828 followers
2,161 followers
RT @RDBinns: Also useful for problem domains where ground truth is subjective/contested and therefore potentially unsuitable for ML, e.g. o…
6,797 followers
Also useful for problem domains where ground truth is subjective/contested and therefore potentially unsuitable for ML, e.g. offensive speech detection (see https://t.co/cv0hCOM0ZK)
5,028 followers
RT @mikarv: We analysed some of these challenges with regard to comment “toxicity” detection in a paper last year: “Like trainer, like bot?…