Modeling Web Sources for Information Integration信息集成建模.pptVIP

Romantic partnerships and the dispersion of social ties Romantic Partnerships and the Dispersion of Social Ties (Backstrom Kleinberg) Questions Who are the most important individuals in a person’s social neighborhood? What are the defining structural signatures of a person’s social neighborhood? Contributions Dispersion: a new measure for estimating tie strength Characterize romantic relationships in terms of network structure Empirical study of this characteristic across Facebook population Who are the most important people in one’s social neighborhood? Following Granovetter, researchers use number of mutual friends (embeddedness) to identify strong ties Close friends, who share much time together Emotionally intense interactions A B C D E F C D E F A B A-B tie is highly embedded in the network A-B tie is not embedded in the network Romantic ties Embeddedness is not able to identify “significant others” (romantic relationships, e.g., spouse, partner, boy/girlfriend) Ego network – social neighborhood of an individual, showing all his/her friends and links between them Ego network of an individual Who is the “significant other”? Social foci People have large clusters of friends corresponding to well-defined foci of interaction in their lives These links have high embeddedness but are not very strong ties In contrast, romantic partners may have lower embeddedness, but they often involve mutual friends from different foci Ego network of an individual College friends Co-workers Embeddedness vs dispersion Dispersion: mutual neighbors of u and v are not well-connected to one another, and hence u and v are the only intermediaries between these different parts of the network. Link u-h has high dispersion: u and h are the only intermediaries between c and f Embeddedness: u and v have many mutual neighbors. Links u-b, u-c, and u-f have embeddedness 5 Link u-h has embeddedness 4 Link dispersion s(u,h)=4 disp(u,b)=1 Evaluation Egonetworks of 1.3 million Facebook users, sel