Mastroianni, Michele

In academic institutions, there is frequently the need to provide new services, in a cloud model, to be used either in teaching or research activities. One of the main decisions to be addressed is related to the cloud model to adopt (private, public or hybrid), and what the mixing of functionalities to use for the hybrid one. In this paper two different methodologies (Cost/features and Semantic-based) are been experimented in order to identify the best suited cloud model to adopt for a specific problem. The long-term perspective is to build a methodology to serve as a tool to be used as decision support for the ICT manager in order to help him in this decision. The comparison between the two different methodologies show the strengths and weaknesses of both approaches.

Local pollution is a problem that affects urban areas and has effects on the quality of life and on health conditions. In order to not develop strict measures and to better manage territories, the national authorities have applied a vast range of predictive models. Actually, the application of machine learning has been studied in the last decades in various cases with various declination to simplify this problem. In this paper, we apply a regression-based analysis technique to a dataset containing official historical local pollution and weather data to look for criteria that allow forecasting critical conditions. The methods was applied to the case study of Napoli, Italy, where the local environmental protection agency manages a set of fixed monitoring stations where both chemical and meteorological data are recorded. The joining of the two raw dataset was overcome by the use of a maximum inclusion strategy as performing the joining action with ”outer” mode. Among the four different r egression models applied, namely the Linear Regression Model calculated with Ordinary Least Square (LN-OLS), the Ridge regression Model (Ridge), the Lasso Model (Lasso) and Supervised Nearest Neighbors Regression (KNN), the Ridge regression model was found to better perform with an R2 (Coefficient of Determination) value equal to 0.77 and low value for both MAE (Mean Absolute Error) and MSE (Mean Squared Error), equal to 0.12 and 0.04 respectively.

Blockchains provide a valid and profitable support for the implementation of trustable and secure distributed ledgers, in support to groups of subjects that are potentially competitors in conflict of interest but need to share progressive information recording processes. Blockchains prevent data stored in blocks from being altered or deleted, but there are situations in which stored information must be deleted or made inaccessible on request or periodically, such as the ones in which GDPR is applicable. In this paper we present literature solutions and design an implementation in the context of a traffic management system for the Internet of Vehicles based on the Pseudonymization/Cryptography solution, evaluating its viability, its GDPR compliance and its level of risk.

In this work, we envise an effective case study concerning a data and a model poisoning attack, consisting in evaluating how much a poisoned word embeddings model could affect the reliability of a deep neural network-based Fake News Checker; furthermore, we plan to train three different word embeddings models among the most performing in the Natural Language Processing field, in order to investigate which of these models can be considered more resilient and robust when such kind of attacks are applied.

The widespread diffusion of Cloud paradigm and its approach based on delegation of resources to service providers, improved greatly the need of protecting personal data. Accordingly, in recent years, governments are going to define and apply new rules, that aims at protecting the personal space of each individual. From 2018, General Data protection Regulation (GDPR) applies in Europe, giving specific rights to each individual and imposing procedures to protect personal data. GDPR addresses a clear need of our social network-based society, but has the side effect of outlining the incapability of many actual enterprises, especially small and medium ones, to address such new requirements. In this paperthe new Regulation is described with a conceptual map approach.