Sentiment analysis can be defined as the process of identifying, categorizing opinions of speakers or viewers in terms of text to determine the attitude toward a particular post, especially on social media. It is the parameter for analyzing if the consumers, in general, dislike or like something.
Sentiment analysis is tough to analyze because of the complicated sentence structure in English. If we consider the following sentences:
“I like that SUV (Sports utility vehicle)”
“That looks like SUV”
If we analyze the sentence the first one would be positive and the other would be neutral. But in most the cases in the systems, it would be marked incorrectly because a word like is expressing positivity in the first sentence but not in the second.
Let’s take another example.
“I’m craving Domino’s so bad”.
The most system will misinterpret seeing word bad. Hence, contextual understanding is imp. To reach human level accuracy.
There are 3 factors in effective sentiment analysis:
- Accuracy: It is a measure of how many times sentiments rating was correct.
- Recall: It is a measure of how many sentiments in a document were actually rated sentimental.
- F1 score: It is a combination of Precision and recall. A perfect score is in between 0.0-1.0.
Sentiment analysis is a part of social media and social listening which utilizes natural language processing system. Accuracy and recall are the critical elements to get the perfect results.