cotohacall

Make it easy to use Japanese processing API, Cotoha, offered by NTT Group.

Description

cotohacall is a tool to use Cotoha API easily. What you need to use the API is installing this package and API account.

What's Cotoha?

"COTOHA is Japanese Native AI originated from Japanese processing technology that NTT Group have developed over 40 years."
by NTT communications.
More details

Demo

Features

Easy use.
You don't have to worry about how to send the request.
You can do Named Entity Extraction, Parsing, Reference Resolution, Keyword Extraction, Similarity Calculation, Sentence Type Classification, User Attribute Estimation (β), Filler Removal (β), Detect Misrecognition (β), Sentiment Analysis, and Summarization (About them).

Requirement

Python3 or more
Cotoha Developers Account (Get it from here, it's free!)

Usage

Install the package.

pip install cotoha_at_python

Import the libraries.

import os
from cotohacall.main import cotoha_call

Get the Cotoha's CLIENT_ID and CLIENT_SECRET from here, and set them as the environment variables.

os.environ['CLIENT_ID'] = 'Your ID'
os.environ['CLIENT_SECRET'] = 'Your secret'

Prepare the sentence/sentences.

sentence = 'えーっと、前線が太平洋上に停滞しています。えー、一方、高気圧が千島近海にあって、あーっと北日本から東日本をゆるやかに覆っています。'

# For Similarity Calculation
sentence_no_2 = '比較対象となる文。'

Select the type below.

Type	ne	parse	coref	keyword	simi	sent_type	user_at	filler	detect	senti	summary
Mean	Named Entity Extraction	Parsing	Reference Resolution	Keyword Extraction	Similarity Calculation	Sentence Type Classification	User Attribute Estimation (β)	Filler Removal (β)	Detect Misrecognition (β)	Sentiment Analysis	Summarization

Detail about these types from here.

Write the code.

cotoha_call('SELECTED TYPE', sentence)

# For Similarity Calculation
cotoha_call('simi', sentence, sentence_2 = sentence_no_2)

# For Summarization (x = the number of the sentences, default = 1)
cotoha_call('summary', sentence, sent_len = x)

Installation

pip install cotoha_at_python

Samples

Ran at the Google Colaboratory.

pip install cotoha_at_python

from cotohacall.main  import cotoha_call
import os

os.environ['CLIENT_ID'] = 'Your ID'
os.environ['CLIENT_SECRET'] = 'Ypur Secret'

1

Try keyword extraction.

sentence = '日本語の高精度な自然言語解析を実現するAPIサービス。NTTグループの40年以上の研究成果を活かした自然言語解析技術をCOTOHA APIでお手軽にご利用いただけます。'
cotoha_call('keyword', sentence)

# Return
"""
[{'form': '高精度', 'score': 20.0},
 {'form': '実現', 'score': 16.8278},
 {'form': 'お手軽', 'score': 10.8133},
 {'form': '研究成果', 'score': 10.0},
 {'form': 'cotoha api', 'score': 10.0}]
"""

2

Try all.

sentence_no_1 = '今日は早起きだね。えーっと、せっかくだし、美味しいと評判の喫茶店に行ってみよっか。'
sentence_no_2 = '朝食は友人に勧められたカフェで食べようか。'

for api_type in ["ne", "parse", "coref", "keyword", "simi", "sen_type", "user_at", "filler", "detect", "senti", "summary"]:
    print('API type : {}'.format(api_type))
    display(cotoha_call(api_type, sentence_no_1, sentence_2 = sentence_no_2))
    print('\n')

Return
"""
omit (too long to show)
"""

Link

Google colab上でCotohaを超手軽に使うためのコピペ用コード

Reference

Cotoha API
自然言語処理を簡単に扱えると噂のCOTOHA APIをPythonで使ってみた

Author

TOOOOOOMY

License

MIT

TOOOOOOMY/Cotoha_at_python