Suggestions from ndthuan

[ndthuan]

Firstly, thanks for your great toolkit. It's very accurate at tokenizing and tagging words.

When doing https://github.com/ndthuan/vi-word-segmenter I found some minor issues that I'd like to suggest here:

• models should be loaded from resources so that we don't need to care about folder structure. I set it as a resource in pom.xml instead of moving the whole models folder into src/main/resources because I don't want to create huge conflicts with the upstream repo.
• Maven build is split into 2 profiles:
  • aio-jar builds the all-in-one runnable jar with all the resources and dependencies. This is default and is the same as current behavior.
  • lib only builds a jar of only this lib and its own resources, without dependencies.
• the pre-built jar is removed from source repo, you could always upload it on releases section of this repo

[tienthanhdhcn]

Thanks for your suggestion. We will have a look to see if we can integrate it soon.