TYK TagYourKnowledge

• TYK is
  • TYK is a vision.
  • It is a visualization akin to browsing a thesaurus, where the words are user–defined tags and their relationships are calculated on the tagging of objects (photos, bookmarks, documents, webpages…). The greater the number of objects using a combination of tags, the stronger the connections between the tags will be.
    Additionally, while browsing it is possible to access the related objects relative to the space one is exploring.
  • TYK allows you to perform real tags browsing on any kind of repository.
  • Real tags browsing means exploring a weighted graph where nodes are tags and the edges are the relationship between the tags. Two tags are connected with an edge when one or more objects are using the same two tags. The weight of each edge is the number of objects that are using both the endpoint tags of the edge.
  • Furthermore TYK allows you for a more effective browsing than just "moving" from one tag to another. In fact, you can also add tags while browsing to find tags connected to each of the tags you have selected and their related objects.
  • TOM (TaggedObjectsManager) is the Java implementation of the core logic.

• This site is
  • This site is a proof of concept.
  • This site is a living example of the overall concept based on a few flickr objects.
  • Once the core logic has been implemented, it was easy to load some flickr data into TOM.
  • Currently TOM stores only 100 objects. Specifically, it stores the last 100 photos with their related tags (about 150) of a specific flickr user. It will be quite easy to crowl much more objects from flickr archive.
  • This site is just one possible application of TOM.
  • The greater the number of objects managed by TOM, the more expressive your browsing will be.

• The next steps are
  • The main UI improvement I am working on (regardless of any layout issue) is to show tags in a logical ORDER based on their weight. The weight has two semantics:
    • the weight of a tag is based on the number of objects that are referencing it
    • the weight of a connected tag is based on the number of objects that are referencing each connected tag.
  • Enhance object discovery by adding some traditional tools like hierarchical structure, index search on creation date, and so on.
  • Solve some Object2Relational problems.
  • Obtain as much user feedback as possible.
  • Welcome your suggestions. You can contact me at vlorenz@users.sourceforge.net

ENJOY your tags browsing