A semantic approach to post-retrieval query performance prediction
作者:
Highlights:
• Query performance predictors are proposed based on measuring semantic similarities.
• Four different semantic similarity models are adopted by the proposed predictors.
• The proposed predictors outperform baselines across different datasets.
• The proposed predictors outperform all baselines on single-entity questions.
• The choice of semantic similarity model impacts the proposed predictors performance.
摘要
•Query performance predictors are proposed based on measuring semantic similarities.•Four different semantic similarity models are adopted by the proposed predictors.•The proposed predictors outperform baselines across different datasets.•The proposed predictors outperform all baselines on single-entity questions.•The choice of semantic similarity model impacts the proposed predictors performance.
论文关键词:Query performance prediction,Semantic-enabled prediction,Post-retrieval prediction,Semantic information retrieval
论文评审过程:Received 12 February 2021, Revised 16 July 2021, Accepted 1 September 2021, Available online 23 September 2021, Version of Record 23 September 2021.
论文官网地址:https://doi.org/10.1016/j.ipm.2021.102746