FISEsecondHackathonReport

From IKS Project
Jump to: navigation, search

The second FISE hackathon took place on May 25-27th, 2011, in Paderborn.

Progress

Not many commits, but many "almost finished" new things that should be committed in the next few days.

Great team building and learning experience for new FISE developers, all participants are now able to create and debug modules for FISE.

Structure of enhancements defined and documented, http://wiki.iks-project.eu/index.php/EnhancementStructure (Rupert and team).

EnhancementEngines modified to use this structure (Rupert).

Simplified search UI implemented, not tested yet as standalone launcher does not have a persistent store or SPARQL endpoint yet (Andrea, Alessandro).

Language detection EnhancementEngine implemented and tested (AFAIK), ready to be committed (Joerg).

Adapter pattern discussed for more flexible use of ContentItem data, see http://code.google.com/p/iks-project/issues/detail?id=5 (Bertrand)

Prototype UI for a content creator user, FISE annotations are highlighted in text and user can resolve ambiguities using graphical contextual menu (Sebastian)

Fise JCR store mostly implemented (Michael).

Improved standalone launcher (launchers/sling), creates a usable runnable jar, need to add a persistent store an SPARQL endpoint to that setup, see http://code.google.com/p/iks-project/issues/detail?id=7 (Bertrand)

Problems

Working with snapshots is a mess, need to see if Clerezza could release the modules that we use, on agree on a common tag or revision number to build those modules ourselves, so that we all get the exact same version.

Missing infrastructure: continuous integration server, maven repository maybe.

Perspectives

Finalize the runnable jar soon, including RESTful API, in-memory triple store, SPARQL endpoint, simple search UI, EnhancementEngines and OSGi console and front page with explanations and links.

Create a demo based on that for the IKS early adopters workshop.

Define how engines can collaborate between themselves, for example run the text extraction engine first, then annotators, etc.

Text extraction engine based on Aperture (Joerg)

New API for user-provided annotations, needed for Sebastian's client where users disambiguate annotations.

Look at OpenCalais, Evri, Freebase, UIMA and check how those services could be integrated in FISE.