Am 05.03.2013 um 13:33 schrieb Steven Bethard <[email protected]>:
> On Mar 4, 2013, at 1:28 PM, "Miller, Timothy" > <[email protected]> wrote: >> The question of when to use uima descriptor parameters vs resources has come >> up offline recently. My understanding is that resources are valuable for >> things like large model files, where it can be loaded once by UIMA and then >> shared between pipelines. I happened to notice that in the dependency >> parser, which has an uncommonly large model file, the model filename is a >> parameter, and from what I can tell the model will be loaded separately for >> each pipeline. Is this something that should be fixed or is there some >> reasoning against using the resource mechanism in this instance or in >> general? > > It's definitely something that should be fixed. Typically, even with resources, the model is loaded separately for each pipeline. That is because UIMA does not support sharing pre-created objects. The model is shared only when multiple AEs or multiple instances of the same AE in a single pipeline (AAE) use the model - e.g. when using running a CPE. Then, UIMA will instantiate the model (as opposed to using a pre-existing instance) and pass it to the different AE instances. Cheers, -- Richard -- ------------------------------------------------------------------- Richard Eckart de Castilho Technical Lead Ubiquitous Knowledge Processing Lab (UKP-TUD) FB 20 Computer Science Department Technische Universität Darmstadt Hochschulstr. 10, D-64289 Darmstadt, Germany phone [+49] (0)6151 16-7477, fax -5455, room S2/02/B117 [email protected] www.ukp.tu-darmstadt.de Web Research at TU Darmstadt (WeRC) www.werc.tu-darmstadt.de -------------------------------------------------------------------
