VADER: Rule Based Approach to Sentiment Analysis

# VADER: for Valence Aware Dictionary for sEntiment Reasoning.

# VADER is a rule-based approach towards doing sentiment analysis.

# As of NLTK version 3.6 [20210721], it uses VADER for sentiment analysis.

VADER: In Code






Try following code snippets in Python CLI
>>> from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer
>>> sa = SentimentIntensityAnalyzer()
>>> sa.lexicon

{ ...
':(': -1.9,
':)': 2.0,
...
'pls': 0.3,
'plz': 0.3,
...
'great': 3.1,
... }

>>> [(tok, score) for tok, score in sa.lexicon.items() if " " in tok]

[("( '}{' )", 1.6),
("can't stand", -2.0),
('fed up', -1.8),
('screwed up', -1.5)]

>>> sa.polarity_scores(text="Python is very readable and it's great for NLP.")
{'compound': 0.6249, 'neg': 0.0, 'neu': 0.661, 'pos': 0.339}

>>> sa.polarity_scores(text="Python is not a bad choice for most applications.")
{'compound': 0.431, 'neg': 0.0, 'neu': 0.711, 'pos': 0.289}

>>> corpus = ["Absolutely perfect! Love it! :-) :-) :-)",
... "Horrible! Completely useless. :(",
... "It was OK. Some good and some bad things."]

>>> for doc in corpus:
... scores = sa.polarity_scores(doc)
... print('{:+}: {}'.format(scores['compound'], doc))

+0.9428: Absolutely perfect! Love it! :-) :-) :-)
-0.8768: Horrible! Completely useless. :(
+0.3254: It was OK. Some good and some bad things.

DRAWBACK OF VADER

The drawback of VADER is that it doesn’t look at all the words in a document, only about 7,500. 

The questions that remained unanswered by VADER are:

What if you want all the words to help add to the sentiment score? 

And what if you don’t want to have to code your own understanding of the words in a dictionary of thousands of words or add a bunch of custom words to the dictionary in SentimentIntensityAnalyzer.lexicon? 

The rule-based approach might be impossible if you don’t understand the language, because you wouldn’t know what scores to put in the dictionary (lexicon)!

References

# “VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text” by Hutto and Gilbert
(http://comp.social.gatech.edu/papers/icwsm14.vader.hutto.pdf)

# You can find more detailed installation instructions with the package source code on github (https://
github.com/cjhutto/vaderSentiment).

Labels: Technology,Natural Language Processing,Python,