Hi AI-ers,

I have this problem in mind that I hope people on this list can give some insights on. I want to know what is the state of the art on determining the sentiment of a piece of text.

For example, "I hate chicken" is pretty negative, but "I hate chicken less than pork" is not as negative on chicken as the first example, but still pretty negative. It is also very negative on pork.

To do a good job, just counting "positive" and "negative" words is not enough. Words can be negated for example. "It doesn't suck" is not negative even though "suck" is a negative word. It seems to me at least some basic understanding of semantics is necessary.

I notice there are some startups that claim to have algorithms that does the job, so I suppose it is not really out there. Can people point me to some textbooks or survey papers that lay out the pros/cons of various approaches? Are there open source code that deals with such problems?

Thanks in advance.
Manch





--
Please Note: If you hit "REPLY", your message will be sent to everyone on this mailing list ([address removed])
This message was sent by Manch ([address removed]) from Bay Area Artificial Intelligence Meetup Group.
To learn more about Manch, visit his/her member profile
To unsubscribe or to update your mailing list settings, click here

Meetup Support: [address removed]
632 Broadway, New York, NY 10012 USA