search

ProjectSidewalk / sidewalk-quality-analysis

An analysis of Project Sidewalk user quality based on interaction logs
5 stars 3 forks source link

Analyzing User Validation Accuracy #48

Closed daotyl000 closed 4 years ago

daotyl000 commented 5 years ago

Galen proposed looking into the accuracy of users doing validations by testing for group conciseness on validations. For examples, if there are 3 users who agree with a validation while 1 disagrees, the validation is probably correct. User's validations importance could be scaled differently based off of how much they agree with other user's validations. This would be looked at separately from labeling interaction data and accuracy.

jonfroehlich commented 4 years ago

Captured here now: https://github.com/ProjectSidewalk/sidewalk-quality-analysis/issues/54