mapmeld / DeepClapback

Queries Reddit comments and trains a model on successful 'clapback' comments

New repo: https://github.com/Georeactor/reddit-one-ups

Datasets: https://huggingface.co/datasets?search=georeactor+one_ups

DeepClapback

A Reddit comment dataset which searches for 'clapbacks' (comments which are scored higher than the original comments) and set up CSVs for Google AutoML to build a classification model.

Article: "Can DeepClapback learn to lol?" https://medium.com/@mapmeld/can-deepclapback-learn-when-to-lol-e4a2092a8f2c

Running with cloud services

• sudo apt-get install python3-pip postgresql-client-10 postgresql-client-common transmission-cli
• cd /mnt/DISK
• bzip -d /mnt/DISK/reddit_data/year/RC_YEAR-MONTH.bz2
• python3 reddit_json_converter.py
• python3 reddit_comment_sql.py

Torrent Warning

Downloading Reddit data as a torrent from AcademicTorrents, may be flagged by your work, school, ISP, VPN or other watchers. Consider your connection, tread carefully.

Content Warning

Comments and responses in the torrent, 'clapback', and NOMEME datasets, all include NSFW language and links!

License

Reddit comments are properties of Reddit and comment owners using their Terms of Service

Code is public domain