1,720,983 research outputs found
Analytic Processing in Data Lakes: A Semantic Query-Driven Discovery Approach
Full text linkData integration and discovery are open issues in Data Lakes potentially storing hundreds of data sources. The present paper addresses these issues targeting multidimensional data sources, that is sources containing atomic or derived measures aggregated along a number of dimensions, typically derived from raw data for analytical and reporting purposes. Combining semantic models of metadata with existing data-driven techniques, the paper proposes an approach for the discovery of mappings between source metadata and concepts in a reference knowledge graph, enabling the definition of reasoning-based techniques to discover, integrate, and rank data sources relevant to a given analytical query. The efficiency and effectiveness of the approach is discussed by means of experiments on real-world scenarios
An Experimental Comparison of Large Language Models for Emotion Recognition in Italian Tweets
No full textIn recent years, the advent of Large Language Models (LLMs), which are task-agnostic models trained on huge amounts of textual data, has given momentum to a wide variety of NLP applications, ranging from chatbots to sentiment classifiers. Currently, many LLMs are publicly available, each with different features and performance, and the selection of the best LLM for a specific task may be challenging. In this work, we focus on the task of emotion recognition in Italian social media content and we present an experimental comparison among three of the most popular LLMs: Google Bidirectional Encoder Representations from Transformers (BERT), OpenAI Generative Pre-trained Transformer 3 (GPT-3) and GPT-3.5. Model specialization in emotion recognition has been achieved by using two different approaches, namely fine-tuning and prompt engineering with few-shot task transfer. The experimentation has been performed on TwIT, a corpus of about 3100 Italian tweets annotated with respect to six emotions. The results show that fine-tuning GPT-3 leads to the best performance on the considered dataset, achieving a remarkable F1=0.90
Find the Right Peers: Building and Querying Multi-IoT Networks Based on Contexts
No full textWith the evolution of the features smart devices are equipped with, the IoT realm is becoming more and more intertwined with people daily-life activities. This has, of course, impacts in the way objects are used, causing a strong increase in both the dynamism of their contexts and the diversification of their objectives. This results in an evolution of the IoT towards a more complex environment composed of multiple overlapping networks, called Multi-IoTs (MIoT). The low applicability of classical cooperation mechanisms among objects leads to the necessity of developing more complex and refined strategies that take the peculiarity of such a new environment into consideration. In this paper, we address this problem by proposing a new model for devices and their contexts following a knowledge representation approach. It borrows ideas from OLAP systems and leverages a multidimensional perspective by defining dimension hierarchies. In this way, it enables roll-up and drill-down operations on the values of the considered dimensions. This allows for the design of more compact object networks and the definition of new strategies for the retrieval of relevant devices
How to Cope with Personnel Unavailability? Process Mining May Help!
No full textReplacement planning is critical to guarantee continuity of operations in business processes in case of personnel unavailability. In this work, we propose a data-driven approach for supporting resource replacement that makes use of logs of past process executions to model a social network of resources. On this top, a similarity measure among resources is exploited to assign tasks of unavailable resource to the available ones through an Integer Linear Model
Multi-dimensional contexts for querying IoT networks
No full textThe pervasiveness of smart objects in people daily life is increasing, as the capabilities of objects are becoming more and more sophisticated. Objects participate to the Internet of Thing (IoT) with changing contexts and scopes, thus resulting in the rise of multiple networks linked to each other to form a new paradigm, called Multi-IoTs (MIoT). Of course, cooperation strategies among objects must follow this innovative trend as classical strategies based on the concept of coexistence appear no more adequate. In this scenario, this paper proposes a contribution by introducing a complex model for devices and contexts that follows a knowledge representation approach. It adopts dimension hierarchies in the multidimensional perspective typical of OLAP systems to represent roll-up relationships between admissible members of the considered dimensions, enabling the retrieval of relevant objects through a supervised algorithm
Evidence-driven appraisal of students’ careers using process mining: a case study
Full text linkToday’s universities are more and more focused on improving their educational programs and supporting their students throughout their academic journey. A key aspect of such an effort is understanding which factors contribute to poor students’ performance. This research illustrates how educational process mining techniques can be used to effectively uncover success and failure factors in students’ academic journeys through a case study at an Italian university. The research reveals patterns related to adherence to curriculum requirements, strategies for taking exams, and the influence of various factors, such as the number of exams passed in the first year on graduation timelines. These findings offer valuable insights for educational institutions that might be used to, e.g., implement support mechanisms to enhance students’ overall success rates
Towards next-location prediction for process executions
No full textPredictive monitoring of business processes aims at predicting the future of an ongoing process execution. In this work, we focus on the prediction of the next activities to be executed in a running case. However, in contrast with most state-of-The-Art approaches, focused on predicting exactly the next activity that will be executed from the current state of the process, we propose an approach aimed at predicting the portion of the process (or 'location') that is likely to be executed next. The notion of location allows us to detect activities belonging to the same portion of a control-flow construct (e.g., at the beginning of a parallelism, or at the end of a loop). It provides an abstraction mechanism from the level of the single activity, which can be used to provide the process analyst with an higher-level overview of what can be expected next in the process execution. We validated the approach over a set of real-world datasets comparing and discussing different strategies for training a classifier in returning a location in place of an activity label
A Knowledge Graph Framework for Impact Calculation in Life-Cycle Assessment
Full text linkSustainability assessments are increasingly critical for evaluating the environmental impacts of business activities. Life-cycle assessment (LCA) is a key methodology in measuring these impacts, but integrating and analyzing data from diverse enterprise data sources to compute LCA indicators remains a challenging task. In this paper, we propose an approach that leverages Knowledge Graphs to formally model LCA indicators and their associated mathematical calculation formulas. The graph serves as a flexible schema supporting enterprises in documentation and mathematical interpretation of indicators for LCA, by linking their internal data sources to the Graph. On top of it, a suite of reasoning-based services is presented to automate the calculation of indicators from available data sources through algebraic manipulation, and to facilitate collaboration among a network of enterprises by enabling consistent comparison of sustainability assessment
A Knowledge Graph Framework for Impact Calculation in Life-Cycle Assessment
Full text linkSustainability assessments are increasingly critical for evaluating the environmental impacts of business activities. Life-cycle assessment (LCA) is a key methodology in measuring these impacts, but integrating and analyzing data from diverse enterprise data sources to compute LCA indicators remains a challenging task. In this paper, we propose an approach that leverages Knowledge Graphs to formally model LCA indicators and their associated mathematical calculation formulas. The graph serves as a flexible schema supporting enterprises in documentation and mathematical interpretation of indicators for LCA, by linking their internal data sources to the Graph. On top of it, a suite of reasoning-based services is presented to automate the calculation of indicators from available data sources through algebraic manipulation, and to facilitate collaboration among a network of enterprises by enabling consistent comparison of sustainability assessments
A Preliminary Study on the Application of Reinforcement Learning for Predictive Process Monitoring
No full textThe present paper explores the opportunity of applying reinforcement learning to various typical tasks in the field of predictive process monitoring. The tasks considered are the prediction of both nextevent activity and time completion as well as the prediction of the whole progression of running cases. Experiments have been conducted on the popular benchmark dataset, BPI’ 2012, on which we compare the pro-posed learning system with state of the art methods adopting LSTM networks trained through supervised learning. Results enlighten promising features of the approach and interesting research issues and challenges, as well as proving the applicability of reinforcement learning to predictive process monitoring
- …
