts there is a lack of training data, which hinders the use of supervised models. This paper focuses on addressing the above issue by applying unsupervised summarization methods. They are exploited to generate sentence-level summaries of the paper sections, which are then refined by applying an optimization step. Furthermore, it evaluates the quality of the output slides by taking into account the original paper structure as well. The results, achieved on a benchmark collection of papers and slides, show that unsupervised models performed better than supervised ones on specific paper facets, whereas they were competitive in terms of overall quality score.

Due to the rapid growth of multimedia data and the diffusion of remote and mixed learning, teaching sessions are becoming more and more multi-modal. To deepen the knowledge of specific topics, learners can be interested in retrieving educational videos that complement the textual content of teaching books. However, retrieving educational videos can be particularly challenging when there is a lack of metadata information. To tackle the aforesaid issue, this paper explores the joint use of Deep Learning and Natural Language Processing techniques to retrieve cross-media educational resources (i.e., from text snippets to videos and vice versa). It applies NLP techniques to both the audio transcript of the videos and to the text snippets in the books in order to quantify the semantic relationships between pairs of educational resources of different media types. Then, it trains a Deep Learning model on top of the NLP-based features. The probabilities returned by the Deep Learning model are used to rank the candidate resources based on their relevance to a given query. The results achieved on a real collection of educational multimodal data show that the proposed approach performs better than state-of-the-art solutions. Furthermore, a preliminary attempt to apply the same approach to address a similar retrieval task (i.e., from text to image and vice versa) has shown promising results.

Quantifying the effects of adverse weather conditions on road traffic is crucial for several reasons, among which monitoring traffic safety, managing vehicle fleets, and ensuring autonomous vehicle security and performance. This paper studies the incidence of adverse weather effects on road traffic safety. The aim is to assign a unified risk level to all spatial regions sharing the same context (e.g., the highly populated areas), which reflects the impact of adverse weather events on road accident occurrences. The proposed approach relies on an incomplete set of historical accident data, which report weather-related accident occurrences in specific risky areas. To estimate the percontext risk level we propose to analyze the weather element measurements acquired by meteorological stations spread over the analyzed area. The series of adverse weather events are embedded into a high-dimensional embedding space to enable the identification of temporal event patterns similar to those observed in risky areas. The experimental results show the effectiveness of the proposed approach in a real case study.

Timeline summarization aims at presenting long news stories in a compact manner. State-of-the-art approaches first select the most relevant dates from the original event timeline then produce per-date news summaries. Date selection is driven by either per-date news content or date-level references. When coping with complex event data, characterized by inherent news flow redundancy, this pipeline may encounter relevant issues in both date selection and summarization due to a limited use of news content in date selection and no use of high-level temporal references (e.g., the past month). This paper proposes a paradigm shift in timeline summarization aimed at overcoming the above issues. It presents a new approach, namely Summarize Date First, which focuses on first generating date-level summaries then selecting the most relevant dates on top of summarized knowledge. In the latter stage, it performs date aggregations to consider high-level temporal references as well. The proposed pipeline also supports frequent incremental timeline updates more efficiently than previous approaches. We tested our unsupervised approach both on existing benchmark datasets and on a newly proposed benchmark dataset describing the COVID-19 news timeline. The achieved results were superior to state-of-the-art unsupervised methods and competitive against supervised ones.

During the last years an increasing number of university-level and post-graduation courses on Data Science have been offered. Practices and assessments need specific learning environments where learners could play with data samples and run machine learning and data mining algorithms. To foster learner engagement many closed-and open-source platforms support the design of data science competitions. However, they show limitations on the ability to handle private data, customize the analytics and evaluation processes, and visualize learners' activities and outcomes. This paper presents Data Science Lab Environment (DSLE, in short), a new open-source platform to design and monitor data science competitions. DSLE offers a easily configurable interface to share training and test data, design group works or individual sessions, evaluate the competition runs according to customizable metrics, manage public and private leaderboards, monitor participants' activities and their progress over time. The paper describes also a real experience of usage of DSLE in the context of a 1st-year M.Sc. course, which has involved around 160 students.

