[ml] 11/24/2010 @ 7:30pm: Random Walks, PageRank, and Social Net Analysis

[Mike Schachter]

Actually those last two papers are a bit mathy.. a more
gentle but still math-sufficient writeup of PageRank can
be found at the end of chapter 2 of a book on numerical computing
with MATLAB:

http://www.mathworks.com/moler/chapters.html

PageRank is important because it's a graph algorithm that
ranks nodes. It does this by randomly walking around a
graph, moving from node to node, and ranks nodes basically
according to how often they will be visited on this "random walk".

The paper on supervised random walks for predicting links on
a social graph basically comes up with a weighted version of
the PageRank algorithm, where the "rank" of a node with respect
to another node is the probability of randomly walking to it on
the graph.

  mike




On Tue, Nov 23, 2010 at 4:57 PM, Mike Schachter <mike at mindmech.com> wrote:

> Also here's an online book for random walks on graphs
> I came across:
>
> http://stat-www.berkeley.edu/users/aldous/RWG/book.html
>
>
>
>
> On Tue, Nov 23, 2010 at 4:48 PM, Mike Schachter <mike at mindmech.com> wrote:
>
>> Hey everyone,
>>
>> Wednesday night I'll be talking about Random Walks,
>> PageRank, and a recent paper Shahin posted to the
>> list about social net analysis:
>>
>> http://arxiv.org/abs/1011.4071
>>
>> You might want to scan through this paper on personalized
>> PageRank as well:
>>
>> http://www.math.ucsd.edu/~fan/wp/lov.pdf<http://www.math.ucsd.edu/%7Efan/wp/lov.pdf>
>>
>> I'll keep the math to a minimum (as much as I'm capable of
>> doing :) and then we'll go on to talk about the Kaggle social
>> net competition and how that's going:
>>
>>
>> https://www.noisebridge.net/wiki/Machine_Learning/Kaggle_Social_Network_Contest
>>
>> See you there!
>>
>>   mike
>>
>>
>>
>>
>>
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.noisebridge.net/pipermail/ml/attachments/20101123/b024649d/attachment-0001.html>