I think you could read broadly on sentiment analysis first. Recommendation in electronic commerce and products purchase could be a product of sentiment analysis of reviews.
Sarcasm Detection: How to detect statements like “Nice perfume. Must you marinate in it?”.
Double negative detection: How to detect “the coffee is not bad” as not a negative statement, and differentiate “Well, your parents are not too bad” which is much more negative.
Emotion Mining
Detecting questions (and discriminating them from non-questions. See 1)
They are trending as a lot of sentiment analysis research is prioritized by big companies, and you can clearly see how it may be beneficial to a social media or analytics/advertising company.