Submitted by Fabrício Silva Assumpção null (assumpcao.f@gmail.com) on 2018-02-18T22:51:55Z No. of bitstreams: 1 modelo-para-publicacao-dados-de-autoridade-linked-data.pdf: 3758953 bytes, checksum: b931683bbc9f76cdbb096f52e63ef88f (MD5) / Approved for entry into archive by Satie Tagara (satie@marilia.unesp.br) on 2018-02-19T17:10:03Z (GMT) No. of bitstreams: 1 assumpcao_fs_dr_mar.pdf: 3758953 bytes, checksum: b931683bbc9f76cdbb096f52e63ef88f (MD5) / Made available in DSpace on 2018-02-19T17:10:03Z (GMT). No. of bitstreams: 1 assumpcao_fs_dr_mar.pdf: 3758953 bytes, checksum: b931683bbc9f76cdbb096f52e63ef88f (MD5) Previous issue date: 2018-02-05 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / A Ciência da Informação tem entre suas preocupações o acesso à informação e aos recursos informacionais, sendo, portanto, objetos de seu interesse os instrumentos utilizados para esse acesso, tais como os catálogos, que incluem dados bibliográficos (representações de recursos informacionais) e dados de autoridade (representações das entidades associadas aos recursos informacionais, tais como pessoas, entidades coletivas e conceitos). A proposta de criação de uma Web Semântica, em que os dados sejam processados não somente por sua sintaxe, mas também por sua semântica, tem impulsionado o desenvolvimento de um conjunto de tecnologias para a representação de dados na Web, assim como para a consulta a esses dados e o raciocínio computadorizado a partir deles. O uso de algumas dessas tecnologias para a publicação e o relacionamento de dados levou ao surgimento do conceito de Linked Data, e o anseio por sua aplicação na Ciência da Informação deu origem a projetos para a publicação de dados de autoridade como Linked Data. No entanto, observa-se que esses projetos, ainda em estágios iniciais ou experimentais, carecem de um quadro teórico construído na Ciência da Informação que possa orientar quanto às políticas, aos procedimentos e às tecnologias empregadas na publicação desses dados. Assim, partindo do problema “como publicar dados de autoridade como Linked Data?” são delineados o objetivo geral – propor, a partir do conceito das funções dos dados de autoridade nos catálogos e de seus benefícios nos ambientes de Linked Data, um modelo de publicação de dados de autoridade como Linked Data compreendendo políticas, procedimentos e tecnologias – a tese e a hipótese desta pesquisa, de caráter bibliográfico, documental e metodológico. Para o alcance desse objetivo, primeiramente é conduzida uma revisão de literatura acerca do controle de autoridade e do desenvolvimento dos dados de autoridade nos catálogos em livros, em fichas e digitais, com destaque para os modelos conceituais FRAD e FRSAD que sintetizam as funções desempenhadas pelos dados de autoridade. Em seguida, são apresentados o conceito de Linked Data e as principais tecnologias da Web Semântica relacionadas a ele: URIs, RDF, RDFS e OWL, apresentação essa que serve de base para a descrição de três vocabulários que podem ser utilizados na publicação de dados de autoridade (SKOS, MADS/RDF e RDA Element Sets), de iniciativas para a publicação desses dados (LC Linked Data Service, datos.bne.es, data.bnf.fr, VIAF e AGROVOC) e dos potenciais benefícios da publicação dos dados de autoridade como Linked Data. A partir dos resultados dessa revisão de literatura, é proposto o modelo para a publicação de dados de autoridade como Linked Data, compreendendo as etapas de planejamento; modelagem e mapeamento; tratamento, relacionamento e conversão; publicação; e feedback e retroalimentação. Após a descrição de cada etapa do modelo, com suas políticas, procedimentos e tecnologias, são apresentadas considerações finais sobre os resultados alcançados e sobre o modelo proposto. / One of the issues that Information Science is concerned with is the access to information resources, therefore the studies in this area include the library catalogs, which comprise bibliographic data (representations of information resources) and authority data (representations of the entities related to the information resources, such as persons, corporate body and concepts). The proposal of a Semantic Web in which data are processed not just by their syntax but also by their semantics has led to the development of a set of technologies for publishing and linking data on the Web, as well as technologies for data querying and for the computerized reasoning. The use of a subset of these technologies for publishing and linking data has led to the Linked Data concept. The experiments with these technologies in Information Science have given rise to projects for publishing authority data as Linked Data. However, we observed that these projects, still in initial or experimental stages, are not based on a theoretical framework developed in Information Science that can guide them regarding to the policies, to the procedures and to the technologies used in the publication of these data. Thus, starting from the question “how to publish authority data as Linked Data?” we define the main goal – conceptualize the functions of authority data in the catalogs and their benefits in Linked Data environments in order to propose a model for authority data publishing, as well as highlighting its policies, procedures and technologies – the thesis and the hypothesis of this research. To achieve this goal, first we made a literature review about authority control and the development of authority data in books, cards and digital catalogs, remarking the FRAD and FRSAD conceptual models which synthesize the functions of authority data. Then, we present the Linked Data concept and the main Semantic Web technologies related to it: URIs, RDF, RDFS and OWL; starting from this introduction, we describe three vocabularies that can be used to publishing authority data (SKOS, MADS/RDF and RDA Element Sets), some initiatives (Library of Congress Linked Data Service, datos.bne.es, data.bnf.fr, VIAF and AGROVOC) and the advantages of authority data published as Linked Data. Based on the results of this literature review, we propose a model for publishing authority data as Linked Data, The model comprases the following stages: planning; modeling and mapping; processing, linking and conversion; publishing; and feedback. After describing each of these stages, with its policies, procedures and technologies, we present the conclusions about the results and about the proposed model.

