Ontology based sentiment analysis for fake review detection

作者：

Highlights：

• Fake reviews can be detected by incorporating review content related features.

• Linguistic, POS tagging and sentiment features can utilize as content related factors.

• Aspect based sentiment analysis was incorporated for exploring sentiment features.

• Outliers from unlabeled dataset were considered as spam reviews for model training.

• Classification results will be generated according to the SWRL rules of the ontology.

摘要

•Fake reviews can be detected by incorporating review content related features.•Linguistic, POS tagging and sentiment features can utilize as content related factors.•Aspect based sentiment analysis was incorporated for exploring sentiment features.•Outliers from unlabeled dataset were considered as spam reviews for model training.•Classification results will be generated according to the SWRL rules of the ontology.

论文关键词：Domain ontology,Rule-based classifier,Outliers,Feature-level sentiment analysis,Review-related features

论文评审过程：Received 7 July 2020, Revised 3 May 2022, Accepted 11 June 2022, Available online 18 June 2022, Version of Record 25 June 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117869