Supervised and weakly supervised counting-by-segmentation: the fluorescent microscopy use case

This thesis focuses on automating the time-consuming task of manually counting activated neurons in fluorescent microscopy images, which is used to study the mechanisms underlying torpor. The traditional method of manual annotation can introduce bias and delay the outcome of experiments, so the author investigates a deep-learning-based procedure to automatize this task. The author explores two of the main convolutional-neural-network (CNNs) state-of-the-art architectures: UNet and ResUnet family model, and uses a counting-by-segmentation strategy to provide a justification of the objects considered during the counting process. The author also explores a weakly-supervised learning strategy that exploits only dot annotations. The author quantifies the advantages in terms of data reduction and counting performance boost obtainable with a transfer-learning approach and, specifically, a fine-tuning procedure. The author released the dataset used for the supervised use case and all the pre-training models, and designed a web application to share both the counting process pipeline developed in this work and the models pre-trained on the dataset analyzed in this work.

L'inclusione lavorativa tra accesso e partecipazione

La Convenzione delle Nazioni Unite sui Diritti delle Persone con Disabilità (UNCRPD) riconosce il diritto di tutte le persone al lavoro “gli Stati Parti adottano misure adeguate a garantire alle persone con disabilità, su base di uguaglianza con gli altri, l’accesso all’ambiente fisico, ai trasporti, all’informazione e alla comunicazione, compresi i sistemi e le tecnologie di informazione e comunicazione e ad altre attrezzature e servizi aperti o forniti al pubblico”(United Nation 2016 p.14). Nonostante i progressi (in ambito politico culturale) che si stanno compiendo in ambito internazionale in termini di pari opportunità e di inclusione, le persone in situazione di disabilità continuano a incontrare barriere che limitano la loro partecipazione attiva al mondo del lavoro. A partire da questo scenario, la ricerca si propone di indagare i bisogni (es. di accoglienza, di accesso al contesto fisico e digitale, di partecipazione nella vita dell’azienda ecc.) delle persone con disabilità e di sviluppare una applicazione digitale (web app), rivolta alle imprese, finalizzata a monitorare e a promuovere l'inclusione lavorativa. Ripercorrendo il modello di progettazione del design thinking e valorizzando un processo di ricerca basato su metodi misti (qualitativi e qualitativi) è stato ideato Job inclusion for all; un ambiente digitale fondato sull’adattamento di due strumenti di “metariflessione”: l’Index for inclusion job version e l’employment role mapping. Lo strumento digitale prototipato è stato testato e validato, durante l’ultimo anno di ricerca, da parte di una equipe multidisciplinare internazionale; tale processo ha consentito di raccogliere feedback (rispetto alla rilevanza e alla chiarezza degli item, rispetto ai punti di forza e di debolezza) che hanno consentito di migliorare e implementare la versione finale del prototipo di web app.

Supporting requirement elicitation and ontology testing in knowledge graph engineering

Knowledge graphs and ontologies are closely related concepts in the field of knowledge representation. In recent years, knowledge graphs have gained increasing popularity and are serving as essential components in many knowledge engineering projects that view them as crucial to their success. The conceptual foundation of the knowledge graph is provided by ontologies. Ontology modeling is an iterative engineering process that consists of steps such as the elicitation and formalization of requirements, the development, testing, refactoring, and release of the ontology. The testing of the ontology is a crucial and occasionally overlooked step of the process due to the lack of integrated tools to support it. As a result of this gap in the state-of-the-art, the testing of the ontology is completed manually, which requires a considerable amount of time and effort from the ontology engineers. The lack of tool support is noticed in the requirement elicitation process as well. In this aspect, the rise in the adoption and accessibility of knowledge graphs allows for the development and use of automated tools to assist with the elicitation of requirements from such a complementary source of data. Therefore, this doctoral research is focused on developing methods and tools that support the requirement elicitation and testing steps of an ontology engineering process. To support the testing of the ontology, we have developed XDTesting, a web application that is integrated with the GitHub platform that serves as an ontology testing manager. Concurrently, to support the elicitation and documentation of competency questions, we have defined and implemented RevOnt, a method to extract competency questions from knowledge graphs. Both methods are evaluated through their implementation and the results are promising.