Angesichts der zentralen Rolle der Root-Zertifizierungsstellen als Vertrauensanker der Web PKI und der in der Vergangenheit aufgetretenen Vorfälle mit unberechtigt oder inkorrekt ausgestellten Zertifikaten, ist die Transparenz und Verantwortlichkeit dieser Root CAs von großer Bedeutung. Seit der Einführung von Certificate Transparency Logs werden alle von Certificate Authorities ausgestellten Zertifikate in diese öffentlichen Logs eingetragen. Die Arbeit stellt die Problematik der eingeschränkten Zugänglichkeit dieser Daten für die Wissenschaft dar und entwickelt ein Werkzeug, dass eine unabhängige Aufzeichnung und Auswertung von Certificate Transparency Logs ermöglicht. Das entwickelte System nutzt eine containerbasierte Architektur und Elasticsearch zur effizienten Speicherung und Analyse der Daten. Es bewältigt ein hohes Datenaufkommen von durchschnittlich 25 Millionen Log-Einträgen pro Tag und ermöglicht eine anpassbare Datenverarbeitung und -auswertung. Die Vorverarbeitung und Indexierung sowie die Auswertung der Daten erfolgte mit Python, was eine flexible Anpassung des Systems an unterschiedliche Forschungsfragen erlaubt. Über einen Zeitraum von 42 Tagen wurden insgesamt 645 Millionen CT Log-Einträge aufgezeichnet und analysiert. Aus den Auswertungen geht hervor, wie verschiedene CAs und deren Root-Zertifikate genutzt werden und wie stark die unterschiedlichen CT Logs von CAs verwendet werden. Die Arbeit identifiziert jedoch auch Herausforderungen, wie den hohen Speicherbedarf und notwendige Optimierungen in der Datenindexierung.:1 Einleitung 1.1 Problemstellung 1.2 Zielstellung 2 Grundlagen 2.1 X509-Zertifikate 2.1.1 Felder 2.1.2 Erweiterungen 2.2 Certificate Transparency 2.2.1 Certificate Transparency Log 2.2.2 Überprüfung durch User Agents 2.2.3 Überprüfung durch Monitors 2.2.4 Eintragung durch Certificate Authorities 3 Konzeptionierung 3.1 Abfrage der CT Logs 3.2 Verarbeitung der Zertifikate 3.3 Speicherung & Auswertung der Daten 3.4 Überwachung 3.5 Docker 4 Implementierung 4.1 Plattform 4.2 Überwachung 4.3 certstream-server 4.4 Verarbeitung 4.4.1 Pufferung (stream-to-queue-publisher) 4.4.2 Vorverarbeitung (cert-indexer) 4.5 Elasticsearch 4.5.1 Speicherverbrauch 4.5.2 Field Mappings 5 Auswertung 5.1 Logs & Log-Betreiber 5.2 Certificate Authorites 5.3 Zertifikats-Größe 5.4 Gültigkeitsdauer 6 Schluss 6.1 Fazit 6.2 Ausblick A Beispiel X509 Leaf-Zertifikat B Beispiel X509 Root-Zertifikat C Beispiele Elasticsearch Abfragen Literatur Abbildungsverzeichnis Tabellenverzeichnis / In view of the central role of the root certification authorities as trust anchors of the Web PKI and the incidents that have occurred in the past with unauthorised or incorrectly issued certificates, the transparency and accountability of these root CAs is of great importance. With the introduction of Certificate Transparency Logs, all certificates issued by Certificate Authorities are now entered in public logs. The work presents the problem of the limited accessibility of this data for science and develops a tool that enables an independent recording and evaluation of Certificate Transparency Logs. The developed system uses a container-based architecture and Elasticsearch to efficiently store and analyse the data. It can handle a high volume of data, averaging 25 million log entries per day, and enables customisable data processing and analysis. Python was used to pre-process, index and analyse the data, allowing the system to be flexibly adapted to different research questions. A total of 645 million CT log entries were recorded and analysed over a period of 42 days. The analyses show how different CAs and their root certificates are used and how much the different CT logs are used by CAs. However, the work also identifies challenges, such as the high memory requirements and necessary optimisations in data indexing.:1 Einleitung 1.1 Problemstellung 1.2 Zielstellung 2 Grundlagen 2.1 X509-Zertifikate 2.1.1 Felder 2.1.2 Erweiterungen 2.2 Certificate Transparency 2.2.1 Certificate Transparency Log 2.2.2 Überprüfung durch User Agents 2.2.3 Überprüfung durch Monitors 2.2.4 Eintragung durch Certificate Authorities 3 Konzeptionierung 3.1 Abfrage der CT Logs 3.2 Verarbeitung der Zertifikate 3.3 Speicherung & Auswertung der Daten 3.4 Überwachung 3.5 Docker 4 Implementierung 4.1 Plattform 4.2 Überwachung 4.3 certstream-server 4.4 Verarbeitung 4.4.1 Pufferung (stream-to-queue-publisher) 4.4.2 Vorverarbeitung (cert-indexer) 4.5 Elasticsearch 4.5.1 Speicherverbrauch 4.5.2 Field Mappings 5 Auswertung 5.1 Logs & Log-Betreiber 5.2 Certificate Authorites 5.3 Zertifikats-Größe 5.4 Gültigkeitsdauer 6 Schluss 6.1 Fazit 6.2 Ausblick A Beispiel X509 Leaf-Zertifikat B Beispiel X509 Root-Zertifikat C Beispiele Elasticsearch Abfragen Literatur Abbildungsverzeichnis Tabellenverzeichnis

