We've seen this develop as a trend – Netflix is the most obvious example, but Yelp is a good business case too, I believe – and I think it's generally positive. I am, however, concerned that there's a capacity for negative social/cultural impact with these recommendation engines when applied to topics that are matters of taste (as opposed to something like "config steps users like you found helpful"). My concern is that Netflix's recommendations and crowdsourced tools like Yelp fuel a regression to the mean in terms of content/experience, and that mean may not actually exist. There might not be an "average taste" when it comes to food or content, and these systems may filter out great, idiosyncratic options. Do you think there's a way the precision vs accuracy measures can be used to control for this, or that these recommendation engines can be implemented in such a way that still surfaces odd choices?
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
We've seen this develop as a trend – Netflix is the most obvious example, but Yelp is a good business case too, I believe – and I think it's generally positive. I am, however, concerned that there's a capacity for negative social/cultural impact with these recommendation engines when applied to topics that are matters of taste (as opposed to something like "config steps users like you found helpful"). My concern is that Netflix's recommendations and crowdsourced tools like Yelp fuel a regression to the mean in terms of content/experience, and that mean may not actually exist. There might not be an "average taste" when it comes to food or content, and these systems may filter out great, idiosyncratic options. Do you think there's a way the precision vs accuracy measures can be used to control for this, or that these recommendation engines can be implemented in such a way that still surfaces odd choices?