wiki:Developer Meetings/2012-11-13


Attendees: Rob Hooft (NBIC), Erik van Mulligen (EMC/S&T), Jan Kors (EMC), Reinout van schouwen (LUMC), Bharat Singh (EMC), Leon Mei (NBIC), David van Enckevort (NBIC), Marco Roos (LUMC)

2012-11-13, 9:30-11:15, Erasmus MC

review of last sprint

  • If some configuration is wrong in, webapp.xml, the error message is just a dump of stack trace and is not very clear.
  • For EMC, multi-lingual support is also very important (for a EU project).
  • For LUMC, a different version of peregrine is used, in particular by Herman. In collaboration with EMC peregrine is and will be used for generating the resources for Anni and Anni web services. ToDo? (Reinout/Marco?): make sure LUMC uses latest version of Peregrine where possible, and gets involved in its development.
  • For S&T work, multi-lingual (e.g. Dutch) indexing seems not working. Dutch stemmer doesn't make any difference.
  • Acceptance test can be improved. e.g., we can try to index all terms in an ontology as a test case.
  • Luca Toldo performed a benchmarking test of various indexers on finding drug & side-effect relationships. Peregrine combining with JREX performs the best. The paper is going to be submitted.

WWW (What Went Well):

  • Erik, Kang are happy. No issue left now.

TALA (Take A Look At):

  • Not all tasks are done. We should make more realistic planning next time.

Other users:

  • Yunqing Chen (Astrazeneca). Leon contacted them again to check if they have Peregrine running or not.

User story

S&T is creating an expertise ontology (hierarchical with parent/child concepts). Erik successfully added new fields into ontology to store such relationships. Would be nice to be able to query these relationships given a known concept via Ontology API.

  • In ontology, split URL into base+ID to allow more efficient storage
  • use SKOS as the ontology format without start-up performance penalty.
  • create a hot-boot for fast ontology loading (e.g. dump the ontology from memory into disk and use it)

Marco wants to have support on concept URLs -> already available via db source/code in the ontology

Luca Toldo:

  • I want to easily install Peregrine and maintain it locally.
  • I want to easily install and use LVG, or other different normalizers. E.g., via a plug-in infrastructure.
  • To support mine relationships, Peregrine should use chunking information to improve the indexing accuracy. Or just passing the chunking information through.


  • Release 1.2 (Sprint till the next technical meeting in the end of January)
  • Release 1.3
  • user meeting the end of April

Parking lot

  • more 3rd party developers involved are welcome. But EMC should make sure the contributions going to the owner (EMC).
  • find more users here in NL.
  • create a PR section at peregrine wiki.
  • BBMRI is potential user group. Lifelines might be an interesting user as they want to improve their catalogue content (probably via Morris Swertz' group).
Last modified 10 years ago Last modified on Nov 15, 2012, 8:40:02 AM