Re: Apache Flink 1.1.4 - Gelly - LocalClusteringCoefficient - Returning values above 1?

Posted by Vasiliki Kalavri on
URL: http://deprecated-apache-flink-user-mailing-list-archive.369.s1.nabble.com/Apache-Flink-1-1-4-Gelly-LocalClusteringCoefficient-Returning-values-above-1-tp11187p11188.html

Hi Miguel,

the LocalClusteringCoefficient algorithm returns a DataSet of type Result, which basically wraps a vertex id, its degree, and the number of triangles containing this vertex. The number 11 you see is indeed the degree of vertex 5113. The Result type contains the method getLocalClusteringCoefficientScore() which allows you to retrieve the clustering coefficient score for a vertex. The method simply divides the numbers of triangles by the number of potential edges between neighbors.

I'm sorry that you this is not clear in the docs. We should definitely improve them to explain what is the output and how to retrieve the actual clustering coefficient values. I have opened a JIRA for this [1].

Cheers,

-Vasia.

[1]: https://issues.apache.org/jira/browse/FLINK-5597

On 20 January 2017 at 19:31, Miguel Coimbra <[hidden email]> wrote:

Hello,

In the documentation of the LocalClusteringCoefficient algorithm, it is said:

The local clustering coefficient measures the connectedness of each vertex’s neighborhood.
Scores range from 0.0 (no edges between neighbors) to 1.0 (neighborhood is a clique).

https://ci.apache.org/projects/flink/flink-docs-release-1.1/apis/batch/libs/gelly.html#local-clustering-coefficient

However, upon running the algorithm (undirected version), I obtained values above 1.

The result I got was this. As you can see, vertex 5113 has a score of 11:
(the input edges for the graph are shown further below - around 35 edges):

(4907,(1,0))
(5113,(11,0))
(6008,(0,0))
(6064,(1,0))
(6065,(1,0))
(6107,(0,0))
(6192,(0,0))
(6252,(1,0))
(6279,(1,0))
(6465,(1,0))
(6545,(0,0))
(6707,(1,0))
(6715,(1,0))
(6774,(0,0))
(7088,(0,0))
(7089,(1,0))
(7171,(0,0))
(7172,(1,0))
(7763,(0,0))
(7976,(1,0))
(8056,(1,0))
(9748,(1,0))
(10191,(1,0))
(10370,(1,0))
(10371,(1,0))
(14310,(1,0))
(16785,(1,0))
(19801,(1,0))
(26284,(1,0))
(26562,(0,0))
(31724,(1,0))
(32443,(1,0))
(32938,(0,0))
(33855,(1,0))
(37929,(0,0))

This was from a small isolated test with these edges:

5113    6008
5113    6774
5113    32938
5113    6545
5113    7088
5113    37929
5113    26562
5113    6107
5113    7171
5113    6192
5113    7763
9748    5113
10191    5113
6064    5113
6065    5113
6279    5113
4907    5113
6465    5113
6707    5113
7089    5113
7172    5113
14310    5113
6252    5113
33855    5113
7976    5113
26284    5113
8056    5113
10371    5113
16785    5113
19801    5113
6715    5113
31724    5113
32443    5113
10370    5113

I am not sure what I may be doing wrong, but is there perhaps some form of normalization lacking in my execution of:

org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient.Result;
org.apache.flink.graph.library.clustering.undirected.LocalClusteringCoefficient;

Am I supposed to divide all scores by the greatest score obtained by the algorithm?

Thank you very much!