13 February 2017-13 july 2017. Norbert Feron Internship (M2)

Visualization and exploration of massive data is a challenge. User interfaces provide tools to manipulate data and carry out information. Many tasks decline from this challenge and imply creating new exploration/manipulation techniques. Some of these techniques require user interaction. For example, the exploration process of a set of data can conduct to a dead-end, the user should be capable of backtracking to continue the search. It is difficult for the human working memory to keep track of all evolution steps. We propose to build an history of user actions, by the use of an “history graph”, saving user choices and providing an additional interactive visualization to assist user awareness during the exploration.

Functionalities:

  • Online building of current interaction history,
  • Interactive navigation including backtracking,
  • Compare differences between paths,
  • Offline replay and mining of the history graph.

Because of user interactions, the “history graph” appeared to be a dynamic graph. We have to define a data structure able to manage interaction history and graph representation to preserve user’s mental map.

In terms of visualization design we want to reveal more the evolution process than the time spent exploring the data. A “custom” technique has to be built to represent user’s progression in the data.

The trace layout algorithm must manage on-line drawing to provide direct response on user interactions on the data being explored, and also off-line representation when the trace is being replayed.