Magister Library and Information Studies - MLIS / Although Resource Description and Access (RDA) has been discussed extensively amongst the ranks of cataloguers internationally, no research on the perceptions of South African cataloguers was available at the time of this research. The aim of this study was to determine how well RDA was faring during the study's timeframe, to give a detailed description regarding cataloguer perceptions within a higher education setting in South Africa. Furthermore, to determine whether the implementation of RDA has overcome most of the limitations that AACR2 had within a digital environment, to identify advantages and/or perceived limitations of RDA as well as to assist cataloguers to adopt and implement the new standard effectively. The study employed a qualitative research design assisted by a phenomenological philosophy to gain insight into how cataloguers experienced the implementation and adoption of RDA by means of two concurrent web-based questionnaires. The study concluded that higher education cataloguing professionals residing in the Western Cape were decidedly positive towards the new cataloguing standard. Although there were some initial reservations, they were overcome to such an extent that ultimately no real limitations were identified, and that RDA has indeed overcome most of the limitations displayed by AACR2. Many advantages of RDA were identified, and participants expressed excitement about the future capabilities of RDA as it continues toward a link-data milieu, making library metadata more easily available.

Magister Library and Information Studies - MLIS / Although Resource Description and Access (RDA) has been discussed extensively amongst the ranks of cataloguers internationally, no research on the perceptions of South African cataloguers was available at the time of this research. The aim of this study was to determine how well RDA was faring during the study's timeframe, to give a detailed description regarding cataloguer perceptions within a higher education setting in South Africa. Furthermore, to determine whether the implementation of RDA has overcome most of the limitations that AACR2 had within a digital environment, to identify advantages and/or perceived limitations of RDA as well as to assist cataloguers to adopt and implement the new standard effectively. The study employed a qualitative research design assisted by a phenomenological philosophy to gain insight into how cataloguers experienced the implementation and adoption of RDA by means of two concurrent web-based questionnaires. The study concluded that higher education cataloguing professionals residing in the Western Cape were decidedly positive towards the new cataloguing standard. Although there were some initial reservations, they were overcome to such an extent that ultimately no real limitations were identified, and that RDA has indeed overcome most of the limitations displayed by AACR2. Many advantages of RDA were identified, and participants expressed excitement about the future capabilities of RDA as it continues toward a link-data milieu, making library metadata more easily available. As this research has revealed a distinctly positive attitude from cataloguers' two main matters for future research remains, being: ? Why South African participants in this study voiced almost no perceived limitations to RDA as a cataloguing standard. Future research might be able to relay information regarding this trend, especially in the light that it was not a global phenomenon. ? A deeper look might have to be taken at how participants' experienced RDA training as this phenomenon might be closely linked to the reasons why the participants did not mention more limitations.