Semantic role labeling?

[xt.k...@gmail.com]

First of all, great work on this library. It really makes processing Korean a lot easier than it was before and doesn't force me to read through broken half-encoded Korean deep in the bowels of the Internet to figure out what I need. ;)

If I wanted to add semantic role labeling, where do you think would be a good starting point? I have researched recurrent neural networks and also other linear-based methods, although I don't know what would be best to use. I wanted to hear your thoughts. I was thinking of first parsing through each sentence (Kkma().sentences()) and then parsing the POS tags of each sentence and putting each sentence's words and matching POS tags through some sort of neural network. Korean, being morphologically complex, means that solutions for English are probably quite ineffective. Therefore I wanted to begin the labeling from a combination of the verb stems, markers, and POS tags. I have an intermediate knowledge of Korean and also its postgrammatical markers. If I come up with anything useful I'd contribute it, but I wondered if you had any advice for me.

Thank you,

Andrew