Assisted Visual Data Exploration for Discovering Information in Digital Libraries

Human access to data plays an increasing role, due to the existing and increasing amount of data. Different disciplines in computer science face this challenge with a variety of approaches, techniques and systems. One promising way is the visual access to the increasing amount of data. Information visualization and visual analytics technologies enable the process of visual exploration. The advantages of visual exploration are that users can discover new and unknown areas and topics in a certain domain. Today's visualization technologies have various limitations that restrict a real access to huge amount of data for searching, analyzing and exploration purposes. The main limitation is the fact that there exists no approach for the entire process of data transformation with data integration, information extraction and visualization for accessing huge amount of data. Most of the systems use either just one dataset, do not use automatic information extraction for enriching data or make no use of interactive visualization methods. A coherent process for integrating data, extracting information and visualizing the information as proposed for visual analytics is missing for real data and heterogeneous data sources. We propose in this thesis such a coherent model for data from different sources. Our model enhances existing approaches by integrating the entire data transformation process for information visualization to support the exploration process. For this, we use ground-truth data from the DBLP digial library as foundation and integrate data from a variety of resources, such as IEEE Xplore or Springer Link. Our main goal is to make use of the surpluses of the different technologies, such as information visualization, visual analytics, data integration and mining, and models of exploratory search to provide a coherent approach that supports the user in the entire exploration and knowledge acquisition process. We propose in this thesis an approach, that makes use of a reference model of interactive information visualization and enhances it with data integration, text and data mining, and assisted search approaches. Further, we implement our model with real world data and enable a proof of feasibility of the entire proposed approach