Sentiment visualization and classification via semi-supervised nonlinear dimensionality reduction

作者：

Highlights：

• A new method is proposed for efficient sentiment visualization and classification.

• The method utilizes semi-supervised learning to save cost of labeling.

• Visibility and prediction accuracy are improved in spite of reduced features.

摘要

Highlights•A new method is proposed for efficient sentiment visualization and classification.•The method utilizes semi-supervised learning to save cost of labeling.•Visibility and prediction accuracy are improved in spite of reduced features.

论文关键词：Text visualization,Semi-supervised dimensionality reduction,Laplacian eigenmaps,Sentiment classification

论文评审过程：Received 26 November 2012, Revised 10 July 2013, Accepted 26 July 2013, Available online 10 August 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.07.022