Hierarchical viewpoint discovery from tweets using Bayesian modelling

作者：

Highlights：

• A novel Bayesian model for hierarchical viewpoint discovery from tweets.

• Phrases with arbitrary length are modelled by a hierarchical Pitman-Yor process.

• Approaches for utilizing prior information like lexicons and hashtags are explored.

• Experiments are conducted against probabilistic models to show the effectiveness.

摘要

•A novel Bayesian model for hierarchical viewpoint discovery from tweets.•Phrases with arbitrary length are modelled by a hierarchical Pitman-Yor process.•Approaches for utilizing prior information like lexicons and hashtags are explored.•Experiments are conducted against probabilistic models to show the effectiveness.

论文关键词：Natural language processing,Opinion mining,Bayesian modelling

论文评审过程：Received 15 April 2018, Revised 11 September 2018, Accepted 12 September 2018, Available online 13 September 2018, Version of Record 23 September 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.09.028