Author |
Topic: social news via targeting within social networks (Read 336 times) |
|
amichail
Senior Riddler
Posts: 450
|
|
social news via targeting within social networks
« on: Nov 21st, 2006, 2:04pm » |
Quote Modify
|
http://newsbuffer.com is interesting, but it still seems easy to spam via fake accounts and/or link sabotage. For example, you could create lots of fake accounts to submit the same link to narrowly targeted groups, thus achieving a high ranking for that link for each of those groups. If multiple submissions of the same link are not allowed, then you can sabotage a link by targeting it poorly. But the combination of targeting and social networks is promising. Maybe the following idea will be more resistant to spam. Consider a social network with unidirectional edges. If A makes B his/her friend, then there is a unidirectional edge from B to A. One can target a link to anyone reachable via a path in the social network. Targeting can be done based on links, tags, and/or other criteria. As with newsbuffer.com, narrowly targeted links would rank more highly. So suppose A targets a link to some group of people B_1, ..., B_k reachable from A in the social network. If a B_i likes the link received from A, then B_i could do nothing. B_i could also make A his friend to ensure that he/she will receive links from A in the future (i.e., to make sure that the path between them will not be broken). If a B_i dislikes the link received from A, then B_i could do nothing. B_i could also ban A, thus preventing A from sending links to B_i. More generally, B_i could ban "key" people along path(s) from A to B_i. This would be important in combating spam with fake accounts. In particular, consider the dominator tree rooted at B_i where social network edges are considered in reverse (see http://en.wikipedia.org/wiki/Dominator). Then B_i may consider banning one or more of his/her descendants in this dominator tree along the path to A.
|
|
IP Logged |
DropZap - a new kind of block elimination game
|
|
|
|