The Tarsqi Toolkit

The Tarsqi Toolkit (TTK) provides one-stop shopping for all your temporal needs (or, hopefully, at least some of your needs). It integrates extraction of events and time expressions with creation of temporal links, using a set of mostly independent modules, while ensuring consistency with a constraint propagation component. The glue that keeps all the modules together is the TimeML language.

TTK contains the following components:

  1. GUTime - extraction of time expressions
  2. Evita - event extraction
  3. Slinket - modal parsing
  4. S2T - temporal repercussions of modal relations
  5. Blinker - opportunistic pattern-based parsing of temporal relations, based on GutenLink
  6. Classifier - MaxEnt classifier trained on TimeBank
  7. Sputlink - constraint propagation (aka temporal closure)
  8. Link Merger - uses Sputlink to ensure consistency of all relations

The long-term vision with TTK is to provide a set of tools that can be adapted to new data and that would let the developer set up a processing chain to fit his or her particular needs. To that end, the toolkit will include tools to inspect and edit dictionaries and pattern sets as well as tools to train machine learning components on new data.

The toolkit is bundled with a GUI, the Tango annotation tool, and TBox, a compact visual representation language for TimeML graphs. See the screenshots page for some snapshots of the toolkit in action.

TTK version 1.0 is available, refer tot the download page for details.