Learning Analytics aims at supporting the understanding of learning mechanisms and their effects by means of data-driven strategies. LA approaches commonly face two big challenges: first, due to privacy reasons, most of the analyzed data are not in the public domain. Secondly, the open data collections, which come from diverse learning contexts, are quite heterogeneous. Therefore, the research findings are not easily reproducible and the publicly available datasets are often too small to enable further data analytics. To overcome these issues, there is an increasing need for integrating open learning data into unified models. This paper proposes UNIFORM, an open relational database integrating various learning data sources. It presents also a machine learning supported approach to automatically extending the integrated dataset as soon as new data sources become available. The proposed approach exploits a classifier to predict attribute alignments based on the correlations among the corresponding textual attribute descriptions. The integration phase has reached a promising quality level on most of the analyzed bechmark datasets. Furthermore, the usability of the UNIFORM data model has been demonstrated in a real case study, where the integrated data have been exploited to support learners' outcome prediction. The F1-score achieved on the integrated data is approximately 30% higher that those obtained on the original data.

The spread of e-commerce and web applications has fostered the integration of cross-domain business activities. To efficiently retrieve products and services, web directories allow customers to browse multiple-level taxonomies to find specific products or services according to a predefined categorization. Providers need to periodically update web directory lists by aligning in-house taxonomies to domain-specific hierarchies coming from external sources. However, such taxonomy mapping procedures are often semi-automatic and rely on traditional word disambiguation techniques to capture the semantics behind categories and products descriptions. Hence, the flexibility and explainability of the underlying models are quite limited. This paper proposes an automated, explainable approach to web directory taxonomy mapping based on text categorization. It exploits two complementary word-based text representations: a frequency-based representation, which captures syntactic text similarities, and an embedding one, which highlights the underlying semantic relationships among words. Since the proposed solution is purely data-driven, it can be successfully applied to business domains where there is a lack of semantic models. The frequency-based text representation has shown to be particularly suitable for driving the automated taxonomy mapping procedure, whereas the embedding space has been profitably used to provide local explanations of the category assignments.

Many universities adopt educational systems where the teacher lecture is video recorded and the video lecture is made available to students with minimum post-processing effort. These cost-effective solutions suffer from the limited amount of annotations associated with the video content, which strongly limits the usability of the service when students need to retrieve specific portions of video, e.g., to revise unclear aspects covered in the past lectures. This paper presents, as a real case study, the system developed and implemented in our university for video lecture annotation and indexing. The original video recordings, which last around 1.5 hour, are first partitioned into smaller segments and then annotated by mapping their content with the entities in a multilingual knowledge base. To this purpose, the proposed approach analyzes both the transcription of the teacher's speech and the text appearing in the video (e.g., the slide content, the note written on the whiteboard) by means of an ad hoc Named Entity Recognition and Disambiguation (NERD) step. NERD relies on a supervised classification approach tailored to the domain under analysis. More specifically, to identify the most salient entities of the knowledge base matching the video content it considers not only text similarity measures but also the semantic pertinence of the candidate entities to the main subject of the video lectures. The performance of the proposed system was validated on a ground truth against the techniques available in the general entity annotation system GERBIL. The preliminary results demonstrate the effectiveness of the proposed approach.

