On 31 Oct 2013 at 11:59, Steven Watanabe wrote:
No automated system can correctly distinguish between - serious bug - minor bug - user error - "yeah, it would be nice. I might get to it someday" - &c.
Such a system isn't looking for that - rather, it is looking for any activity by the maintainer at all. Comments, for example, count. The most important thing, I think, is detecting user frustration and anger. If a library is seeing a lot of new issues containing frustration, that is a good sign a closer look is needed. I agree that coding up the appropriate Bayesian filters is probably a bit hard ... but I did read somewhere about some Google web service API which can be used to grok text for sentiments. Anyway, this is all surely many years out after modularisation has been around for a while. Niall -- Currently unemployed and looking for work. Work Portfolio: http://careers.stackoverflow.com/nialldouglas/