Sentiment analysis

One the most widespread applications of NLP is sentiment analysis of short texts (tweets, posts, comments, reviews, and so on). From a marketing viewpoint, it's very important to understand the semantics of these pieces of information in terms of the sentiment expressed. As you can understand, this task can be very easy when the comment is precise and contains only a set of positive/negative words, but it becomes more complex when in the same sentence there are different propositions that can conflict with each other. For example, I loved that hotel. It was a wonderful experience is clearly a positive comment, while The hotel is good, however, the restaurant was bad and, even if the waiters were kind, I had to fight with a receptionist to have another pillow. In this case, the situation is more difficult to manage, because there are both positive and negative elements, resulting in a neutral review. For this reason, many applications aren't based on a binary decision but...