“…"Or vedi l'anime di color cui vinse l'ira." 1 -Dante Alighieri, Divina Commedia, Inferno Online conversations have a reputation for going awry (Hinds and Mortensen, 2005;Gheitasy et al, 2015): antisocial behavior (Shepherd et al, 2015) or simple misunderstandings (Churchill and Bly, 2000;Yamashita and Ishida, 2006) hamper the efforts of even the best intentioned collaborators. Prior computational work has focused on characterizing and detecting content exhibiting antisocial online behavior: trolling (Cheng et al, 2015(Cheng et al, , 2017, hate speech (Warner and Hirschberg, 2012;Davidson et al, 2017), harassment (Yin et al, 2009), personal attacks (Wulczyn et al, 2017) or, more generally, toxicity (Chandrasekharan et al, 2017;Pavlopoulos et al, 2017b).…”