Dedicated to my AI teacher, Kavita Kelkar ma'am who introduced me to the field that I love so much now.
'Sentiment Analysis' is a branch of AI, linguistics particularly, which deals with understanding the 'opinion' behind words. It is for this that it seems to have immense potential in business applications.
Sentiment analysis deals with processing text so as to understand what the text says about a particular subject in terms of positive or negative content. So, sentiment analysis would read the sentence 'India beat Pakistan in the cricket match' and present an output that infers that the sentence says something positive for India and something negative for Pakistan.
A first thought is identifying words that say 'good' and 'bad' things. For example, the words 'fabulous', 'marvellous', 'excellent', 'amazing' denote positive opinion while 'dull', 'disappointing' denote negative opinion.
Sentiment analysis, however, goes deeper than words - it enters the world of cognition, the way we think what we think. Sentiments are not conveyed by the presence of mere words .
Consider this very sarcastic statement : "It is amazing to see how the camera makes sure all the photos you take with it are overexposed to light." Now this statement has no word that denotes a heavy negative sentiment. But still the overall opinion is negative.
With so many blogs and review websites coming up, the textual data available is already in abundance and with techniques like these, one would only be able to retrieve 'maximum information' from the data. Sentiment Analysis, thus, turns out to be a very amazing task that will help business houses, product vendors, filmmakers know what people are thinking and saying about their product.
Like, you know, Sentiment Analysis will be able to predict that this blogpost is saying all good things about it! :-)
Sentiment analysis deals with processing text so as to understand what the text says about a particular subject in terms of positive or negative content. So, sentiment analysis would read the sentence 'India beat Pakistan in the cricket match' and present an output that infers that the sentence says something positive for India and something negative for Pakistan.
A first thought is identifying words that say 'good' and 'bad' things. For example, the words 'fabulous', 'marvellous', 'excellent', 'amazing' denote positive opinion while 'dull', 'disappointing' denote negative opinion.
Sentiment analysis, however, goes deeper than words - it enters the world of cognition, the way we think what we think. Sentiments are not conveyed by the presence of mere words .
Consider this very sarcastic statement : "It is amazing to see how the camera makes sure all the photos you take with it are overexposed to light." Now this statement has no word that denotes a heavy negative sentiment. But still the overall opinion is negative.
With so many blogs and review websites coming up, the textual data available is already in abundance and with techniques like these, one would only be able to retrieve 'maximum information' from the data. Sentiment Analysis, thus, turns out to be a very amazing task that will help business houses, product vendors, filmmakers know what people are thinking and saying about their product.
Like, you know, Sentiment Analysis will be able to predict that this blogpost is saying all good things about it! :-)
thank you for dedicating this article!
ReplyDeleteI wish you do great research in AI!!!
Good one :)
ReplyDelete