|Posted by kyle on June 3, 2006 4:16 PM||bookmark / share:|
Sources and Background
Wordnet is a project at Princeton University's Cognitive Science Laboratory that provides a map of the English language. Visual Wordnet currently presents only a narrow view of Wordnet's resources.
The idea of using Force Directed and Hyperbolic graphs to display and navigate Wordnet is nothing new. ThinkMap's Visual Thesaurus is just awesome and it does a much better job scaling to support large graphs. So why write this implementation? ThinkMap's solution is implemented in Java and available as an applet. It's just not appropriate for some applications. Also ... it ain't free.
Known Issues (echoed from my initial post on Force Directed Graphs)
- Performance: As currently implemented, the graphing engine consumes a lot of processing power, even after the graph has settled. An entropy-based throttle would help.
- Scaling: The layout algorithm has a complexity of On2 :(. It just doesn't scale very well.
- Graphs for words with a lot of synonyms, like "work" and "go", contain a lot of nodes. With the scaling problems, this drives performance to a near-stall.
Update on Licensing:
This work is licensed under a Creative Commons Attribution 2.5 License.
As promised, I'm providing an open source license for the files referenced here. I've chosen a Creative Commons Attribution License, which means you are free to distribute the files and create derivative works as long as you include the license and attribution information contained in each file. I will follow up with a post describing each of these files to help you integrate with your project.