Goodreads Project

Data were downloaded from Mengting Wan .

The datasets were collected in late 2017 from goodreads. Details of the datasets are described in the dataset website

Goals

1. Explore high dimensional user data from Goodreads.
2. Learn new-to-me approaches in NLP and implicit collaborative filtering/recommender systems.

Main Notebooks

Used for my analyses so far:

• makeDBs_complete.ipynb and makeDBs_romance.ipynb: Create SQLite database and tables for data organization/storage.
• statistics.ipynb: EDA notebook to compute basic statistics of data and understand rating/review trends across genres.
• reviews_sentiment.ipynb: Sentiment analysis on >3.4 million romance book reviews.
• romance_recommender.ipynb: Implicit collaborative filtering on >16 million romance book ratings.

Dataset Citations

• Mengting Wan, Julian McAuley, "Item Recommendation on Monotonic Behavior Chains", in RecSys'18. [bibtex]
• Mengting Wan, Rishabh Misra, Ndapa Nakashole, Julian McAuley, "Fine-Grained Spoiler Detection from Large-Scale Review Corpora", in ACL'19. [bibtex]