Sentence-based summarization aims at extracting concise summaries of collections of textual documents. Summaries consist of a worthwhile subset of document sentences. The most effective multilingual strategies rely on Latent Semantic Analysis (LSA) and on frequent itemset mining, respectively. LSA-based summarizers pick the document sentences that cover the most important concepts. Concepts are modeled as combinations of single-document terms and are derived from a term-by-sentence matrix by exploiting Singular Value Decomposition (SVD). Itemset-based summarizers pick the sentences that contain the largest number of frequent itemsets, which represent combinations of frequently co-occurring terms. The main drawbacks of existing approaches are (i) the inability of LSA to consider the correlation between combinations of multiple-document terms and the underlying concepts, (ii) the inherent redundancy of frequent itemsets because similar itemsets may be related to the same concept, and (iii) the inability of itemset-based summarizers to correlate itemsets with the underlying document concepts. To overcome the issues of both of the abovementioned algorithms, we propose a new summarization approach that exploits frequent itemsets to describe all of the latent concepts covered by the documents under analysis and LSA to reduce the potentially redundant set of itemsets to a compact set of uncorrelated concepts. The summarizer selects the sentences that cover the latent concepts with minimal redundancy. We tested the summarization algorithm on both multilingual and English-language benchmark document collections. The proposed approach performed significantly better than both itemset- and LSA-based summarizers, and better than most of the other state-of-the-art approaches.

Most engineering courses include fundamental practice activities to be performed by students in computer labs. During lab sessions, students work on solving exercises with the help of teaching assistants, who often have a hard time for guaranteeing a timely, optimized, and "democratic" support to everybody. This paper presents a learning environment to improve the experience of the lab sessions participants, both the students and the teaching assistants. In particular, the environment was designed, implemented, and experimented in the context of a database course. The application designed to support the learning environment stores all the events occurring during a SQL practice lab, i.e., task progression, query submissions, error feedback, assistance requests and interventions, and it provides information useful both for use on-the-fly and for later analysis. Thanks to the analysis of these data, the application dynamically provides teaching assistants with a graphical interface highlighting where assistance is most needed, by considering different factors such as the progression rate, the percentage of correct solutions, and the difficulties in solving the current exercise. Furthermore, the stored data allow teachers later on to analyze and to interpret the behavior of the students during the lab, and to have insights on their main mistakes and misconceptions. After describing the environment, the interfaces, and the approaches used to identify the students' teams that need timely assistance, the paper presents the results of different analyses performed using the collected data, to help the teacher better understand students' educational needs.

Many articles on the same news are daily published by online newspapers and by various social media. To ease news article exploration sentence-based summarization algorithms aim at automatically generating for each news a summary consisting of the most salient sentences in the original articles. However, since sentence selection is error-prone, the automatically generated summaries are still subject to manual validation by domain experts. If the validation step not only focuses on pruning less relevant content but also on enriching summaries with missing yet relevant sentences this activity may become extremely time consuming. The paper focuses on summarizing news articles by means of an itemset-based technique. To tune summarizer performance a relevance feedback given on sentences is exploited to drive the generation of a new, more targeted summary. The feedback indicates the pertinence of the sentences that are already in the summary. Among the words or the word combinations selected by the summarization model, those occurring in sentences with high feedback score represent concepts that may be deemed as particularly relevant. Therefore, they are exploited to drive the new sentence selection process. The proposed approach was tested on collections of news articles reporting emergency situations. The results show the effectiveness of the proposed approach.

New information and communication technologies offer today many opportunities to improve the quality of educational services in universities and in particular they allow to design and implement innovative learning models. This paper describes and validates our university blended learning model, and specifically the massive educational video service that we offer to our students since 2010. In these years, we have gathered a huge amount of detailed data about the students' access to the service, and the paper describes a number of analyses that we carried out with these data. The common goal was to find out experimentally whether the main objectives of the educational video service we had in our mind when we designed it, namely appreciation, effectiveness and flexibility, were reflected by the users' behavior. We analyzed how many students used the service, for how many courses, and how many videos they accessed within a course (appreciation of the service). We analyzed the correlation between the use of the service and the performance of the students in terms of successful examination rate and average mark (effectiveness of the service). Finally, by using data mining techniques we profiled users according to their behavior while accessing the educational video service. We found out six different patterns that reflect different uses of the services matching different learning goals (flexibility of the service). The results of these analyses show the quality of the proposed blended learning model and the coherency of its implementation with respect to the design goals.

