Skip to content

Implementation of CIKM'18 paper: "MEgo2Vec: Embedding Matched Ego Networks for User Alignment Across Social Networks∗".

Notifications You must be signed in to change notification settings

allanchen95/MEgo2Vec-Embedding-Matched-Ego-Networks-for-User-Alignment-Across-Social-Networks

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MEgo2Vec: Embedding Matched Ego Networks for User Alignment Across Social Networks∗

This is basic implementation of our CIKM'18 paper:

Jing Zhang, Bo Chen, Xianming Wang, Hong Chen*, Cuiping Li, Fengmei Jin, Guojie Song, and Yutao Zhang. 2018. MEgo2Vec: Embedding Matched Ego Networks for User Alignment Across Social Networks. In Proceedings of ACM conference (CIKM’18).

Requirements

  • Ubuntu 16.04
  • Python 2.7
  • Tensorflow-gpu
  • GPU,CUDA,CUDNN

Note: Running this project will consume upwards of 20GB hard disk space. The overall pipeline will take several hours. You are recommended to run this project on a Linux server.

Data Description

Training data in this demo is about AMiner - Linkedin networks which is placed in the data directory. If you want to download the original network data (AMiner, Linkedin), please use the link : https://pan.baidu.com/s/1b6_8jd8J9CGiCpyFBfZgoQ 密码:xacn . If you want to get other networks (Twitter, MySpace LastFm...), please click the link.

simplified_feature.p : Users' profile. Each line contains 4 fields of attributes(name, affiliation, education, publication). The line number is nodes' id.

network_A.p : Each line stands for a user's match-ego-network, which has the format : pairid, label, neighbor-num(n)(contatin itself), neighbor-id1...neighbor-idn, (n*n)adjacency matrix(1: connect, 0: disconnect).

train/test_network.p: Split the network_A.p into train or test network set, which is convenient to train or test other baselines.

new_pairs.p : The line number is the same as the network's pairid and neighbor-id. Each line contains two nodes id - node1_id, node2_id, which belong to one pair. The node_id is the same as simplified_feature.p's nodes_id.

How to run

cd code
python main.py

Note: Hyper parameter and training data in this demo is a little different than what we used in the experiments, so the performance (F1-score) will be a little bit lower than reported scores.

About

Implementation of CIKM'18 paper: "MEgo2Vec: Embedding Matched Ego Networks for User Alignment Across Social Networks∗".

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages