Sentiment analysis using novel and interpretable architectures of Hidden Markov Models
作者：
Highlights：
• We introduce Hidden Markov Models for sentiment analysis.
• The introduced HMMs have high interpretability and visualize the decisions made.
• They show the way that sentiments change from the start to the end of sentences.
• Different architectures, orders, ensembles and training methods are examined.
• The introduced HMMs outperform traditional HMMs and machine learning methods.
摘要
•We introduce Hidden Markov Models for sentiment analysis.•The introduced HMMs have high interpretability and visualize the decisions made.•They show the way that sentiments change from the start to the end of sentences.•Different architectures, orders, ensembles and training methods are examined.•The introduced HMMs outperform traditional HMMs and machine learning methods.
论文关键词：Sentiment analysis,Hidden Markov Models,Interpretability,Ensemble learning,High-order Hidden Markov Models
论文评审过程：Received 14 December 2020, Revised 13 July 2021, Accepted 20 July 2021, Available online 27 July 2021, Version of Record 12 August 2021.
论文官网地址：https://doi.org/10.1016/j.knosys.2021.107332
