Lignes du temps connectées pour l'analyse visuelle d'adaptations et de relations entre artistes

de l'artiste). Les listes d'artistes énumèrent les artistes qui ont inspiré, collaboré ou se sont inspirés de l'artiste. La notice de chanson affiche des informations supplémentaires sur une chanson sélectionnée. Le résumé de l'artiste donne une vue globale sur les rôles occupés et les types de productions de l'artiste.

Abstract

The rise of Linked Open Data in the recent years is emocratizing access to complex data structures, usually taking the form of very large multivariate graphs. These are typically hard to understand for humans and require the use of visualization tools. While this data structure is a very well-researched topic in data visualization, large and complex graphs — with several heterogeneous node and link attributes — are hard to visualize with existing techniques. During a project in collaboration with the national library and archives agency of the Quebec government, we have been mandated to visualize a complex dataset of relationships between artists. Eleven key exploratory tasks were selected during the initial design process. Standard graph visualization techniques weren't suited to answers those specific tasks. Furthermore, since the dataset's structure is an unusual combination of a bipartite graph and trees, a thorough literature review found very few related or similar work. This research proposes a novel visualization method called MuzLink to efficiently explore and discover relationships between artists. The objective is to evaluate the effectiveness of this method for the selected task with a formal user study. Based on all eleven selected tasks, a fully functional prototype was developed. The proposed solution is a multi-facet visualization tool centered around a specific artist. The first view contains 3 connected timelines: (1) the inspirations, (2) the productions, and (3) the influences. Beeswarms of songs are positioned on the timelines according to their relationship with the artist. Songs on the middle timeline are the artist's production. Songs that inspired the artist are on the top timeline. Songs inspired by the artist are on the bottom timeline. Links are drawn between songs having an influence relationships. This connected timeline allows a topological exploration of inspiration relationships around an artist's musical production. The second view shows inspired, collaborators, and influenced artists ordered by their number of implications with the current artist. Interactions between the two complementary views allow answering a larger range of tasks. The user can easily navigate between artists with a search bar and by clicking on related artists. Clicking on a song reveals more details regarding the relationships among other songs.