This project is meant to perform sentiment analysis on text documents using two different approaches:
- Bag of words
- Machine Learning

As of today (12th of September 2018) only the bag of words approach is implemented.

This approach is based on lists of positive and negative words.

First, all the words in the documents are lemmatized (eg: "leaves" -> "leaf").
Then, it will compute a score for each word in the document:
- if the word is present in the list of positive words, its score is 1 (eg: "good")
- if the word is present in the list of negative words, its score is -1 (eg: "bad")
- otherwise it is 0

Then the score of the word is updated to take into account the context: - if there is a diminisher in the previous words it will multiply the score by 0.5 ("This is a partially nad"; "bad" will have a score of: -1 * 0.5 = -0.5 )
- if there is an intensifier in the previous words it will multiply the score by 2 (eg: "It is very bad"; "bad" will have a score of: -1 * 2 = -2)
- if there is a negation in the previous words it will multiply the score by -1 (eg: "It is not bad"; "bad" will have a score of: -1 * -1 = 1)

The user has the choice of different list of words:
- Loughan and McDonald: (https://sraf.nd.edu/textual-analysis/resources/#LM%20Sentiment%20Word%20Lists)
- The Harvard Inquirer: (http://www.wjh.harvard.edu/~inquirer/) (to be implemented)

Example:

get_sentiment_doc("He is a good developper, I like the way he adds insightful comments in his code")
-> 2

To be implemented.