Temporal

This page will host temporal information processing resources developed by Naushad UzZaman. I also host another webpage for general Temporal Information Processing related resources [here]. 

 
Temporal Evaluation with Temporal QA
Will be available soon...

Temporal Evaluation Toolkit [code]

This toolkit is used to evaluate the TempEval-3 participants. It evaluates the extraction of temporal entities (events, temporal expressions) using entity-based evaluation. For temporal relations, it evaluates systems that extract temporal information from text. It uses temporal closure to reward relations that are equivalent but distinct. This metric measures the overall performance of systems with a single score, making comparison between different systems straightforward. The fast, scalable temporal closure implementation is based on Timegraph (Miller and Schubert, 1990). 

The toolkit is described at our ACL 2011 short paper and the TempEval-3 task paper.

Temporal Annotation Merger [code]
A tool for merging different TimeML annotations of the same text. This uses a set of heuristics and temporal reasoning to get a high recall, high precision or balanced merged annotation that only contains consistent relations

Temporal Expression Normalizer [code
This program takes a temporal expression and returns the normalized type and value according to TimeML scheme. This is a regular expression based naive program, but this normalizer had the second best performance in TempEval 2010 (temporal evaluation competition). The reason for sharing this program is to help others start from this basic program and extend it to get better performance. It handles basic temporal expressions decently, so most of the people don't need to modify it. Any modification will be highly appreciated. Give me your updated copy, I will add your modification with credit. 


TRIOS-TimeBank corpus and TRIPS Ontology [link]
TRIOS-TimeBank corpus is an extension to TimeBank corpus with suggestions for missed events and temporal expressions, new event feature (ontology type) and extra SLINKs and new RLINK. Details can be found in the paper. TRIPS ontology has the mapping from ontology type to WordNet.