The diffusion of learning technologies has fostered the use of mobile and Web-based applications to assess the knowledge level of learners. In parallel, an increasing research interest has been devoted to studying new learning analytics tools able to summarize the content of large sets of learning documents. To bridge the gap between formative assessment tools and document summarization systems, this paper addresses the problem of recommending short summaries of large sets of learning documents based on the outcomes of multiple-choice tests. Specifically, it presents a new methodology for integrating formative assessment through mobile applications and summarization of learning documents in textual form. The content of the multiple-choice tests is exploited to drive the generation of document summaries tailored to specific topics. Furthermore, the outcomes of the tests are used to automatically recommend the generated summaries to learners based on their actual needs. As a case study, we performed an evaluation experience of students' progresses, which was conducted in the context of a university-level course. The achieved results show the applicability of the proposed methodology.

Our university has offered a massive educational video service since 2010, as part of a blended learning model that allows students to balance active participation in the classroom with remote access to video-recorded lectures. In these years, we have collected a huge amount of very detailed data about the students' access to the service. Together with additional information that characterize a university system (e.g. students' performance or course population), these data represent a precious ground set to assess the educational model. The paper describes an experimental set to profile the use of the educational video service, whose results will contribute to improve the model. Specifically the paper analyzes the students' service use relatively to different transversal course characteristics, such as level, main topic, population, success rate. As a result, it outlines the profile of the “ideal” courses for which students highly appreciate the service. This information will help educational designers to select the future courses to be included in the service, but it will also give directions on the sectors where improvements are necessary. Finally, the paper experimentally demonstrates a positive impact of the educational video service on students' performance, and specifically on the exam success rate.

Today's citizens and city administrations have an increasing interest in monitoring the air quality in urban areas. Studying the causes of air pollution entails analyzing the correlations between heterogeneous data, among which pollutant concentrations, traffic flow measurements, and meteorological data. To this end, innovative data analytics solutions able to acquire, integrate, and analyze very large amounts of data are needed. This paper presents a new data mining system, named GEneralized Correlation analyzer of pOllution data (GECKO), to discover interesting and multiple-level correlations among a large variety of open air pollution-related data. Specifically, correlations among pollutant levels and traffic and climate conditions are discovered and analyzed at different abstraction levels. The knowledge extraction process is driven by a taxonomy to generalize low-level measurement values as the corresponding categories. To ease the manual inspection of the result, the extracted correlations are classified into few classes based on the semantics of underlying data. The experiments, performed on real data acquired in a major Italian Smart City, demonstrate the effectiveness of the proposed analytics engine in discovering correlations among pollutant data that are potentially useful for supporting city administrators in decision-making.

E-learning systems commonly rely on advanced ICT technologies to enable users to access and browse electronic resources. Document summarization is an established text mining technique which focuses on extracting succinct summaries of potentially long textual documents. The application of summarization algorithms in the e-learning context is particularly appealing, because readers may want to pinpoint the key concepts by reading short summaries instead of the whole document content. This paper investigates the application of a state-of-the-art summarization algorithm to English-written academic teaching material. The summarizer produces an ordered sequence of key phrases extracted from learning material organized in different sections. The generated summaries are provided to students as additional material for study and revision. A crowd-sourcing experience of evaluation of the generated summaries was conducted by involving the students of a B.S. Course given by a technical university. The results show that the automatically generated summaries reflect, to a large extent, the student's expectations and therefore they can be useful for supporting individual and collective learning activities.

Physicians and health care organizations always collect large amounts of data during patient care. These large and high-dimensional datasets are usually characterized by an inherent sparseness. Hence, analyzing these datasets to figure out interesting and hidden knowledge is a challenging task. This article proposes a new data mining framework based on generalized association rules to discover multiple-level correlations among patient data. Specifically, correlations among prescribed examinations, drugs, and patient profiles are discovered and analyzed at different abstraction levels. The rule extraction process is driven by a taxonomy to generalize examinations and drugs into their corresponding categories. To ease the manual inspection of the result, a worthwhile subset of rules (i.e., nonredundant generalized rules) is considered. Furthermore, rules are classified according to the involved data features (medical treatments or patient profiles) and then explored in a top-down fashion: from the small subset of high-level rules, a drill-down is performed to target more specific rules. The experiments, performed on a real diabetic patient dataset, demonstrate the effectiveness of the proposed approach in discovering interesting rule groups at different abstraction levels.

