Abstract
CloudAnchor is a multi-agent brokerage platform for the negotiation of Infrastructure as a Service cloud resources between Small and Medium Sized Enterprises, acting either as providers or consumers. This project entails the research, design, and implementation of a smart contract solution to permanently record and manage contractual and behavioural stakeholder data on a blockchain network. Smart contracts enable safe contract code execution, increasing trust between parties and ensuring the integrity and traceability of the chained contents. The defined smart contracts represent the inter-business trustworthiness and Service Level Agreements established within the platform. CloudAnchor interacts with the blockchain network through a dedicated Application Programming Interface, which coordinates and optimises the submission of transactions. The performed tests indicate the success of this integration: (i) the number and value of negotiated resources remain identical; and (ii) the run-time increases due to the inherent latency of the blockchain operation. Nonetheless, the introduced latency does not affect the brokerage performance, proving to be an appropriate solution for reliable partner selection and contractual enforcement between untrusted parties. This novel approach stores all brokerage strategic knowledge in a distributed, decentralised, and immutable database.

Abstract
This study presents the dependence of strain sensitivity on cavity length in conventional Fabry-Perot (F-P) sensors. A high number of F-P sensors were required and to ensure their reproducibility, a manufacturing process was developed to obtain similar sensors but with different types of lengths. A hollow-core silica tube was used to fabricate several F-P cavities by fusion splicing it between two sections of SMF28 fiber. The fabricated F-P has a varying length ranging from 15 to 2500 mu m. The cavities were measured under a microscope and the reflected spectrum was acquired for each one. Strain measurements were performed for a maximum strain of 1000 mu epsilon. The strain sensitivity showed a highly linear correlation with increment lambda(FSR). Small length variations for short cavities heavily affect the FSR value. The smallest and longest cavities present sensitivities of 8.71 and 2.68 pm/mu epsilon, respectively. Thermal characterization for low- and high-temperature regimes was also performed and is constant for tested sensors.

Abstract
This research aims to extract knowledge of sensors behavior resorting to Bayesian networks (BNs) and dynamic Bayesian networks (DBNs), a time-based BN version. These two types of models belong to the group of probabilistic graphical models (PGMs). These graphical models can be very useful to get insights from data in order to improve sensor capabilities in the industry of fire detection systems, since it can provide the conditional dependence structure among various sensor variables. Relevant sensors with fire alerts were selected and studied at device level. We conduct a data fusion analysis since we deal with heterogeneous data sources, Remote Alert (RA) with sensor states and Condition Monitoring (CM) with numerical data. To achieve an accurate fusion of the data, a pipeline was designed to align both sources of data in a regular time interval. Furthermore, a change point detection (CPD) method was used to discretize the numerical variables. In addition, one-hot encoding was used to create binarized datasets and combine all data (RA+CM). Our modeling helps understanding the dependencies among the sensor variables, highlighting that individual devices of the same type can have a very different probabilistic behavior along the time, probably due to be installed in distinct regions. Moreover, the models helped capturing strange probabilistic sensor behavior such as a low probability of a NORMAL state happening given that states FIRE, WARNING and TROUBLE did not happen. © 2022 IEEE.

Abstract
Ferramentas de e-learning no contexto ensino-aprendizagem

Abstract
Introduction: Breast volume estimation is considered crucial for breast cancer surgery planning. A single, easy, and reproducible method to estimate breast volume is not available. This study aims to evaluate, in patients proposed for mastectomy, the accuracy of the calculation of breast volume from a low-cost 3D surface scan (Microsoft Kinect) compared to the breast MRI and water displacement technique. Material and Methods: Patients with a Tis/T1-T3 breast cancer proposed for mastectomy between July 2015 and March 2017 were assessed for inclusion in the study. Breast volume calculations were performed using a 3D surface scan and the breast MRI and water displacement technique. Agreement between volumes obtained with both methods was assessed with the Spearman and Pearson correlation coefficients. Results: Eighteen patients with invasive breast cancer were included in the study and submitted to mastectomy. The level of agreement of the 3D breast volume compared to surgical specimens and breast MRI volumes was evaluated. For mastectomy specimen volume, an average (standard deviation) of 0.823 (0.027) and 0.875 (0.026) was obtained for the Pearson and Spearman correlations, respectively. With respect to MRI annotation, we obtained 0.828 (0.038) and 0.715 (0.018). Discussion: Although values obtained by both methodologies still differ, the strong linear correlation coefficient suggests that 3D breast volume measurement using a low-cost surface scan device is feasible and can approximate both the MRI breast volume and mastectomy specimen with sufficient accuracy. Conclusion: 3D breast volume measurement using a depth-sensor low-cost surface scan device is feasible and can parallel MRI breast and mastectomy specimen volumes with enough accuracy. Differences between methods need further development to reach clinical applicability. A possible approach could be the fusion of breast MRI and the 3D surface scan to harmonize anatomic limits and improve volume delimitation.

Abstract
In Portugal, the dropout rate of university courses is around 29%. Understanding the reasons behind such a high desertion rate can drastically improve the success of students and universities. This work applies existing data mining techniques to predict the academic dropout mainly using the academic grades. Four different machine learning techniques are presented and analyzed. The dataset consists of 331 students who were previously enrolled in the Computer Engineering degree at the Universidade de Tras-os-Montes e Alto Douro (UTAD). The study aims to detect students who may prematurely drop out using existing methods. The most relevant data features were identified using the Permutation Feature Importance technique. In the second phase, several methods to predict the dropouts were applied. Then, each machine learning technique's results were displayed and compared to select the best approach to predict academic dropout. The methods used achieved good results, reaching an Fl-Score of 81% in the final test set, concluding that students' marks somehow incorporate their living conditions.