MRT: Tracing the Evolution of Scientific Publications

Official Site | Demo

The implementation for the paper "MRT: Tracing the Evolution of Scientific Publications" (TKDE 2021 accepted).

Introduction

The MRT (Master Reading Tree) is designed to help researchers quickly find the evolution roadmap of a target paper, for example, tracing how the famous paper BERT evolves. A demo of the BERT roadmap.

Specifically, given a research paper, the designed algorithm will try to extract a small citation network with metainfo of papers and then build a roadmap to sketch different tracks of the target paper's evolution.

This repo includes two code libraries:

1. mrtframework, the backend algorithm for generating roadmaps. You can either use the code to discover how the library works or how evaluations are conducted.
2. react-mrt, the frontend UI (a React component) for displaying a generated roadmap. You can use this framework to integrate dynamic roadmaps in your own website.

This whole framework has already been integrated into the AMiner online system where you can click the "Generate MRT" button on paper pages to generate roadmaps for your interested papers.

Citation

Please cite the following works if you find the work of MRT helps you in your research.

@article{yin2021mrt,
  title={MRT: Tracing the Evolution of Scientific Publications},
  author={Yin, Da and Tam, Weng Lam and Ding, Ming and Tang, Jie},
  journal={IEEE Transactions on Knowledge and Data Engineering},
  year={2021},
  publisher={IEEE}
}