A data-driven visual approach to explore linked open data environments

Resumen

Humans have registered their environment since ancient times, either verbally, through writing or by using graphics and images. These records have allowed us to learn from the past, analysing previous scenarios and extracting new knowledge that has transformed our societies, making us capable to address and deal with new challenges. Since the invention of the World Wide Web by Sir Tim Berners-Lee, the ease to publish, update, discover and access new data has grown exponentially, and it is estimated that every two years we generate as much data as in the whole history before. Tim Berners-Lee envisaged that machines could help humans in data processing and understanding tasks, giving birth to the Semantic Web field, an scenario in which data is provided together with semantic annotations, allowing its comprehension by algorithms. Years later, in 2006, the Linked Data principles were proposed as a method of publishing structured facts, so that they could be connected (linked) to other resources through the World Wide Web. It relies on standard Web technologies, and is intended to be consumed by computers. Despite the benefits brought by Linked Data, the adoption of its related developments has normalised after the initial years, and little attempts are performed outside the research community. To make Internet users aware of Linked Data’s advantages, we propose an approach to explore its datasets using visual means, relying on our ability to discover patterns and insights through graphic imageries and depictions. In order to deal with the diversity of structured data published as Linked Data, our proposal takes a data-driven approach, that is, we base our whole analysis on the data itself, avoiding preconceptions that might lead to wrong conclusions. The main objective is to ease semantic data exploration through suitable visualizations, making any user able to interact with novel datasets with no prior knowledge nor skills required. In this dissertation, we explain the visualization pipeline that allows to take raw semantic data as input, and produces visual representations as output, together with the involved modules and the contributions we have designed and implemented to push forward the State of the Art on Linked Data Visualization.