This repo conatins the Python implementation of the paper HRank: A Path based Ranking Framework in Heterogeneous Information Network. Of the 3 methods mentioned in the paper, two have been implemented:
- HRank-SY (for symmetric metapaths)
- HRank-AS (for asymmetric metapaths)
An additional count-based method was also implemented to compare the rankings. The rankings are compared using the RBO (Rank-biased Overlap) method.
- Write your own metapaths in metapaths.py in the required format or use the already existing metapaths.
- Run
python3 hrank.py. - The output would consist of the information related to the graph of metapaths, the rankings as well as the RBO score. The ranking results would also be stored in ranks.csv.