Earn 36,000 ($360.00)
Fortune: Use AI for Content Moderation
Bounty Description
In Fortune, people can enter any text they want, this might include profanities. While the current Recording Oracle has some profanity filter, it is very weak. Part of this challenge is to implement a better profanity filter using AI and integrate it into Fortune.
- Set up Fortune locally
- Write a prediction script.
- Write a training script.
- Train and save first model (any text task, does not have to be content moderation).
Resources
Current Profanity Filter: https://github.com/humanprotocol/human-protocol/blob/f9f646952d8670b0476772f03e4be56fe1708dab/packages/examples/fortune/recording-oracle/src/services/recordingOracle.ts#L50
Text classification with Sklearn: https://scikit-learn.org/stable/auto_examples/text/plot_document_classification_20newsgroups.html
Hate-speech Datasets: https://hatespeechdata.com/
Text Classification Models with Transformers: https://huggingface.co/docs/transformers/tasks/sequence_classification
Free Cloud GPU compute to train neural models: https://colab.research.google.com/?utm_source=scs-index
Job Launcher : https://job-launcher-client-nine.vercel.app/
For further questions and discussion, please join our Hackathon discord channel: https://discord.gg/z7Ecwns3