Network analysis and machine learning assist drug repurposing and safety assessment in neurological diseases

In recent decades, two prominent trends have influenced the data modeling field, namely network analysis and machine learning. This thesis explores the practical applications of these techniques within the domain of drug research, unveiling their multifaceted potential for advancing our comprehension of complex biological systems. The research undertaken during this PhD program is situated at the intersection of network theory, computational methods, and drug research. Across six projects presented herein, there is a gradual increase in model complexity. These projects traverse a diverse range of topics, with a specific emphasis on drug repurposing and safety in the context of neurological diseases. The aim of these projects is to leverage existing biomedical knowledge to develop innovative approaches that bolster drug research. The investigations have produced practical solutions, not only providing insights into the intricacies of biological systems, but also allowing the creation of valuable tools for their analysis. In short, the achievements are: • A novel computational algorithm to identify adverse events specific to fixed-dose drug combinations. • A web application that tracks the clinical drug research response to SARS-CoV-2. • A Python package for differential gene expression analysis and the identification of key regulatory "switch genes". • The identification of pivotal events causing drug-induced impulse control disorders linked to specific medications. • An automated pipeline for discovering potential drug repurposing opportunities. • The creation of a comprehensive knowledge graph and development of a graph machine learning model for predictions. Collectively, these projects illustrate diverse applications of data science and network-based methodologies, highlighting the profound impact they can have in supporting drug research activities.

Modelling the infrastructure of the “Śivadharma Database”

A Digital Scholarly Edition is a conceptually and structurally sophisticated entity. Throughout the centuries, diverse methodologies have been employed to reconstruct a text transmitted through one or multiple sources, resulting in various edition types. With the advent of digital technology in philology, these practices have undergone a significant transformation, compelling scholars to reconsider their approach in light of the web. In the digital age, philologists are expected to possess (too) advanced technical skills to prepare interactive and enriched editions, even though, in most cases, only mechanical or documentary editions are published online. The Śivadharma Database is a web Content Management System (CMS) designed to facilitate the preparation, publication, and updating of Digital Scholarly Editions. By providing scholars with a user-friendly CRUD web application to reconstruct and annotate a text, they can prepare their textus with additional components such as apparatus, notes, translations, citations, and parallels. It is possible by leveraging an annotation system based on HTML and graph data structure. This choice is made because the text entity is multidimensional and multifaceted, even if its sequential presentation constrains it. In particular, editions of South Asian texts of the Śivadharma corpus, the case study of this research, contain a series of phenomena that are difficult to manage formally, such as overlapping hierarchies. Hence, it becomes necessary to establish the data structure best suited to represent this complexity. In Śivadharma Database, the textus is an HTML file readily displayable. Textual fragments, annotated via an interface without requiring philologists to write code and saved in the backend, form the atomic unit of multiple relationships organised in a graph database. This approach enables the formal representation of complex and overlapping textual phenomena, allowing for good annotation expressiveness with minimal effort to learn the relevant technologies during the editing workflow.

Self-supervised Information Retrieval basato su deep metric learning

La tesi ha lo scopo di ricercare, esaminare ed implementare un sistema di Machine Learning, un Recommendation Systems per precisione, che permetta la racommandazione di documenti di natura giuridica, i quali sono già stati analizzati e categorizzati appropriatamente, in maniera ottimale, il cui scopo sarebbe quello di accompagnare un sistema già implementato di Information Retrieval, istanziato sopra una web application, che permette di ricercare i documenti giuridici appena menzionati.

Un ponte tra Crowdsourcing e Crowdsensing: L'implementazione di SmartCrowd

L’elaborato presenta lo sviluppo di un’applicazione web che unisce il con- cetto di crowdsourcing, ovvero un paradigma in cui una colletività esegue una mansione per il raggiungimento di un obiettivo; e il concetto di crowd- sensing, paradigma nel quale un gruppo di persone, utilizzando il proprio dispositivo mobile, condivide dati, che possono essere successivamente stu- diati. SmartCrowd si pone come intermediario tra questi due modelli, e la sua implementazione permette di raccogliere dati da una “folla” che faccia uso dei dispositivi mobili, usufruendo di una piattaforma di crowdsourcing. Con SmartCrowd vengono create le campagne, ovvero insieme di attività svol- te dagli utenti finali; esiste un sistema di interazione con la piattaforma di crowdsourcing di Microworkers, volta al reclutamento delle persone; la con- divisione dei dati tramite smartphone, usando il sensore GPS, eseguita dagli utenti ed infine, tramite SmartCrowd è possibile analizzare i dati ricevuti e fare una valutazione positiva o negativa.

Smart Mirror: remote verification of an identity document holder

Nowadays, some activities, such as subscribing an insurance policy or opening a bank account, are possible by navigating through a web page or a downloadable application. Since the user is often “hidden” behind a monitor or a smartphone, it is necessary a solution able to guarantee about their identity. Companies are often requiring the submission of a “proof-of-identity”, which usually consists in a picture of an identity document of the user, together with a picture or a brief video of themselves. This work describes a system whose purpose is the automation of these kinds of verifications.

