Deeq NLP Morph Analyzer / POS Tagger

[rsn0315]

deeq-nlp v1.2.1 : https://github.com/deeq-ai
Deeq-nlp client is a natural language processing engine developed by Baikal AI(https://www.baikal.ai). Our engine is :

• Designed by the professor of Korean language and literature department at Seoul National University, and further developed and commercialized by Baikal AI.
• A deep learning engine that understands how agglutinative language like Korean works
• A deep learning engine that satisfies the new morpheme tagging standard proposed by the National Institute of Korean Language in 2019
• Based on go lang, and thus works very fast
• 99.7% morph tagging accuracy
• 99.6% word segmentation accuracy

We've followed all the contribution requirements, and checked that our modification passed all the required tests.
(We also wanted to check whether the code works not only in python3 environment but also in python 2, but we noticed that the package Jpype was only supported in Python3 environment.)

[Thrameos]

JPype dropped support for Python 2 at the start of the year. Older versions still support Python 2 but speed penalty for Python 2 support is a factor of 10-500 depending on the call. We had to move to supporting only one object model as supporting two different models was keeping a huge pile of code in Python rather than C.