This project implements collaborative filtering using a graph traversal pattern to generate recommendations for GitHub user follows. More information is available in this paper.

GitHub follow data was collected from Githubarchive.org for the period April 1, 2013 - April 1, 2014. This data was anonymized and serialized to edgelist format:

1	2
3	4
5	6
7	8
9	10
11	12
11	13
14	15
16	17
18	19

Where the first column corresponds to the user id of the source id, the second column indicates the destination user id. The example above therefore represents user 1 follows user 2, user 3 follows user 4... The edgelist file for this project is data/git_follow/edgelist in this repository.

This data was then inserted into a Neo4j graph database instance for analysis. All user data is anonymized and is stored in Neo4j format in the data directory in this project.

Maven - this project uses Maven for dependency resolution and building, install Maven first:

• sudo apt-get update
• sudo apt-get install maven

• Clone this repository: git clone https://github.com/johnymontana/githubneo.git
• cd into the githubneo directory: cd githubneo
• Run with maven: mvn exec:java -Dexec.mainClass="RecommendFollows"

The default configuration will run cross fold validation using random samples of the GitHub user graph for various configurations of the parameters. NOTE: this will take a while to run.