Multidocument summarization addresses the selection of a compact subset of highly informative sentences, i.e., the summary, from a collection of textual documents. To perform sentence selection, two parallel strategies have been proposed: (a) apply general-purpose techniques relying on data mining or information retrieval techniques, and/or (b) perform advanced linguistic analysis relying on semantics-based models (e.g., ontologies) to capture the actual sentence meaning. Since there is an increasing need for processing documents written in different languages, the attention of the research community has recently focused on summarizers based on strategy (a). This article presents a novel multilingual summarizer, namely MWI-Sum (Multilingual Weighted Itemset-based Summarizer), that exploits an itemset-based model to summarize collections of documents ranging over the same topic. Unlike previous approaches, it extracts frequent weighted itemsets tailored to the analyzed collection and uses them to drive the sentence selection process. Weighted itemsets represent correlations among multiple highly relevant terms that are neglected by previous approaches. The proposed approach makes minimal use of language-dependent analyses. Thus, it is easily applicable to document collections written in different languages. Experiments performed on benchmark and real-life collections, English-written and not, demonstrate that the proposed approach performs better than state-of-the-art multilingual document summarizers.

In the era of smart cities huge data volumes are continuously generated and collected, thus prompting the need for efficient and distributed data mining approaches. Generalized itemset mining is an established data mining technique, which entails the discovery of multiple-level patterns hidden in the analyzed data by exploiting analyst-provided taxonomies. Among the generalized itemsets, the most peculiar high-level patterns are those with many contrasting correlations among items at different abstraction levels. They represent misleading situations that are worth analyzing separately by experts during manual inspection. This paper proposes a novel cloud-based service, named MGI-CLOUD, to efficiently mine misleading multiple-level patterns, i.e., the Misleading Generalized Itemsets, on a distributed computing environment. MGI-CLOUD consists of a set of distributed MapReduce jobs running in the cloud. As a case study, the system has been contextualized in a real-life scenario, i.e., the analysis of traffic law infractions committed in a smart city environment. The experiments, performed on real datasets, demonstrate the efficiency and effectiveness of MGI-CLOUD.

Frequent weighted itemsets represent correlations frequently holding in data in which items may weight differently. However, in some contexts, e.g., when the need is to minimize a certain cost function, discovering rare data correlations is more interesting than mining frequent ones. This paper tackles the issue of discovering rare and weighted itemsets, i.e., the infrequent weighted itemset (IWI) mining problem. Two novel quality measures are proposed to drive the IWI mining process. Furthermore, two algorithms that perform IWI and Minimal IWI mining efficiently, driven by the proposed measures, are presented. Experimental results show efficiency and effectiveness of the proposed approach.

Gene Expression Datasets (GEDs) usually consist of the expression values of thousands of genes within hundreds of samples. Frequent itemset and association rule mining algorithms have been applied to discover significant co-expressions among multiple genes from GEDs. To perform these data analyses, gene expression values are commonly discretized into a predefined number of bins. Such an expert-driven and not trivial preprocessing step could bias the quality of the mining result. This paper presents a novel approach to discovering gene correlations from GEDs which does not require data discretization. By representing per-sample gene expression values as item weights, frequent weighted itemsets can be extracted. The discovery of weighted itemsets instead of traditional (not weighted) ones prevents experts from discretizing GEDs before analyzing them and thus improves the effectiveness of the knowledge discovery process. Experiments performed on real GEDs demonstrate the effectiveness of the proposed approach.

