Difference between revisions of "IntrospectionBot"
Jump to navigation
Jump to search
(Created page with "Category: SFO Category: Projects") |
|||
| Line 1: | Line 1: | ||
[[Category: SFO]] [[Category: Projects]] | [[Category: SFO]] [[Category: Projects]] | ||
| + | |||
| + | IntrospectionBot is a stretch break for your System 1. It reads whatever public online interaction of yours you ask it to track, analyses its emotional content ([http://nlp.stanford.edu/sentiment/ sentiment]) over time, and contacts you when the trendline moves abruptly to suggest taking a break for some introspection. | ||
| + | |||
| + | ==Features== | ||
| + | * Authenticated opt-in (you add yourself to the app via OAuth/Facebook Connect; harassing other people via helpful bots is not ok) | ||
| + | * Channels for notification: public message (it tweets at you/posts to your wall), private message, email, any of these? | ||
| + | |||
| + | ==Potential issues== | ||
| + | * Suppose John dies and his friends find out about it on social media. An abrupt, negative, public response is *entirely appropriate* here; IntrospectionBot would be interrupting their shared grieving if it advised all of them to back off and that would be wrong. | ||
| + | ** Different kinds of responses for different types of detected sentiment? | ||
| + | *** sadness | ||
| + | *** anger/hatred | ||
| + | *** fear | ||
| + | *** others? | ||
Latest revision as of 00:21, 18 May 2015
IntrospectionBot is a stretch break for your System 1. It reads whatever public online interaction of yours you ask it to track, analyses its emotional content (sentiment) over time, and contacts you when the trendline moves abruptly to suggest taking a break for some introspection.
Features
- Authenticated opt-in (you add yourself to the app via OAuth/Facebook Connect; harassing other people via helpful bots is not ok)
- Channels for notification: public message (it tweets at you/posts to your wall), private message, email, any of these?
Potential issues
- Suppose John dies and his friends find out about it on social media. An abrupt, negative, public response is *entirely appropriate* here; IntrospectionBot would be interrupting their shared grieving if it advised all of them to back off and that would be wrong.
- Different kinds of responses for different types of detected sentiment?
- sadness
- anger/hatred
- fear
- others?
- Different kinds of responses for different types of detected sentiment?