This is Novel and Lightnovel rating dataset from MyAnimelist. I use Jikan API v4 to retrieve novel data, user data, and reviews. If you want to Jikan API as well, check out Jikan API v4 Documentation. Note that it limit the API call to 3 per second, and 60 API call per minutes.

This Dataset contains five csv files. One for novel details (novels.csv), one for users details (users.csv), and three for interactions between users and novels (interaction_ori.csv, interaction_equal.csv, and interaction_all.csv). The three interaction files are for searching the optimal way to build NN-based recommender system.

• The original is pure from API scrapping to mangaUserUpdate, where status is 'Add To Read List', 'Reading' or 'Complete' is considered as positive feedback and the others are negative feedback. This version has positive feedback 92% of the entire data
• The most biggest file, interaction_all.csv include all novel. First, we get the 'unthouched' novel, and then we pair those novel with every users and mark it as negative feedback. The positive feedback percentage of this version is 0.13%
• The Equal is a subset from interaction_all where the number of positive and negative feedback are equal (50% of the entire data).

Note: Please Ignore The concatenated.csv, as its file is not updated along with other files, and this file will be removed in the next version