[jira] [Commented] (OPENNLP-496) DictionaryNameFinder only deals with a single physical dictionary, thus a single type of entities

["Jim Piliouras (Commented) (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/OPENNLP-496?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13253395#comment-13253395 ] 

Jim Piliouras commented on OPENNLP-496:
---------------------------------------

They can also be used with the AggregateNameFinder class with no problems. It seems to me that with these improvements the namefind package is way ahead of the game from other NER APIs. People can do whatever they want more or less. Use multiple models, multiple dictionaries, do custom feature generation, evaluate the combined results etc. Everything works just fine in my private build (experiment-OPENNLP.jar)...I really hope we can have this sometime at least on trunk...
                
> DictionaryNameFinder only deals with a single physical dictionary, thus a single type of entities
> -------------------------------------------------------------------------------------------------
>
>                 Key: OPENNLP-496
>                 URL: https://issues.apache.org/jira/browse/OPENNLP-496
>             Project: OpenNLP
>          Issue Type: Improvement
>          Components: Name Finder
>    Affects Versions: tools-1.5.3
>         Environment: Ubuntu x64 Java 7 update 3
>            Reporter: Jim Piliouras
>              Labels: patch
>             Fix For: tools-1.5.3
>
>         Attachments: OPENNLP-496-1.patch, OPENNLP-496-2.patch
>
>   Original Estimate: 2h
>  Remaining Estimate: 2h
>
> The Dictionary itself should store the type of entities it includes - NOT the finder. All the code (global field, extra constructor,getter method) should go in the Dictionary class. This will allow passing several actual xml dictionaries to the finder and the fidner can assign the correct type depending on which dictionary gave the prediction. This is extremely simple to do , in fact all the code needed is almost identical to yesterday's patch (OPENNLP-495). Since opennlp-495 has already been commited i will provide new patches for the latest head revision. One of the patches will revert the DictioanryNameFinder to its original state without breaking anything.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira