Social network datasets (directed) from KONECT

Introduction

The joint degree (in- and out-degree) distributions of social networks have some unique features which are different from traditional Web hyperlink and citation networks. The differences motivated us to explore an explanation by studying the origin of the distribution. We proposed several two-dimensional mathematical generative models to capture the evolution of the joint degree in social networks. The social network databases were downloaded from KONECT (Koblenz Network Collection, a project to collect large network datasets of all types). We use simulation to compare between the synthetic data from our generative model and the data from real networks. We validate the success of our models by fitting and comparing the asymptotic dependence between synthetic and real data. An abstract was presented in NetSci 2015 and a paper was presented in ACC, 2016 (http://ieeexplore.ieee.org/document/7525372/). Github: https://github.com/ShanLu1984/KONECT-Social-Network-Datasets.

What's inside:

This projects include extracting degree information from TSV file downloaded from KONECT Website http://konect.uni-koblenz.de/; and data visiualization.

1. network_read.m

• Usage: get indegree and outdegree of the nodes in the network;
• Output: degrees.mat It contains a matrix data with dimension (n, 2), n is the number of nodes in the network; the first column is the indegree of the node; and the second column is the outdegree.

2. network_show.m

• Usage: plot the joint degree distribution of nodes in the network. The correlation between in-degree and out-degree is computed and show on the title of the figure;
• Output: .fig, joint degree of each node in log-log scale.

3. overlap_fraction.m

• Usage: get the nodes overlap percentage between top x% nodes ranked by outdegree and indegree in descending order. This figure reflects the asymptotic dependency between in-degree and out-degree of nodes with the increasing of nodes' degree;
• Output: .fig, overlap percentage as a function of fraction of top users ordered by out/in-degree.

4. Inside PCSDE model for 2d power law data:

| function | description | output | |-----------------|------------------------| ---------------------------| | GenerativeModel2D_Bollobas.m | Generate networks from Bollobas model | A: adjacency matrix; D: degree matrix, D(i, 1) indegree of node i; D(i,2): outdegree of node i | | PCSDEwithB_gamma_05_for_Bollobas.m | Generate sample data from PCSDE model with the same distribution as Bollobas model | Figure | | PCSDEwithB_gamma_vary_for_social_network_data.m | Generate sample data from PCSDE model with the same distribution as social network datasets | Figure |

Datasets:

Facebook wall posts http://konect.uni-koblenz.de/networks/facebook-wosn-wall

The is the directed network of a small subset of posts to other user's wall on Facebook. The nodes of the network are Facebook users, and each directed edge represents one post, linking the users writing a post to the users whose wall the post is written on. Since users may write multiple posts on a wall, the network allows multiple edges connecting a single node pair. Since users may write on their own wall, the network contains loops.

• Size: 46,952 vertices (users)
• Volume: 876,993 edges
• Average degree (overall): 37.357 edges / vertex
• Reciprocity: 62.5%

Youtube links http://konect.uni-koblenz.de/networks/youtube-links

This is the social network of Youtube users and their connections.

• Size: 1,138,499 vertices (users)
• Volume: 4,942,297 edges (links)
• Average degree (overall): 8.6821 edges / vertex
• Reciprocity: 79.0%

Wikipedia talk, English http://konect.uni-koblenz.de/networks/wiki_talk_en

This is the communication network of the English Wikipedia. Nodes represent users of the English Wikipedia, and an edge from user A to user B denotes that user A wrote a message on the talk page of user B at a certain timestamp.

• Size: 2,987,535 vertices (users)
• Volume: 24,981,163 edges (messages)
• Average degree (overall): 16.724 edges / vertex
• Reciprocity: 21.5%

Flickr Friendship http://konect.uni-koblenz.de/networks/flickr-growth

This is the social network of Flickr users and their friendship connections.

• Size: 2,302,925 vertices (users)
• Volume: 33,140,017 edges (friendships)
• Average degree (overall): 28.781 edges / vertex
• Reciprocity: 62.2%

LiveJournal Frendship http://konect.uni-koblenz.de/networks/soc-LiveJournal1

This is the social network of LiveJournal. Nodes are users of LiveJournal, and directed edges represent friendships.

• Size: 4,847,571 vertices (users)
• Volume: 68,475,391 edges (friendships)
• Average degree (overall): 28.251 edges / vertex
• Reciprocity: 74.8%

Google Hyperlink http://konect.uni-koblenz.de/networks/web-Google

This is a network of web pages connected by hyperlinks. The data was released in 2002 by Google as a part of the Google Programming Contest.

• Size: 875,713 vertices (webpages)
• Volume: 5,105,039 edges (hyperlinks)
• Average degree (overall): 11.659 edges / vertex
• Reciprocity 30.7%

How to run:

1. Select the dataset you are interested in, download the TSV file from the KONECT website.
2. Extract the .tar file.
3. Change the code in network_read.m. For example, for Facebook dataset:

node_size = 46952;
dataset = 'out.FACEBOOK-WOSN-WALL';

4.Type "network_read" in Command Window or open file "network_read.m" and click "Run" button on the top bar. 5. To show the joint degree distribution of the network, change the code in network_show.m. For Facebook dataset:

load('.\facebook-wosn-wall\degrees.mat');
title(['facebook-wosn-wall, correlation: ' num2str(corr_inout)]);

6. Type "network_show" in Command Window or open file "network_show.m" and click "Run" button on the top bar.
7. To show the asymptotic dependency of the nodes' in- and out-degree, change the code in overlap_fraction.m. For Facebook dataset:

load('.\facebook-wosn-wall\degrees.mat');
legend('facebook-wosn-wall');

8. Type "overlap_fraction in Command Window or open file "overlap_fraction.m" and click "Run" button on the top bar.