gg

gg

Github users community detection visualisation. gg will extract user related resources from Github and construct a graph presenting relationships among the users.

For example if user A reviewed a pull request of user B then gg will connect users A and B with relation REVIEWED.

Neo4j graph database was used to present Github users and detect hidden communities using neo4j native community detection algorithms.

Find out more implementation details chekout docs/.

Design

Requirements

• docker 19.03.X

Run

Start neo4j

NOTE: /plugins and /conf should be mounted in order to detect clusters

docker run \
    --name testneo4j \
    -p7474:7474 -p7687:7687 \
    -d \
    --env NEO4J_AUTH=neo4j/test \
    --memory 4gi \
    -v ./conf:/conf \
    -v ./plugins:/plugins \
    neo4j:latest

or

./run.sh

GITHUB_TOKEN can be obtained at: https://github.com/settings/tokens

When neo is up and running run gg

docker run \
    --rm --name gg \
    --network host \
    -e NEO_URI=bolt://localhost:7687 \
    -e NEO_USER=neo4j \
    -e NEO_PASS=test \
    -e GITHUB_TOKEN=$GITHUB_TOKEN \
    piotrpersona/gg

or ask for help

$ docker run piotrpersona/gg --help
Fetch repositories from a github and build a graph

Usage:
  gg [flags]

Flags:
  -h, --help                help for gg
      --issue-comment int   Weight of issue comment (default 10)
      --loglevel string     Log level (default "info")
  -p, --password string     Neo4j connection password
      --pr-comment int      Weight of pull request comment (default 16)
  -q, --query string        Github repositories search query (default "stars:>=1000")
      --review int          Weight of review (default 10)
  -t, --token string        GitHub API Token String
      --uri string          Neo4j compatible URI
  -u, --username string     Neo4j connection username

Detect communities

Enter http://localhost:7474 and pass credentials defined with NEO_AUTH.

Call Louvain algorithm:

CALL algo.beta.louvain.stream('User', null,
{ graph: 'huge', direction: 'BOTH', weightProperty: 'Weight', includeIntermediateCommunities: true })
YIELD nodeId, community, communities
RETURN algo.asNode(nodeId).Name as name, community, communities
ORDER BY name ASC

Example result of louvain algorithm