1000 Movie Reviews (Review + Attached rating + Sentiment polarity) for Reputation Generation

Published: 9 March 2019| Version 1 | DOI: 10.17632/38j8b6s2mx.1
Contributor:
Abdessamad Benlahbib

Description

We have created manually 10 datasets for 10 different movies, each one contains 100 reviews (comment + rating + sentiment polarity) randomly extracted and we made sure that the datasets are representatives based on IMDb users weighted average vote. For each movie, we have created 100 text files named from 1 to 100, each one contains an opinion towards the target movie expressed in natural language (English). Furthermore, we have created a text file named "rating.txt" that contains an array of numbers, each number represents the rating (numeric scale ranging from 1 to 10) attached to a specific opinion towards the target movie. Also we have created a text file named "polarity.txt" that contains an array of 0 and 1. 0 stands for a negative review and 1 stands for positive review. This is what the rating file looks like: rating = [ 9, 10, 3, ...................] Interpretation: The user that posted the review stored in file "1.txt" has given a 9/10 as rating towards the target movie. The user that posted the review stored in file "2.txt" has given a 10/10 as rating towards the target movie. The user that posted the review stored in file "3.txt" has given a 3/10 as rating towards the target movie. This is what the polarity file looks like: polarity = [1, 1, 0, ...................] Interpretation: The review stored in file "1.txt" holds positive opinion towards the target movie. The review stored in file "2.txt" holds positive opinion towards the target movie. The review stored in file "3.txt" holds negative opinion towards the target movie. We mention that we have annotated manually the sentiment polarity of each review.

Files

Categories

Natural Language Processing, Reputation, Sentiment Analysis

Licence