An empirical study of the design choices for local citation recommendation systems

作者：

Highlights：

• Certain design choices affect the accuracy of local citation recommendation systems.

• Training reranker model with a strict regime improves the model’s performance.

• Triplet-based reranking models benefit from non-random negative sampling strategies.

• The best negative sampling strategy for triplet construction depends on the dataset.

摘要

•Certain design choices affect the accuracy of local citation recommendation systems.•Training reranker model with a strict regime improves the model’s performance.•Triplet-based reranking models benefit from non-random negative sampling strategies.•The best negative sampling strategy for triplet construction depends on the dataset.

论文关键词：Natural language processing,Citation recommendation,Information retrieval,BM25,SPECTER,Negative sampling

论文评审过程：Received 30 July 2021, Revised 18 January 2022, Accepted 7 March 2022, Available online 19 March 2022, Version of Record 29 March 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.116852