README

Sentiment of Reviews 


Algorithm used:
Set sentiment score is 0;
Scan each word A in a review R
If A is a positive word, sentiment score = sentiment score+1;
If A is a negative word, sentiment score = sentiment score -1;
After scanning all words, if sentiment score ≥ 0, the review is positive; if the sentiment < 0, the sentiment score is negative.


Datasets Description:
UnlabelReview.json: It contains review id and review contents, the review is raw data from customers, organized in paragraph.
Example: { id: “5201_1”, review: “I like the movie, it’s fantastic although I hate  the actor  James in the movie”}

UnlabelReviewAfterSpliting.json: It contains review id and review contents, each review has been split to word and word count. It is processed from UnlabelReview.json. We have omitted part of useless words, like stop words, like “is”,” are”, “do” and so on).
Example: { id: “5201_1”, review: [{word: “like”,count:2}, {word: “movie”,count:2}, {word: “actor”: 1}, {word: “fantastic”, count:1}, {word: “hate”, count: 1} ]}


Additional File:
positive words.txt: a list of positive words you may use when categorizing the review
<<<<<<< HEAD
negative words.txt: a list of negative words you may use when categorizing the review
=======
negative words.txt: a list of negative words you may use when categorizing the review
>>>>>>> 8f8bd9c6093b57384b46c1691a880e60253077dc