ProjectSidewalk / sidewalk-quality-analysis

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

Meeting 8/6 #44

Closed daotyl000 closed 4 years ago

daotyl000 commented 5 years ago
    • Could look through old project sidewalk paper for the "Data Validation Study" section where Project Sidewalk has labeled 44 miles and had Turkers go through those routes to test accuracy. Could look at their interaction/label data
    • Ask @misaugstad if it is possible for Neil & I to go through the same missions that users have gone through and we label them ourselves to see how accurate they were
    • Potentially create a test mission that we have labeled as ground truth for users to run through that label themselves. We can use these results to calculate their accuracy
    • Look back at the HCOMP proceedings & get a sense for what has been published. Is there a prevalence of papers that have been published that relate to our study