schatzopoulos / ArtSim

Framework that can be applied on top of any existing popularity estimation method to improve its accuracy.

ArtSim & ArtSim+

This repository contains the code for the article "ArtSim: Improved estimation of current impact for recent articles." published in AIMinScience workshop @ TPDL 2020 and its extension ArtSim+ published in the QSS journal.

Execution

ArtSim and ArtSim+ can be executed with the scripts artsim.py and artsim_plus.py respectively as follows:

python3 artsim.py <paper_file> <scores_file> <pap_similarities_file> <ptp_similarities_file> <pv_connections_file> <cold_start_year> <evaluation_method> <ndcg:k>

where

• paper_file is a tsv file for mapping internal numeric ids to actual paper ids
• scores_file contains paper ids and their popularity scores computed by a popularity method
• pap_similarities_file and ptp_similarities_file are files containing similarities based on authors and topics respectively; they contain tuples of numeric paper ids with their respecitive similarity score. pv_connections_file contains the paper to venue relationships. The files we used for our experiments with the DBLP dataset can be found at (Zenodo)[https://zenodo.org/record/4567527]
• cold_start_year is the year after which we consider articles being in their cold start period.
• evaluation_method can be one of 'tau' or 'ndcg'
• in case of selecting 'ndcg' as an evaluation method in the previous parameter, we should also provide the k parameter of ndcg as an extra parameter

Please cite:

@article{chatzopoulos2021further,
  title={Further improvements on estimating the popularity of recently published papers},
  author={Chatzopoulos, Serafeim and Vergoulis, Thanasis and Kanellos, Ilias and Dalamagas, Theodore and Tryfonopoulos, Christos},
  journal={Quantitative Science Studies},
  pages={1--36},
  year={2021}
}

@inproceedings{chatzopoulos2020artsim,
  title={Artsim: improved estimation of current impact for recent articles},
  author={Chatzopoulos, Serafeim and Vergoulis, Thanasis and Kanellos, Ilias and Dalamagas, Theodore and Tryfonopoulos, Christos},
  booktitle={ADBIS, TPDL and EDA 2020 Common Workshops and Doctoral Consortium},
  pages={323--334},
  year={2020},
  organization={Springer}
}