I am using a software tool SentiWordNet that can map an English word to a pair of numbers showing how positive and negative the word is. The pair of numbers p and n satisfies the 3 conditions.
- p ≥ 0
- n ≥ 0
- p + n ≤ 1
One can process each word of, say, a movie review and then find out the overall positiveness/negativeness.
I was thinking about whether or not I can find out the similarity of 2 movie reviews to compare the 2 sets of (p, n) points.
Are there any similarity metrics to compare between 2 sets of 2D points.
Any comments are welcome!
Thanks, Patrick