The output file in the directory contains the sorted expected data.

Description: In this project we will use Hadoop MapReduce to implement a very basic “Sentiment Analysis” using the review text in the Yelp Academic Dataset as training data.

The Scoring Algorithm: For each word that appears in the text of a review, the job will compute a score summarizing the "positivity" or "negativity" of that word. The score for a word is computed by applying the following rules: • Each occurrence in a 5-star review increases the word's score by 2. • Each occurrence in a 4-star review increases the word's score by 1. • 3-star reviews have no impact on word score. • Each occurrence in a 2-star review decreases the word's score by 1. • Each occurrence in a 1-star review decreases the word's score by 2.

Consider an input dataset containing only the following three reviews: {"stars": 4, "text": "Good food, great times."} {"stars": 5, "text": "Been many times. Will be back."} {"stars": 1, "text": "Rude staff. Sent my tacos back."}

The scores for each of the words are computed as follows: score("Good") = 1 score("food,") = 1 score("great") = 1 score("times.") = 1 + 2 = 3 score("Been") = 2 score("many") = 2 score("Will") = 2 score("be") = 2 score("back.") = 2 + -2 = 0 score("Rude") = -2 score("staff.") = -2 score("Sent") = -2 score("my") = -2 score("tacos") = -2

The following is one example of a correct final output for the above input: 3 times. 2 be 2 many 2 Will 2 Been 1 great 1 food, 1 Good 0 back. -2 tacos -2 Sent -2 staff. -2 my -2 Rude