Maoyan Movie Review Dataset(M-Dataset)
Introduction
-
coming from "GEK: A Spam Movie Review Detection Model Based on Graph Convolutional Neural Network Embedding of Movie Knowledge"
-
M-Dataset contains spam movie review records and non-spam movie review records collected from Maoyan(one of the largest platforms for movie tickets and movie reviews in China).
-
The user may not violate the privacy protection policy of Maoyan.
-
M-Dataset Statistics
Name #Reviews #Spam #Non-spam #Movie Genre Movie Released Year Spam Candidates 734,130 -- -- 2352 40 2017-2021 M-Dataset 65,696 20092 45604 457 40 2017-2021
Labeling Strategy
-
The content of the review has absolutely nothing to do with the movie being reviewed;
-
If the evaluation and rating do not match, the content of the review is rated as a high score for criticism, and the content of the review as a low score for praise;
-
The review description is exaggerated and overly praised, full of empty adjectives and pure warm praise, without any shortcomings;
-
The content of the reviews is unified, and there are a lot of similar reviews in other movie reviews;
-
The review that show too high or too low ratings simply because they like or hate a certain movie star or character
We consider spam reviews as long as they meet one or more of the above criteria.
Details
-
The detailed information fields of a review include:
it will be updated soon...
-
The detailed information fields of a movie introduction include:
it will be updated soon...
-
The detailed information fields of a long comment include:
it will be updated soon...