Towards a Smart Campus Digital Twin: Promoting Awareness and Sustainability Through Wayfinding and Real-Time Environmental Data

The newly inaugurated Navile District of the University of Bologna is a complex created along the Navile canal, that now houses various teaching and research activities for the disciplines of Chemistry, Industrial Chemistry, Pharmacy, Biotechnology and Astronomy. A Building Information Modeling system (BIM) gives staff of the Navile campus several ways to monitor buildings in the complex throughout their life cycle, one of which is the ability to access real-time environmental data such as room temperature, humidity, air composition, and more, thereby simplifying operations like finding faults and optimizing environmental resource usage. But smart features at Navile are not only available to the staff: AlmaMap Navile is a web application, whose development is documented in this thesis, that powers the public touch kiosks available throughout the campus, offering maps of the district and indications on how to reach buildings and spaces. Even if these two systems, BIM and AlmaMap, don't seem to have many similarities, they share the common intent of promoting awareness for informed decision making in the campus, and they do it while relying on web standards for communication. This opens up interesting possibilities, and is the idea behind AlmaMap Navile 2.0, an app that interfaces with the BIM system and combines real-time sensor data with a comfort calculation algorithm, giving users the ability not just to ask for directions to a space, but also to see its comfort level in advance and, should they want to, check environmental measurements coming from each sensor in a granular manner. The end result is a first step towards building a smart campus Digital Twin, that can support all the people who are part of the campus life in their daily activities, improving their efficiency and satisfaction, giving them the ability to make informed decisions, and promoting awareness and sustainability.

Web-based application programming interface to solve nonlinear optimization problems

Nonlinear Optimization Problems are usual in many engineering fields. Due to its characteristics the objective function of some problems might not be differentiable or its derivatives have complex expressions. There are even cases where an analytical expression of the objective function might not be possible to determine either due to its complexity or its cost (monetary, computational, time, ...). In these cases Nonlinear Optimization methods must be used. An API, including several methods and algorithms to solve constrained and unconstrained optimization problems was implemented. This API can be accessed not only as traditionally, by installing it on the developer and/or user computer, but it can also be accessed remotely using Web Services. As long as there is a network connection to the server where the API is installed, applications always access to the latest API version. Also an Web-based application, using the proposed API, was developed. This application is to be used by users that do not want to integrate methods in applications, and simply want to have a tool to solve Nonlinear Optimization Problems.

Sigla WEBGIS portal: an open source Web-based GIS application for the portuguese bluetongue Entomologiacal surveillance program.

Dissertação apresentada como requisito parcial para obtenção do grau de Mestre em Ciência e Sistemas de Informação Geográfica

Intégration des connaissances ontologiques dans la fouille de motifs séquentiels avec application à la personnalisation Web

Thèse numérisée par la Division de la gestion de documents et des archives de l'Université de Montréal

Ka-Map: an interactive web-mapping application

ka-Map ("ka" as in ka-boom!) is an open source project that is aimed at providing a javascript API for developing highly interactive web-mapping interfaces using features available in modern web browsers. ka-Map currently has a number of interesting features. It sports the usual array of user interface elements such as: interactive, continuous panning without reloading the page; keyboard navigation options (zooming, panning); zooming to pre-set scales; interactive scalebar, legend and keymap support; optional layer control on client side; server side tile caching

Using the web graph to influence application behaviour

The Web's link structure (termed the Web Graph) is a richly connected set of Web pages. Current applications use this graph for indexing and information retrieval purposes. In contrast the relationship between Web Graph and application is reversed by letting the structure of the Web Graph influence the behaviour of an application. Presents a novel Web crawling agent, AlienBot, the output of which is orthogonally coupled to the enemy generation strategy of a computer game. The Web Graph guides AlienBot, causing it to generate a stochastic process. Shows the effectiveness of such unorthodox coupling to both the playability of the game and the heuristics of the Web crawler. In addition, presents the results of the sample of Web pages collected by the crawling process. In particular, shows: how AlienBot was able to identify the power law inherent in the link structure of the Web; that 61.74 per cent of Web pages use some form of scripting technology; that the size of the Web can be estimated at just over 5.2 billion pages; and that less than 7 per cent of Web pages fully comply with some variant of (X)HTML.

DiffServer: Application Level Differentiated Services for Web Servers

Web content hosting, in which a Web server stores and provides Web access to documents for different customers, is becoming increasingly common. For example, a web server can host webpages for several different companies and individuals. Traditionally, Web Service Providers (WSPs) provide all customers with the same level of performance (best-effort service). Most service differentiation has been in the pricing structure (individual vs. business rates) or the connectivity type (dial-up access vs. leased line, etc.). This report presents DiffServer, a program that implements two simple, server-side, application-level mechanisms (server-centric and client-centric) to provide different levels of web service. The results of the experiments show that there is not much overhead due to the addition of this additional layer of abstraction between the client and the Apache web server under light load conditions. Also, the average waiting time for high priority requests decreases significantly after they are assigned priorities as compared to a FIFO approach.