The HPI TimeML Corpus

Update (Feb 9th, 2012)

We have temporarily taken this corpus offline because of some exciting new developments. While we can't announce anything publicly right now, we expect to be able to soon. We also expect the data to be available again in the near future. In the meanwhile, check out our paper, and, if you have questions about the corpus, please do get in touch.
-- LG & NB
* * *

Quick Instructions

  1. Download
  2. Unzip & Expand
  3. Read the README file
  4. Install


The HPI TimeML Corpus is a small corpus of TimeML-annotated "History of Present Illness" sections of 44 clinical reports (discharge summaries). Details about the corpus can be found in the following paper:

L. Galescu, N. Blaylock. A Corpus of Clinical Narratives Annotated with Temporal Information. Proceeedings of the 2nd ACM SIGHIT International Health Informatics Symposium (IHI 2012). January 28-30, 2012, Miami, Florida, USA. [pdf]
We ask that this paper be cited in all references to the HPI TimeML Corpus.

Included in the download is not the actual corpus, but a set of standoff annotations and scripts to (re-)generate the HPI TimeML Corpus itself. The corpus cannot be generated unless you already have the original i2b2/VA 2010 NLP Challenge Data Set, from which our corpus is derived. Information on how to acquire the i2b2 2010 Data Set can be found on the webpage for the 2010 NLP Shared Task.


Lucian Galescu
Nate Blaylock