Towards Visualization Recommendation: A Semi-Automated Ecological Data Specific Learning Approach, GfÖ 2015,

Ecological  studies  produce highly  complex,  heterogeneous  and distributed  data from its wider research activities.  For efficiently communicating the research work and presentation of the related data, visualization plays an important role, due to its  ability  to  condense  large  amounts  of  data  into  effective  and  understandable graphics. The  decision  of  optimal  choice  of  visualization,  not only  produces more interpretable graphics, but support the community to understand, analyse the data and reuse it for their respective studies.  However, studies have shown that the potential of visualization has not been fully utilized in scientific journals, due to in-appropriate visualization selection with respect to the nature of data and message to convey.  This does not only impede analysis but also  results in misleading conclusions. To provide a solution for the problem of visualization selection, we propose a semi‐automated  context‐aware  visualization  recommendation  model. In  the model, information will be extracted from data and metadata, and annotated with suitable ecological operations (analytical tasks like spatial distribution, relative species abundance).  This information will  be mapped  to  the visualization  semantics; like in each extracted operation which variables are involved and how they are visually  represented.  This  helps  in deriving  the  relevant  visualizations  for  that  data. We  also  propose  an  interactive  learning  workflow  for  visualization  recommendation that will enrich the model from the knowledge gathered from each interaction with  the user. In our work, we will develop our base knowledge (which visualizations have been used  to represent what ecological operations)  from the visualization presented in the ecological publications. This knowledge is integral in making decisions based on  the current  trends in visualizations  for representing ecological concepts and data.