Here is an interesting new article about how network analysis can help invade privacy. And you know what the biggest cited culprit is? That's right: computation of community structure (which, as many of you know, is something I work on).
And, of course, I noticed when reading this that one of the linked pieces of evidence had a bad characterization of what modularity values indicate good community structure.
This is a good article, and when the author writes that ideas about network communities can be used to glean information people don't disclose, he is talking "community structure" in the technical sense. (And, yes, computing community structure can give one a very good idea about information that has not been revealed based on information in one's community that has been revealed. See, e.g. this paper of mine.)
Now I better read the final 6 pages of that Masters-dissertation draft. I keep avoiding finishing this...
(Tip of the cap to Peter Mucha.)
No comments:
Post a Comment