Context Models for Understanding Images and Videos

A computer vision system that has to interact in natural language needs to understand the visual appearance of interactions between objects along with the appearance of objects themselves. Relationships between objects are frequently mentioned in queries of tasks like semantic image retrieval, image captioning, visual question answering and natural language object detection. Hence, it is essential to model context between objects for solving these tasks. In the first part of this thesis, we present a technique for detecting an object mentioned in a natural language query. Specifically, we work with referring expressions which are sentences that identify a particular object instance in an image. In many referring expressions, an object is described in relation to another object using prepositions, comparative adjectives, action verbs etc. Our proposed technique can identify both the referred object and the context object mentioned in such expressions. Context is also useful for incrementally understanding scenes and videos. In the second part of this thesis, we propose techniques for searching for objects in an image and events in a video. Our proposed incremental algorithms use the context from previously explored regions to prioritize the regions to explore next. The advantage of incremental understanding is restricting the amount of computation time and/or resources spent for various detection tasks. Our first proposed technique shows how to learn context in indoor scenes in an implicit manner and use it for searching for objects. The second technique shows how explicitly written context rules of one-on-one basketball can be used to sequentially detect events in a game.

Abstractive Long Document Summarization: Studio e Sperimentazione di Modelli Generativi Retrieval-Augmented

In questa tesi si trattano lo studio e la sperimentazione di un modello generativo retrieval-augmented, basato su Transformers, per il task di Abstractive Summarization su lunghe sentenze legali. La sintesi automatica del testo (Automatic Text Summarization) è diventata un task di Natural Language Processing (NLP) molto importante oggigiorno, visto il grandissimo numero di dati provenienti dal web e banche dati. Inoltre, essa permette di automatizzare un processo molto oneroso per gli esperti, specialmente nel settore legale, in cui i documenti sono lunghi e complicati, per cui difficili e dispendiosi da riassumere. I modelli allo stato dell’arte dell’Automatic Text Summarization sono basati su soluzioni di Deep Learning, in particolare sui Transformers, che rappresentano l’architettura più consolidata per task di NLP. Il modello proposto in questa tesi rappresenta una soluzione per la Long Document Summarization, ossia per generare riassunti di lunghe sequenze testuali. In particolare, l’architettura si basa sul modello RAG (Retrieval-Augmented Generation), recentemente introdotto dal team di ricerca Facebook AI per il task di Question Answering. L’obiettivo consiste nel modificare l’architettura RAG al fine di renderla adatta al task di Abstractive Long Document Summarization. In dettaglio, si vuole sfruttare e testare la memoria non parametrica del modello, con lo scopo di arricchire la rappresentazione del testo di input da riassumere. A tal fine, sono state sperimentate diverse configurazioni del modello su diverse tipologie di esperimenti e sono stati valutati i riassunti generati con diverse metriche automatiche.

Social Presence Approach Within the Question and Answering eLearning Model: An Experiment with a Multi-Agent System

The model of Questions Answering (Q&A) for eLearning is based on collaborative learning through questions that are posed by students and their answers to that questions which are given by peers, in contrast with the classical model in which students ask questions to the teacher only. In this proposal we extend the Q&A model including the social presence concept and a quantitative measure of it is proposed; besides it is considered the evolution of the resulting Q&A social network after the inclusion of the social presence and taking into account the feedback on questions posed by students and answered by peers. The social network behaviorwas simulated using a Multi-Agent System to compare the proposed social presence model with the classical and the Q&A models

Comparative analysis: Public and private school management systems

A Work Project, presented as part of the requirements for the Award of a Masters Degree in Economics from the NOVA – School of Business and Economics

Comparateive analysis: Public and private school management systems

The Portuguese educational system has counted, for many years, with the co-existence of both public and private schools. In fact, the country’s growth and development led, in the past, to an increasing demand for free of charge public education that could only be matched through the creation of “publicly-subsidized and privately owned and managed schools”. Still, the demographic evolution of Portugal recently generated a decrease on the demand for public educational services. This situation has raised doubts about the true contribution of this type of school for the public education system. This paper aims at answering this question by isolating the impact of different property and management schemes on the performance of students, resorting to cross-section data on 9th grade students from 2010. The results corroborate the well known result on the relevance of the family socio-economic background for students’ performance, but do also sustain the existence of a significant positive impact of private ownership and management schemes on the overall performance of students. These results suggest that there might be gains associated with the expansion of such schemes within the public education system.

How much flower-rich habitat is enough for wild pollinators? Answering a key policy question with incomplete knowledge

In 2013, an opportunity arose in England to develop an agri-environment package for wild pollinators, as part of the new Countryside Stewardship scheme launched in 2015. It can be understood as a 'policy window', a rare and time-limited opportunity to change policy, supported by a narrative about pollinator decline and widely supported mitigating actions. An agri-environment package is a bundle of management options that together supply sufficient resources to support a target group of species. This paper documents information that was available at the time to develop such a package for wild pollinators. Four questions needed answering: (1) Which pollinator species should be targeted? (2) Which resources limit these species in farmland? (3) Which management options provide these resources? (4) What area of each option is needed to support populations of the target species? Focussing on wild bees, we provide tentative answers that were used to inform development of the package. There is strong evidence that floral resources can limit wild bee populations, and several sources of evidence identify a set of agri-environment options that provide flowers and other resources for pollinators. The final question could only be answered for floral resources, with a wide range of uncertainty. We show that the areas of some floral resource options in the basic Wild Pollinator and Farmland Wildlife Package (2% flower-rich habitat and 1 km flowering hedgerow), are sufficient to supply a set of six common pollinator species with enough pollen to feed their larvae at lowest estimates, using minimum values for estimated parameters where a range was available. We identify key sources of uncertainty, and stress the importance of keeping the Package flexible, so it can be revised as new evidence emerges about how to achieve the policy aim of supporting pollinators on farmland.

