As I am quite new for ICA world, we would like to start to work on Turkish language support for ICA.
I have found some links (one of them is this forum!) and started to look at Language Ware workbench while warming up.
I would like to hear your experiences for localization of ICA environment.
As you can understand I am not just talking about the UI (these are Turkish at the moment). We would like to add semantic analysis and etc.
I also heard that new version of ICA will be using System-T instead of LanguageWare ..
what are the differences?
Should we wait for the new version?
May be we can start on ICA 2.2env and move our work to new major version of ICA... but I am not sure if we can do this with such an architectural change.
bfoyle 060001WDQ360 Posts
Re: ICA Localization and new version2012-06-01T22:14:00ZThis is the accepted answer. This is the accepted answer.We have announced the new version now so I am free to talk about what's in the new release.
First off, ICA has always used a combination of UIMA annotators and SystemT rules. In this release it's no different but a couple of things that used to be SystemT have shifted to UIMA. On the other hand, some of the new stuff like Sentiment support is utilizing SystemT for now. So the overall balance is about the same.
For the most part, however, if you are adding a new language for support I would expect you to work with UIMA and ICA Studio (the new branding for what used to be LanguageWare Resource Workbench).
I will let other partners talk about their past experiences specifically but from my perspective, I've been pleased with how quickly they have been able to extend ICA with new language support.
Additionally, there are some new features in ICA Studio that are added specifically for supporting the addition of new languages by partners...so I would recommend starting with ICA 3.0 and ICA Studio rather than ICA 2.2.
All the documentation and enablement / training should be on the partner portal shortly.
SystemAdmin 110000D4XK197 Posts
Re: ICA Localization and new version2013-01-21T16:29:44ZThis is the accepted answer. This is the accepted answer.
- bfoyle 060001WDQ3
just came bacj from BigInsights Bootcamp and we did some Text Analytics Workshop with System T (AQL)
My question is how can I use this annotations directly within ICA (so add a tam file as custom annotator to ICA)
I guess this will not work in this way?
Any information or link to PPT(PDFs) to find System T (AQL) and ICA Integration?
bfoyle 060001WDQ360 Posts
Re: ICA Localization and new version2013-01-23T19:08:31ZThis is the accepted answer. This is the accepted answer.
- SystemAdmin 110000D4XK
Typically, I see people building UIMA annotators via ICA Studio and then deploying them in ICA and then, via the integration with BigInsights, analyzing their large content volumes.
What it sounds like you are trying to do is take a SystemT annotator and run it through ICA. I believe the thing that needs to be done there is create a UIMA wrapper for that AQL and then incorporate it into the UIMA pipeline of ICA.
I'll see if I can get one of the developers to confirm / comment.