Network Based Inference

This project contains code implementing the Network Based Inference technique for use as a recommendation engine.

Getting Started

The code implemented works as a sci-kit learn classifier. It will fit the NBI weighted adjacency matrix on a bipartite graph of person-by-object. It will then predict likely objects for a new person, given any past object acquisition behaviour. Please see the reference below for the details of algorithm.

Prerequisites

The following packages will be needed to run the code.

sklearn
numpy

Authors and References

• Iain J. Cruickshank
• Tao Zhou, Jie Ren, Matus Medo, and Yi-Cheng Zhang, Bipartite network projection and personal recommendation. Physical Review E 76(4): 046115, 2007

License

This project is licensed under the MIT License - see the LICENSE.md file for details