Luís Nunes

The present work reports the impacts on urban mobility and air quality in Lisbon, Portugal, of the imposed restrictions to curb the transmission of SARS-CoV-2 virus, which causes COVID-19 disease. We performed a data-driven approach over Lisbon Smart cities data, collected from several sources, such as traffic and pollution. During the first Portuguese emergency period (18-03-2020 to 03-05-2020) the sharp reductions in anthropogenic activities, most importantly road traffic, resulted in generally reduced criteria air pollutant concentration compared to an homologous baseline from 2013–2019 measured in the six air quality monitoring stations throughout the city. The most negatively impacted air pollutants were NO2, with a reduction of 54.35% in traffic stations and 28.62% in background stations. Google mobility indicator for local commerce was found to be the main anthropogenic activity indicator for Lisbon, with a moderate and positive correlation with NO ₂ concentration ( r =+0.54). A regressor ML pipeline was trained to predict NO ₂ concentration with the available anthropogenic activity, weather, and air pollutant inputs from March/2020 to March/2021, achieving R ² = 0.925 on the test set.

Explainable Artificial Intelligence (XAI) is a growing area of research that aims to improve the interpretability of the not-so-informative black-box models. However, it is currently difficult to categorize an existing method in terms of its intrinsic characteristics and explainability. We provide a new unified yet simple taxonomy for the categorization of XAI methods and present the explainability methods currently being applied in finance applications. For both purposes, we present two separate systematic literature reviews: an anthological search for surveys on XAI methods in order to present a unified taxonomy, followed by an exposition of the XAI methods currently in use that have been found. We also concisely define the existing explainability methods using the proposed categories based on the ones most commonly addressed in the reviewed literature and pinpoint specific XAI methods being used in practical applications in Finance.

This study presents a trajectory-based person re-identification algorithm, embedded in a tool to detect and track customers present in a large retail store, in a multi-camera environment. The customer trajectory data are obtained from video surveillance images captured by multiple cameras, and customers are detected and tracked along the frames that compose the videos. Due to the characteristics of a multi-camera environment or the occurrence of occlusions, caused by objects such as shelves or counters, different identifiers are assigned to each person when, in fact, they should be identified with a unique identifier. Thus, the proposed tool tries to solve this problem in a scenario where there are constraints in using images of people due to data privacy concerns. The results show that our method was able to correctly re-identify the customers present in the store with a re-identification rate of 82%.

The widespread usage of mobile devices and their seamless adaptation to each user’s needs through useful applications (apps) makes them a prime target for malware developers. Malware is software built to harm the user, e.g., to access sensitive user data, such as banking details, or to hold data hostage and block user access. These apps are distributed in marketplaces that host millions and therefore have their forms of automated malware detection in place to deter malware developers and keep their app store (and reputation) trustworthy. Nevertheless, a non-negligible number of apps can bypass these detectors and remain available in the marketplace for any user to download and install on their device. Current malware detection strategies rely on using static or dynamic app extracted features (or a combination of both) to scale the detection and cover the growing number of apps submitted to the marketplace. In this paper, the main focus is on the apps that bypass the malware detectors and stay in the marketplace long enough to receive user feedback. This paper uses real-world data provided by an app store. The quantitative ratings and potential alert flags assigned to the apps by the users were used as features to train machine learning classifiers that successfully classify malware that evaded previous detection attempts. These results present reasonable accuracy and thus work to help to maintain a user-safe environment.

Traffic accidents in urban areas lead to reduced quality of life and added pressure in the cities’ infra-structures. In the context of smart city data is becoming available that allows a deeper analysis of the phenomenon. We propose a data fusion process from different information sources like road accidents, weather conditions, local authority reports tools, traffic, fire brigade. These big data analytics allow the creation of knowledge for local municipalities using local data. Data visualizations allow big picture overview. This paper presents an approach to the geo-referenced accident-hotspots identification. Using ArcGIS Pro, we apply Kernel Density and Hot Spot Analysis (Getis-Ord Gi*) tools, identifying the existence of black spots in terms of location and context conditions, and evaluate the possible human, environmental and circumstantial factors that may influence the severity of accidents. The results were validated by an expert committee. This approach can be applied to other cites wherever this data is available.

Mobile computing is on the rise. More and more users rely on mobile applications and mobile devices to execute the most basic tasks of their lives while depositing their most private and critical data on them. Mobile application stores act as the trust anchors that sit between applications developed by third parties and the user’s mobile devices. Therefore, app stores should provide the means to ensure that the apps installed by the users follow high security and quality standards, minimising the user’s data exposure risks. A critical path towards that security and quality standards is to early test and detect mobile application vulnerabilities resulting from incorrect development practices and to provide developers feedback about the problems found and some additional information on how to correct them. This paper presents a system, developed to help Android app stores (in this case, on the Aptoide app store) timely detect vulnerabilities on submitted apps and provide appropriate feedback to developers. The provide feedback makes developers aware of the secure development processes while improving the quality and security of their apps before they are made available to end-users and installed on their devices.

This study examines the relationship between distance measures and a Portuguese data set consisting of 34,622 online hotel reviews extracted from Booking.com and TripAdvisor written in Portuguese, Spanish, and English. Based on the country of origin of each review author, a geographic and a psychic distance measure is calculated for Portugal. Data and text mining analysis provides additional insights into online hotel ratings. The authors confirm that online travelers’ evaluations are multifaceted constructs displaying varying patterns of rating behavior among the traveler base. By investigating the contemporary relevance of geographic and psychic distance, a key finding of this study is that travelers with less distance both in terms of psychic and geographic distance give a lower rating score than travelers with greater distance. The inclusion of psychic and geographic distance is advocated as a salient aspect for future researchers and for those practitioners who wish to enhance hotel product and service features.

Booking cancellations negatively contribute to the production of accurate forecasts, which comprise a critical tool in the hospitality industry. Research has shown that with today's computational power and advanced machine learning algorithms it is possible to build models to predict bookings cancellation likelihood. However, the effectiveness of these models has never been evaluated in a real environment. To fill this gap and investigate how these models can be implemented in a decision support system and its impact on demand-management decisions, a prototype was built and deployed in two hotels. The prototype, based on an automated machine learning system designed to learn continuously, lead to two important research contributions. First, the development of a training method and weighting mechanism designed to capture changes in cancellations patterns over time and learn from previous days' predictions hits and errors. Second, the creation of a new measure-Minimum Frequency-to measure the precision of predictions over time. From a business standpoint, the prototype demonstrated its effectiveness, with results exceeding 84% in accuracy, 82% in precision, and 88% in Area Under the Curve (AUC). The system allowed hotels to predict their net demand and thus making better decisions about which bookings to accept and reject, what prices to make, and how many rooms to oversell. The systematic prediction of bookings with high probability of being canceled allowed hotels to reduce cancellations by 37 percentage points by acting to avoid their cancellation.