New perspectives on analyzing data from biological collections based on social network analytics

• Ano de defesa: 2018
• Autor(a) principal: Siracusa, Pedro Correia de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Laboratório Nacional de Computação Científica; Coordenação de Pós-Graduação e Aperfeiçoamento (COPGA); Brasil; LNCC; Programa de Pós-Graduação em Modelagem Computacional
• Palavras-chave em Português: Biodiversidade; Processamento de dados; biodiversity; Data processing; CNPQ::CIENCIAS BIOLOGICAS::ECOLOGIA; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO::SISTEMAS DE INFORMACAO
• Link de acesso: https://tede.lncc.br/handle/tede/279

Biological collections have been historically regarded as fundamental sources of scientific information on biodiversity, supporting a wide range of scientific and management initiatives in the scope of natural resources conservation. As they are typically composed of discrete records of specimens (most of which derived from non-random and opportunistic sampling), biological collection datasets are commonly associated with a variety of biases, which must be characterized and mitigated before data can be consumed. In this dissertation, we are particularly motivated by taxonomic and collector biases, which can be understood as the effect of particular recording preferences of key collectors on shaping the overall taxonomic composition of biological collections they contribute to. In this context, we propose two network models as the first steps towards a network-based conceptual framework for understanding the formation of biological collections as a result of the composition of collectors’ interests and activities. Both models extend the well-established framework of social network analytics, benefiting from a whole set of metrics and algorithms for characterizing network topological features. Species-Collector Networks (SCNs) model the interests of collectors towards particular species, and are structured by linking collectors to each species they have recorded in biological collection datasets. From complementary perspectives, SCNs allow one to investigate which collectors share common interest for sets of species; and conversely, which species are usually recorded by similar sets of collectors. Collector CoWorking Networks (CWNs) are a special type of collaboration networks, structured from collaboration ties that are formed between collectors who record specimens together in field. Such collaborative ties are created between pairs of collectors whenever they are both included as collectors in the same record. Building upon the defined network models, we also present a case study in which we use our models to explore the community of collectors and the taxonomic composition of the University of Brasília herbarium. We describe general topological features of the networks and point out some of the most relevant collectors in the biological collection as well as their taxonomic groups of interest. We also investigate the collaborative behavior of collectors while recording specimens. Finally, we discuss future perspectives for incorporating temporal and geographical dimensions to the models. Moreover, we indicate some possible investigation directions that could possibly benefit from our approach based on social network analytics to model and analyze biological collections.