• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 2
  • Tagged with
  • 2
  • 2
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

TRIVIR: A Visualization System to Support Document Retrieval with High Recall / TRIVIR: Um sistema de visualização para apoio à recuperação de documentos com alta cobertura

Dias, Amanda Gonçalves 08 July 2019 (has links)
A high recall problem in document retrieval is described by scenarios in which one wants to ensure that, given one (or multiple) query document(s), (nearly) all relevant related documents are retrieved, with minimum human effort. The problem may be expressed as a document similarity search: a user picks an example document (or multiple ones), and an automatic system recovers similar ones from a collection. This problem is often handled with a so-called Continuous Active Learning strategy: given the initial query, which is a document described by a set of relevant terms, a learning method returns the most-likely relevant documents (e.g., the most similar) to the reviewer in batches, the reviewer labels each document as relevant/not relevant and this information is fed back into the learning algorithm, which uses it to refine its predictions. This iterative process goes on until some quality condition is satisfied, which might demand high human effort, since documents are displayed as ranked lists and need to be labeled individually, and impact negatively the convergence of the learning algorithm. Besides, the vocabulary mismatch issue, i.e., when distinct terminologies are employed to describe semantically related or equivalent concepts, can impair recall capability. We propose TRIVIR, a novel interactive visualization tool powered by an information retrieval (IR) engine that implements an active learning protocol to support IR with high recall. The system integrates multiple graphical views in order to assist the user identifying the relevant documents in a collection. Given representative documents as queries, users can interact with the views to label documents as relevant/not relevant, and this information is used to train a machine learning (ML) algorithm which suggests other potentially relevant documents. TRIVIR offers two major advantages over existing visualization systems for IR. First, it merges the ML algorithm output into the visualization, while supporting several user interactions in order to enhance and speed up its convergence. Second, it tackles the vocabulary mismatch problem, by providing terms synonyms and a view that conveys how the terms are used within the collection. Besides, TRIVIR has been developed as a flexible front-end interface that can be associated with distinct text representations and multidimensional projection techniques. We describe two use cases conducted with collaborators who are potential users of TRIVIR. Results show that the system simplified the search for relevant documents in large collections, based on the context in which the terms occur. / No âmbito de recuperação de documentos, há situações em que é preciso assegurar que todos os documentos relevantes para uma dada consulta serão recuperados, de preferência com um esforço humano mínimo. Uma das maneiras de formular este problema de recuperação com alta cobertura é com uma consulta por similaridade: um usuário seleciona um (ou vários) documento(s), e um sistema automático é utilizado para recuperar, de uma coleção, os documentos semelhantes aos apresentados. Uma maneira usual de abordar o problema adota uma estratégia denominada Continuous Active Learning, em que dado o(s) documento(s) de consulta, descrito por seus termos relevantes, um método de aprendizado de máquina retorna e apresenta ao analista, em lotes, os documentos mais provavelmente relevantes, ou mais similares a esse(s). O analista classifica cada documento quanto à relevância, realimentando o algoritmo de aprendizado, o qual pode então refinar suas previsões. Esse processo interativo continua até que alguma condição de qualidade seja satisfeita, o que pode exigir grande esforço do usuário, já que os documentos são oferecidos no formato de listas ranqueadas e devem ser marcados individualmente, e impactar negativamente a convergência do algoritmo de aprendizado. Ademais, uma das dificuldades é a incompatibilidade de vocabulário, quando terminologias distintas são empregadas para descrever conceitos semanticamente relacionados, o que pode prejudicar a identificação dos documentos relevantes. Neste trabalho propomos TRIVIR, uma visualização interativa alimentada por um motor de recuperação de informação (RI) que implementa o protocolo Continuous Active Learning com o fim de auxiliar RI de alta cobertura. O sistema integra várias representações gráficas para auxiliar o usuário a identificar documentos relevantes em uma coleção. Dados documentos representativos como entrada, usuários podem interagir com as visualizações e marcar documentos como relevantes/não relevantes. Esta informação é utilizada para treinar um algoritmo de aprendizado de máquina que, por sua vez, sugere documentos potencialmente relevantes. TRIVIR oferece duas principais vantagens em relação a outros sistemas de visualização para RI. Primeiro, integra a visualização a um algoritmo de aprendizado de máquina com o qual usários podem interagir para melhorar e acelerar a convergência do algoritmo. Segundo, o sistema trata o problema de incompatibilidade de vocabulário, provendo sinônimos dos termos e o contexto no qual termos são utilizados na coleção. TRIVIR foi desenvolvido como uma interface web flexível podendo ser associado com diferentes técnicas de representação de documentos e projeção multidimensional. Descrevemos dois casos de uso conduzidos com potenciais usuários do TRIVIR. Resultados mostraram que o sistema facilitou a pesquisa por documentos relevantes em grandes coleções, por meio da utilização da informação do contexto no qual os termos ocorrem.
2

Paul Verhoeven, media manipulation, and hyper-reality

Malchiodi, Emmanuel William 01 May 2011 (has links)
Does the individual really matter in the post-modern world, brimming with countless signs and signifiers? My main objective in this writing is to demonstrate how this happens in Verhoeven's films, exploring his central themes and subtext and doing what science fiction does: hold a mirror up to the contemporary world and critique it, asking whether our species' current trajectory is beneficial or hazardous.; Dutch director Paul Verhoeven is a polarizing figure. Although many of his American made films have received considerable praise and financial success, he has been lambasted on countless occasions for his gratuitous use of sex, violence, and contentious symbolism--1995s Showgirls was overwhelmingly dubbed the worst film of all time and 1997s Starship Troopers earned him a reputation as a fascist. Regardless of the controversy surrounding him, his science fiction films are a move beyond the conventions of the big blockbuster science fiction films of the 1980s (E.T. and the Star Wars trilogy are prime examples), revealing a deeper exploration of both sociopolitical issues and the human condition. Much like the novels of Philip K. Dick (and Verhoeven's 1990 film Total Recall--an adaptation of a Dick short story), Verhoeven's science fiction work explores worlds where paranoia is a constant and determining whether an individual maintains any liberty is regularly questionable. In this thesis I am basically exploring issues regarding power. Although I barely bring up the term power in it, I feel it is central. Power is an ambiguous term; are we discussing physical power, state power, objective power, subjective power, or any of the other possible manifestations of the word? The original Anglo-French version of power means "to be able," asking whether it is possible for one to do something. In relation to Verhoeven's science fiction work each demonstrates the limitations placed upon an individual's autonomy, asking are the protagonists capable of independent agency or rather just environmental constructs reflecting the myriad influences surrounding them.

Page generated in 0.0605 seconds