[backend] Recommendation feature

[hadooboo]

1. Textblob : https://textblob.readthedocs.io/en/dev/

2. Spacy : https://spacy.io/

I'm thinking of a way to tagging before implementing recommendation. @ByungHeeCha has found 2 and I've found 1. I think of using 1 to analyze sentiment in sentence unit(https://textblob.readthedocs.io/en/dev/quickstart.html#tokenization) and using 2 to filter adjectives.(https://spacy.io/usage/linguistic-features#pos-tagging) In this way, we can find adjectives in comment and feelings on them.
We can just show adjectives or nouns corresponding that adjectives.

How about?

[ByungHeeCha]

3. Stanford Core NLP: https://nlp.stanford.edu/sentiment/index.html
   This is pretty great too!

[hadooboo]

I chose 1 and 2 as they are python codes! However 3 seems to use java.. isn't it?

[ByungHeeCha]

#!pip install pycorenlp
from pycorenlp import StanfordCoreNLP

They support python too!

[ByungHeeCha]

Let's make a simple model using these 3 NLP api and find the best one!