Exploiting Heterogeneity

Event Date: 

Thursday, October 11, 2012 - 12:00pm

Event Location: 

  • SSMS 2135
Krzysztof Janowicz
Assistant Professor, Dept. of Geography, UC Santa Barbara
The so-called Fourth Paradigm describes a novel, data-intensive approach to scientific discovery. It is often characterized as the scientific perspective on Big Data. The underlying assumption is that the availability of data at a higher spatial, temporal, and thematic resolution will enable us to answer complex scientific and social questions that cannot be answered from within one domain but span across multiple disciplines. It will also result in a more holistic understanding of phenomena that are a consequence of the interplay of physical and social factors, e.g., global change. To address the arising new challenges and possibilities will require novel data management infrastructures that can handle scientific data across domains, embed data in scientific frameworks, and feed them into complex models and simulations. Surprisingly, at their very core these envisioned new infrastructures are mostly distributed and integrated knowledge archives. The crucial first step, namely data publishing and retrieval, are mostly neglected and approached following old paradigms and often (leaving simple taxonomies aside) semantics-free keyword search in metadata catalogs. Instead, next-generation knowledge infrastructures could be envisioned as distributed knowledge engines. For example, using analogy-based reasoning such engines could automatically propose relevant data sources and study areas for evaluation. Finally, working with interdisciplinary data and models will require novel interaction paradigms and user interfaces that actively support scholars in finding relevant data. Instead of requiring precise queries and a detailed knowledge of the accessed data and catalogs, these new interfaces should support browsing and navigating the global graph of interlinked data. This talk will give an overview and open the discussion how heterogeneous data sources can be exploited to answer complex search questions and how interaction paradigms can support users in exploring these data.