Similarly, it’s difficult to train systems to identify irony and sarcasm, and this can lead to incorrectly labeled sentiments. Algorithms have trouble with pronoun resolution,
