How does sentiment analysis use natural language processing to identify emotions and opinions in text? What role do machine learning models, text preprocessing, and language context play in analyzing customer feedback? What challenges can occur when processing sarcasm, mixed emotions, and multilingual content?