Enhancing QA systems with complex temporal question processing capabilities

This paper presents a multilayered architecture that enhances the capabilities of current QA systems and allows different types of complex questions or queries to be processed. The answers to these questions need to be gathered from factual information scattered throughout different documents. Specifically, we designed a specialized layer to process the different types of temporal questions. Complex temporal questions are first decomposed into simple questions, according to the temporal relations expressed in the original question. In the same way, the answers to the resulting simple questions are recomposed, fulfilling the temporal restrictions of the original complex question. A novel aspect of this approach resides in the decomposition which uses a minimal quantity of resources, with the final aim of obtaining a portable platform that is easily extensible to other languages. In this paper we also present a methodology for evaluation of the decomposition of the questions as well as the ability of the implemented temporal layer to perform at a multilingual level. The temporal layer was first performed for English, then evaluated and compared with: a) a general purpose QA system (F-measure 65.47% for QA plus English temporal layer vs. 38.01% for the general QA system), and b) a well-known QA system. Much better results were obtained for temporal questions with the multilayered system. This system was therefore extended to Spanish and very good results were again obtained in the evaluation (F-measure 40.36% for QA plus Spanish temporal layer vs. 22.94% for the general QA system).

Irrigation development. History, customs, laws, and administrative systems relating to irrigations, water-courses, and waters in France, Italy, and Spain. The introductory part of the Report of the state engineer of California, on irrigation and the irrigation question.

"Authorities" at end of each chapter.

Estimates of the real structured radius of stability of linear dynamic systems

A question is examined as to estimates of the norms of perturbations of a linear stable dynamic system, under which the perturbed system remains stable in a situation R:here a perturbation has a fixed structure.

Holt, Johnson and the 1966 Federal Election. A Question of Causality

US President Lyndon Johnson's state visit to Australia in October 1966, came at the pinnacle of support for Australia's military involvement in the Vietnam War. Johnson's visit also occurred just weeks before an election for the House of Representatives at which the ruling Liberal-Country Party Coalition won its eighth successive, and largest victory, The proximity of these events has led many to argue that a causal relationship exists between the two. Advocates of this thesis, however, have failed to support their position with any evidence other than the anecdotal. Contrary to the assertions made by numerous political historians and observers of the period, this paper finds no evidence to support a thesis of causality. This paper argues that the Coalition's landslide victory in 1966 was both a rejection of the tired and lacklustre leadership of Labor's Arthur Calwell and a measure of the electorate's overwhelming support for Holt and his Government's policies of conscription and military involvement in Vietnam.

Research commentary: Information systems and conceptual modeling - A research agenda

Within the information systems field, the task of conceptual modeling involves building a representation of selected phenomena in some domain. High-quality conceptual-modeling work is important because it facilitates early detection and correction of system development errors. It also plays an increasingly important role in activities like business process reengineering and documentation of best-practice data and process models in enterprise resource planning systems. Yet little research has been undertaken on many aspects of conceptual modeling. In this paper, we propose a framework to motivate research that addresses the following fundamental question: How can we model the world to better facilitate our developing, implementing, using, and maintaining more valuable information systems? The framework comprises four elements: conceptual-modeling grammars, conceptual-modeling methods, conceptual-modeling scripts, and conceptual-modeling contexts. We provide examples of the types of research that have already been undertaken on each element and illustrate research opportunities that exist.

Shopping for apparel: how can kiosk systems help?

When shopping for apparel, many consumers seek advice from friends and family or store personnel. In-store kiosk systems might serve as an alternative decision support system. In the present study we address the key question of how such kiosk systems are evaluated by consumers. We conducted three focus group discussions with regular apparel shoppers aged between 23 and 39 years. In sum, qualitative information from 15 participants was subject to a qualitative content analysis with the aim of gaining a more comprehensive understanding of how apparel shoppers experience the shopping process. Getting a more in-depth understanding of the needs and wishes associated with the apparel shopping process gives a basis for evaluating the potential acceptance of electronic decision support systems in apparel shopping. Although our study is exploratory in nature, we are able to draw an initial picture of how kiosk systems could be used in apparel shopping.

Group decision making and Quality-of-information in e-Health systems

Knowledge is central to the modern economy and society. Indeed, the knowledge society has transformed the concept of knowledge and is more and more aware of the need to overcome the lack of knowledge when has to make options or address its problems and dilemmas. One’s knowledge is less based on exact facts and more on hypotheses, perceptions or indications. Even when we use new computational artefacts and novel methodologies for problem solving, like the use of Group Decision Support Systems (GDSSs), the question of incomplete information is in most of the situations marginalized. On the other hand, common sense tells us that when a decision is made it is impossible to have a perception of all the information involved and the nature of its intrinsic quality. Therefore, something has to be made in terms of the information available and the process of its evaluation. It is under this framework that a Multi-valued Extended Logic Programming language will be used for knowledge representation and reasoning, leading to a model that embodies the Quality-of-Information (QoI) and its quantification, along the several stages of the decision-making process. In this way, it is possible to provide a measure of the value of the QoI that supports the decision itself. This model will be here presented in the context of a GDSS for VirtualECare, a system aimed at sustaining online healthcare services.