CS599: Structure and Dynamics of Networked Information (Spring 2005) 02/07/2005: The Search for Communities
نویسنده
چکیده
We previously looked at communities as simply dense subgraphs graphs, or as subgraphs such that each node has a large fraction of its edges inside. Another view, taken in [3], starts from the hubs and authorities model. It argues that a structure of densely linked hubs and authorities is a common feature of communities. Such a core, a dense bipartite graph, can be considered the “signature” of a community Ideally, we would like to enumerate all such signatures, and expand them to communities. However, the complexity of doing so would be prohibitive. In fact, even finding just one large complete bipartite graph is NP-hard. However, when the given graph is dense enough, it always has a large complete bipartite subgraph.
منابع مشابه
CS599: Structure and Dynamics of Networked Information (Spring 2005) 02/02/2005: α-communities
Last time, we looked at a notion of community where we required high overall edge density among groups of nodes. This made no requirement on individual nodes. In particular, the definition favors the inclusion of high-degree nodes (in WWW terms: popular sites, e.g., yahoo.com, google.com, cnn.com). One may argue that communities should consist of nodes which predominantly belong to that communi...
متن کاملCs599: Structure and Dynamics of Networked Information (spring 2005) 02/23/2005: Rank Aggregation Scribes: Ranjit Raveendran and Animesh Pathak
In previous lectures, we had discussed the problem of searching for relevant results on the WWW by exploiting the link structure. Nowadays, there are already multiple search engines giving quite good results. However, given that there are so many search engines already, employing different techniques, we may be interested in combining their positive features, and constructing a meta-search engi...
متن کاملCs599: Structure and Dynamics of Networked Information (spring 2005) 04/25/2005: Analysis of Gossip-based Averaging Scribes: David Kempe 1 Averaging Using Gossip
In order to analyze this protocol, we will track what “share” of each node j’s number xj is currently contributing to each node’s sum si. So we look at a vector ~vi, in which the j component vi,j denotes the fraction of node j’s value that currently is part of node i’s sum. This means that initially, we have vi,i = 1 for all i, and vi,j = 0 for all i 6= j. In this view, our protocol can be expr...
متن کاملCs599: Structure and Dynamics of Networked Information (spring 2005) 02/16/2005: Uniform Labeling Problem Scribes: Yuriy Brun and Nupur Kothari 1 Uniform Labeling Problem via Lp Approximation
In the last lecture, we started studying the uniform labeling problem, where labels are to be assigned to nodes minimizing the combination of two cost objectives: a labeling cost for assigning a particular label at a particular node, and a separation cost for having different labels at adjacent nodes. The problem is known to be NP-hard for 3 or more labels, so we are looking for approximation a...
متن کاملCS599: Structure and Dynamics of Networked Information (Spring 2005) 03/23/2005: Small-World Networks
The fact that social networks seem to be small worlds has long been the subject of anecdotal observation. The hypothesis that “any two persons in the world are connected by a small chain of intermediate social acquaintances” is known as the small world phenomenon. The hypothesis was first tested by Stanley Milgram in 1967 [2]. Milgram found that in the experiment he conducted, there were an ave...
متن کامل