NLTK: Sentiment Strength Detection in Bahasa Indonesia

SentiStrengthID

Sentiment Strength Detection in Bahasa Indonesia. This is unsupervised version of SentiStrength (http://sentistrength.wlv.ac.uk/) in Bahasa Indonesia. Core Feature:

• Sentiment Lookup
• Negation Word Lookup
• Booster Word Lookup
• Emoticon Lookup
• Idiom Lookup
• Question Word Lookup
• Slang Word Lookup
• Spelling Correction (optional) using Pater Norvig (http://norvig.com/spell-correct.html)
• Negative emotion ignored in question
• Exclamation marks count as +2
• Repeated Punctuation boosts sentiment

Ignored Rule:

   repeated letters more than 2 boosts sentiment score. This rule do not applied due to my own pre-processing rule which removing word's extra character
   score +2, -2 in word "miss". Do not apply in Bahasa Indonesia.

Warning!

This is work in progress. Experimental for my Master Thesis

Source Code Asli

import pymongo, requests, json, urllib, re, collections
from pymongo import MongoClient

client = MongoClient('mongodb://localhost:27017/')
#mongodb database and collection
db = client.dataset_agnezmo
dbPre = db.preprocessing

def main():
    ss = sentiStrength()
    sc = spellCheck()
    tweets = dbPre.find().skip(0).limit(100)
    for t in tweets:
        print ss.main(t['text'])
    print "====================="        
    print ss.getSentimenScore()
    
main()

Test Kalimat

import requests, json, urllib, re, collections

def main():
    ss = sentiStrength()
    sc = spellCheck()
    t = 'film godzilla amat sangat menarik sekali'
    print ss.main(t)
    print "====================="        
    print ss.getSentimenScore()
    
main()

Hasilnya:

film godzilla amat sangat menarik [3] sekali [score:-1,3][result: +positive]
=====================
[Positive:1] [Negative:0] [Neutral:0]

Ubah Source Code

import argparse

def parse_args():
    parser = argparse.ArgumentParser()
    parser.add_argument('-i', '--infile', default=, help='input filename')
    return parser.parse_args() 

def main():
    args = parse_args()
    infile = args.infile

    filename = open(infile,'r')
    fcontent=filename.readlines()
    filename.close()

    ss = sentiStrength()
    sc = spellCheck()
    for t in fcontent:
        print ss.main(t)
    print "====================="        
    print ss.getSentimenScore()
    
main()

Referensi

• https://github.com/masdevid/SentiStrengthID