Difficulty In Sentiment Analysis & Opinion Mining

When I want to pick a film, I always go to DOUBAN for reference. Firstly, I will focus on which style of film would attract me better through finding the key words. Then I will compare the scores of films. Finally I will read some comments of films having a high score.

Recently more and more individuals prefer shopping online. I think many people also visit similar platform to gain some information before making their choice, just like my picking films. Our opinions and behaviors are influenced by others'. Therefore, sentiment analysis is increasingly valuable to study.

Through the fourth course, we have learnt relative knowledge in the field of mining opinion. I want to take DOUBAN mentioned above as a example. How to deal with the enormous comments? In an intuitive sense, adjectives represent the sentiment. In addition, we regard unigrams as the feathers, because they are more simple and give satisfactory results. And we have to dispose the negative sentences. We add NOT_ to every word between negation and following punctuation. "don't like this movie" is changed to "don't NOT_like NOT_this NOT_movie". Then we can use an appropriate classifier to finish the classification：

However, sentiment analysis is not that easy. When browsing the web pages, I found so many euphemism or sarcastic comment. "If you are reading this because it is your darling fragrance, please wear it at home exclusively, and tape the windows shut." The writer runs the gamut of emotion from A to B. "This film should be brilliant.  It sounds like a great plot, the actors are first grade, and the supporting cast is good as well, and Stallone is attempting to deliver a good performance. However, it can’t hold up." I expecte a communication with classmates, who have some ideas of solving this problem.