Comments on: Using social networks for persuasion profiling http://www.deaneckles.com/blog/146_using-social-networks-for-persuasion-profiling/?utm_source=rss&utm_medium=rss&utm_campaign=using-social-networks-for-persuasion-profiling Dean Eckles blogs on people and technology Wed, 28 Apr 2010 05:13:23 -0700 http://wordpress.org/?v=2.8.6 hourly 1 By: Dean Eckles http://www.deaneckles.com/blog/146_using-social-networks-for-persuasion-profiling/comment-page-1/#comment-9703 Dean Eckles Thu, 03 Dec 2009 08:57:46 +0000 http://www.deaneckles.com/blog/?p=146#comment-9703 In a situation where the profiler has already observed considerable behavior by the user (e.g., on Amazon they have already looked and responded to some offers), then I think the user's own behavior and psychographics will be the best predictors. When this isn't available (e.g., new user for the profiler, noisy behavioral data), then using the behaviors of their social neighborhood will be valuable. In these cases, it will be critical to be able to identify which of the many members of their neighborhood matter the most -- perhaps based on engagement with the content they are creating. (I gather this is what you're thinking of?) Both of the latter cases highlight the potential value of shrinkage estimators (as in multilevel models) in a social network context... I wonder if there is much work on this. In a situation where the profiler has already observed considerable behavior by the user (e.g., on Amazon they have already looked and responded to some offers), then I think the user’s own behavior and psychographics will be the best predictors. When this isn’t available (e.g., new user for the profiler, noisy behavioral data), then using the behaviors of their social neighborhood will be valuable. In these cases, it will be critical to be able to identify which of the many members of their neighborhood matter the most — perhaps based on engagement with the content they are creating. (I gather this is what you’re thinking of?)

Both of the latter cases highlight the potential value of shrinkage estimators (as in multilevel models) in a social network context… I wonder if there is much work on this.

]]> By: Enrique Allen http://www.deaneckles.com/blog/146_using-social-networks-for-persuasion-profiling/comment-page-1/#comment-9239 Enrique Allen Thu, 19 Nov 2009 20:07:30 +0000 http://www.deaneckles.com/blog/?p=146#comment-9239 Great post Dean, I'm also excited about the mechanisms underlying effective persuasion profiling. In particular I suspect that the processes of relationship creation/management, measured by various chained engagement behaviors such as photo tagging, contain the most rich "scent"of explanation. Great post Dean, I’m also excited about the mechanisms underlying effective persuasion profiling. In particular I suspect that the processes of relationship creation/management, measured by various chained engagement behaviors such as photo tagging, contain the most rich “scent”of explanation.

]]> By: Joan Griffin http://www.deaneckles.com/blog/146_using-social-networks-for-persuasion-profiling/comment-page-1/#comment-5834 Joan Griffin Tue, 16 Jun 2009 21:09:42 +0000 http://www.deaneckles.com/blog/?p=146#comment-5834 interesting and far-reaching implications attached to one's web relationships. interesting and far-reaching implications attached to one’s web relationships.

]]>