We have developed and implemented a pipeline that can take peer review text from MWrite courses, and perform the text classification with PeerBERT algorithm as a deep learning architecture.
This is a multi-faceted approach that uses multiple tools:
- OpenShift & Jenkins for daily cron job to retrieve course info and peer review results
- Google Cloud Services
- BigQuery for flexible data storage
- Vertex AI to build and deploy machine learning solutions
- PyTorch and the RoBERTa transformer architecture to perform powerful natural language processing and inferences
- Tableau to build interactive and intuitive dashboards that delivers focused insights for instructors and students
Insights into Peer Reviews: A Text Analysis Pipeline Built on Google Cloud - Slide Presentation