About 2 weeks ago I changed my view. In a flash of realization, after reading a paper about analogy detection by Peter Turney, I realized that sentiment is just another relationship between two things located in the text. Furthermore, I realized that what I should really be doing is figuring out what the characteristics of this relationship are so that I might be able to differentiate it from a host of other relationships.
After seeing a quote in Matt's recent post about Opinion Mining Startups:
Jodange specializes in sentiment mining, deriving tuples of the form <holder, opinion, target>I was pleased to see that at least someone else holds my newfound world view, but I think that there are other types of tuples applicable to sentiment analysis.
No comments:
Post a Comment