Latent Space Point Process Models for Dynamic Networks

Python 2.7 code for the paper:

Decoupling homophily and reciprocity with latent space network models.
Jiasen Yang, Vinayak Rao, and Jennifer Neville.
In Proceedings of the 33rd Conference on Uncertainty in Artificial Intelligence (UAI), 2017

Description of Files

point_process.py: Base class for point process models.

hawkes_simple.py, hawkes_simple_test.py: Implements the HP model.

hawkes_embedding.py, hawkes_embedding_test.py: Implements the DLS model.

hawkes_embedding2.py, hawkes_embedding2_test.py: Implements the RLS model.

The PLS and BLS models can be obtained by regularizing the DLS model.

embedding.py: Various functions for evaluating embeddings.

helper.py: Miscellaneous utility functions.

Demo Example

Learning the DLS model from simulated data: Hawkes-DLS-demo.ipynb

Enron data file (in cPickle format): enron-events.pckl

Please contact Jiasen Yang for questions or comments.