Chrome extension and backend service using the Prediction API to predict the assignee of an issue based on the issue title/text/labels
Repos would have to be explicitly enabled for this, though possibly a self-serve process on demand, since the API requires you to gather training data and train a model, and predictions won't work until the model is trained.
Chrome extension and backend service using the Prediction API to predict the assignee of an issue based on the issue title/text/labels
Repos would have to be explicitly enabled for this, though possibly a self-serve process on demand, since the API requires you to gather training data and train a model, and predictions won't work until the model is trained.
Training data would be in the form of:
(the first column trains the "output" to predict)
Prediction requests would take the title/body/labels and predict (with confidences) the likelihood for each potential assignee.
The backend could poll for new issues and assignment changes to update the model in ~realtime.
golang/goand there weren't enough to get high-confidence predictions yet