[
https://issues.apache.org/jira/browse/GIRAPH-144?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13224096#comment-13224096
]
Avery Ching commented on GIRAPH-144:
@Jakob, any more thoughts?
My personal take is that they do have similar function (they extract
communities), but they have a general different type of output. in
label propagation you'd end up with an id for each vertex (the vertex
id that is the centroid for the community each vertex belongs to),
while i'd expect from
while i'd expect from linlog something like an (x,y) coordinate. is
that correct?
(x,y,z) is also common -- i think of it as extreme dimension reduction
-- after which you're free to use your favorite gis tool --
http://postgis.refractions.net/ -- you can also pull in tools from
statistical
Ok, but why a bipartite graph? I can imagine you can have 2 or 3
coordinates associated with each vertices without actually having multiple
vertex types.
On Wednesday, March 7, 2012, Timmy Wilson tim...@smarttypes.org wrote:
while i'd expect from linlog something like an (x,y) coordinate. is
HBase/Accumulo Input and Output formats
---
Key: GIRAPH-153
URL: https://issues.apache.org/jira/browse/GIRAPH-153
Project: Giraph
Issue Type: New Feature
Components: bsp
Affects Versions:
[
https://issues.apache.org/jira/browse/GIRAPH-153?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Brian Femiano updated GIRAPH-153:
-
Labels: (was: gir)
Description:
Four abstract classes that wrap their respective