Extract features from issue's body text

mozilla / webcompat-ml

Webcompat machine learning models

Mozilla Public License 2.0

4 stars 3 forks source link

Extract features from issue's body text #3

Open johngian opened 5 years ago

johngian commented 5 years ago

Currently we tokenize the issue body as a single free text input. What we could try to see if it improves the metrics is parse the distinct parts that consist a webcompat report and extract features. Potential features:

URL
Browser
Category
Description
Steps to reproduce
Tested another browser