Propagating sentiment signals for estimating reputation polarity
作者：
Highlights：
• Sentiment lexicons can be augmented to create reputation polarity lexicons.
• Learning PMI values from training data is very effective for reputation polarity.
• Sentiment signals can be propagated to annotate reputation polarity.
• Pairwise similarity performs better than clustering tweets thematically.
• Weakly supervised annotation of reputation polarity is feasible.
摘要
•Sentiment lexicons can be augmented to create reputation polarity lexicons.•Learning PMI values from training data is very effective for reputation polarity.•Sentiment signals can be propagated to annotate reputation polarity.•Pairwise similarity performs better than clustering tweets thematically.•Weakly supervised annotation of reputation polarity is feasible.
论文关键词：Online reputation analysis,Sentiment propagation,Social media analysis
论文评审过程：Received 20 December 2018, Revised 11 January 2019, Accepted 6 July 2019, Available online 22 July 2019, Version of Record 22 July 2019.
论文官网地址：https://doi.org/10.1016/j.ipm.2019.102079
