> Sentiment analysis has no way of defining sarcasm or other natural language terms. Of course it can. If humans are capable of detecting a given inflection, c

> Sentiment analysis has no way of defining sarcasm or other natural language terms. Of course it can. If humans are capable of detecting a given inflection, computers absolutely can as well (given enough data). Any sentiment analysis algorithm which classifies "that new machine learning algorithm is so sick" as negative is not worth an ounce of consideration. Compared to other problems, that is absolutely trivial to classify, especially since you're typically training off data sets which already include such vernacular.