Abstract
Traditional social bookmarking systems often rely on linear lists or rigid hierarchies, obscuring the rich, multidimensional relationships between users, tags, and web resources. This paper introduces GalViz, an interactive network visualization interface designed to map these "tripartite" relationships. By presenting bookmarks as a contextual graph, GalViz enables users to explore semantic connections and discover content through a "social view" that integrates personal and global data. Preliminary log analysis indicates that this visual approach fosters higher user engagement, faster task completion, and more meaningful interactions compared to standard list-based interfaces.
Introduction: The Limitations of Linear Lists
Social tagging systems, such as Delicious or Flickr, are built upon a foundational "triple": the association between a user, a tag, and a web resource. Managing this data is inherently complex due to its non-linear and multidimensional nature. However, most existing bookmark managers force this rich information into simple list views, flat tag clouds, or rigid folder hierarchies. These design choices obscure valuable contextual information, masking the socio-semantic connections between users with shared interests and making it difficult to discover new content outside of a user's immediate collection.
The GalViz Interface
To address these shortcomings, the authors developed GalViz, a tool integrated into the GiveALink platform. Instead of a static list, GalViz presents bookmarks as an interactive network graph.
Key features of this approach include:
- Network Visualization: Nodes represent web resources or tags, with edges connecting related objects. Users can manipulate nodes directly via drag-and-drop and contextual menus.
- Contextual Maps: The system visualizes a web page's semantic context—related tags and pages—helping users understand a resource at a glance without deep reading.
- Social View: Users can toggle a view that expands their personal graph with globally related triples from the community, facilitating discovery through collective intelligence.
Evaluation and Results
The system was evaluated using log data from the GiveALink platform, consisting of 38,181 actions from 6,310 users. During an A/B testing period, users were randomly assigned to either a traditional "list mode" or the new "visualization mode."
The preliminary log study revealed several distinct advantages for the visualization interface:
- Higher Engagement: The visualization mode recorded 1,192 sessions, significantly more than the 960 sessions in list mode.
- Efficiency: Average session length was shorter in visualization mode (172.3s vs. 197.4s), suggesting faster information retrieval and understanding.
- Meaningful Interaction: A striking 83.1% of clicks in the visualization mode were "content-related" (e.g., editing tags or resources), compared to only 57.6% in list mode. This suggests the graph encourages users to interact directly with data rather than just navigating the UI.
Conclusion and Future Work
GalViz demonstrates that visualizing the semantic context of bookmarks as a graph is more effective than traditional scrolling lists. By bridging personal and global discovery, the system encourages deeper interaction with data. While these preliminary results are promising, the authors acknowledge the need for more rigorous, task-based usability testing. Future work will involve deploying the system for crowdsourced evaluation to further quantify cognitive load and user satisfaction across specific information management tasks.
Summary updated