ClassifierForJ - mehdimo/janett GitHub Wiki

Classifier4J is a text classification library.

Translating Classifier version 0.6 is the reason for starting Janett development. It includes vector space algorithm not included in version 0.5, which is available as NClassifier.

Classifier has been our good friend in Janett development. Janett can currently translate Classifier in both VirtualizationMode and NativeMode, resulting code will be built instantly, test-cases will pass instantly in VirtualizationMode and with 3 manual changes in NativeMode all test cases will also pass.

Note: We don't have any Classifier specific tweaks in Janett, Therefore, we excluded some parts of Classifier while translating them prematurely was an easy task:

• JDBC libraries are not mapped so JDBC support was excluded.
• Because of the reason explained in ClassifierResourceClass, CustomizableStopWordProvider and its test-case are excluded.
• AbstractWordsDataSourceSupport (base class for test-cases) should be abstract even in Java

Currently, Classifier translation, building and passing unit-tests is Janett regression-testing tool.

You can translate Classifier yourself in these steps:

• Checkout classifier source from CVS or extract latest distribution to a folder.

• Make AbstractWordsDataSourceSupport in core4Jabstract

• Copy translate.xml and DotNet.patch files, attached to this page, to this folder

• Call janett.exe passing this folder. Repeat for IKVM mode. For example:

  janett E:\Workspaces\Classifier janett E:\Workspaces\Classifier -m IKVM

translate code will be at E:-DotNet and E:-!IKVM. See magic of Janett there.