The increasing availability of user-generated content coming from online communities allows the analysis of common user behaviors and trends in social network usage. This paper presents the TweM (Tweet Miner) framework that entails the discovery of h

Tag recommendation is focused on recommending useful tags to a user who is annotating a Web resource. A relevant research issue is the recommendation of additional tags to partially annotated resources, which may be based on either personalized or collective knowledge. However, since the annotation process is usually not driven by any controlled vocabulary, the collections of user-specific and collective annotations are often very sparse. Indeed, the discovery of the most significant associations among tags becomes a challenging task. This article presents a novel personalized tag recommendation system that discovers and exploits generalized association rules, that is, tag correlations holding at different abstraction levels, to identify additional pertinent tags to suggest. The use of generalized rules relevantly improves the effectiveness of traditional rule-based systems in coping with sparse tag collections, because: (i) correlations hidden at the level of individual tags may be anyhow figured out at higher abstraction levels and (ii) low-level tag associations discovered from collective data may be exploited to specialize high-level associations discovered in the user-specific context. The effectiveness of the proposed system has been validated against other personalized approaches on real-life and benchmark collections retrieved from the popular photo-sharing system Flickr.

A promising approach to Bayesian classification is based on exploiting frequent patterns, i.e., patterns that frequently occur in the training data set, to estimate the Bayesian probability. Pattern-based Bayesian classification focuses on building and evaluating reliable probability approximations by exploiting a subset of frequent patterns tailored to a given test case. This paper proposes a novel and effective approach to estimate the Bayesian probability. Differently from previous approaches, the Entropy-based Bayesian classifier, namely EnBay, focuses on selecting the minimal set of long and not overlapped patterns that best complies with a conditional-independence model, based on an entropy-based evaluator. Furthermore, the probability approximation is separately tailored to each class. An extensive experimental evaluation, performed on both real and synthetic data sets, shows that EnBay is significantly more accurate than most state-of-the-art classifiers, Bayesian and not.

Frequent itemset mining is a widely exploratory technique that focuses on discovering recurrent correlations among data. The steadfast evolution of markets and business environments prompts the need of data mining algorithms to discover significant correlation changes in order to reactively suit product and service provision to customer needs. Change mining, in the context of frequent itemsets, focuses on detecting and reporting significant changes in the set of mined itemsets from one time period to another. The discovery of frequent generalized itemsets, i.e., itemsets that 1) frequently occur in the source data, and 2) provide a high-level abstraction of the mined knowledge, issues new challenges in the analysis of itemsets that become rare, and thus are no longer extracted, from a certain point. This paper proposes a novel kind of dynamic pattern, namely the History GENeralized Pattern (HIGEN), that represents the evolution of an itemset in consecutive time periods, by reporting the information about its frequent generalizations characterized by minimal redundancy (i.e., minimum level of abstraction) in case it becomes infrequent in a certain time period. To address HIGEN mining, it proposes HIGEN MINER, an algorithm that focuses on avoiding itemset mining followed by postprocessing by exploiting a support-driven itemset generalization approach. To focus the attention on the minimally redundant frequent generalizations and thus reduce the amount of the generated patterns, the discovery of a smart subset of HIGENs, namely the NONREDUNDANT HIGENs, is addressed as well. Experiments performed on both real and synthetic datasets show the efficiency and the effectiveness of the proposed approach as well as its usefulness in a real application context.

This paper presents a novel semi-automatic approach to construct conceptual ontologies over structured data by exploiting both the schema and content of the input dataset. It effectively combines two well-founded database and data mining techniques, i.e., functional dependency discovery and association rule mining, to support domain experts in the construction of meaningful ontologies, tailored to the analyzed data, by using Description Logic (DL). To this aim, functional dependencies are first discovered to highlight valuable conceptual relationships among attributes of the data schema (i.e., among concepts). The set of discovered correlations effectively support analysts in the assertion of the Tbox ontological statements (i.e., the statements involving shared data conceptualizations and their relationships). Then, the analyst-validated dependencies are exploited to drive the association rule mining process. Association rules represent relevant and hidden correlations among data content and they are used to provide valuable knowledge at the instance level. The pushing of functional dependency constraints into the rule mining process allows analysts to look into and exploit only the most significant data item recurrences in the assertion of the Abox ontological statements (i.e., the statements involving concept instances and their relationships).