Anomaly analysis based on meta-subspace approach for sentiment classification

Sudha, K.K.; Suguna, N.

doi:10.3233/jifs-181138

Cited by 1 publication

(2 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Detecting abnormal emotion becomes an important field of social media sentiment analysis. By extracting the valuable parts of the user's twitter corpus, it can help enterprises to make correct decisions on product evaluation, improve product quality, reduce unnecessary loss, and increase corporate profits according to users' emotional tendencies 2 …”

Section: Introductionmentioning

confidence: 99%

“…By extracting the valuable parts of the user's twitter corpus, it can help enterprises to make correct decisions on product evaluation, improve product quality, reduce unnecessary loss, and increase corporate profits according to users' emotional tendencies. 2 Many scholars have researched on detecting users' abnormal emotions on social media. Lin et al 3 proposed a factor graph convolutional neural network.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Smart abnormal emotion analysis from English sentences

Xu¹,

Chen

2020

Internet Technology Letters

View full text Add to dashboard Cite

Smart abnormal emotion analysis refers to identifying abnormal sentiments, opinions, or attitudes from massive patterns automatically. The abnormal emotion may be hidden in a paragraph of text to reflect the sentiment suddenly changes. The suddenly changed emotion should be identified in time to avoid severe consequence. For instance, a customer would not buy a product any more if he or she has a negative emotion to this product. With the popularization of social media, more and more information is available. For example, feedbacks, comments, or opinions widely exist on Twitter. Identifying abnormal emotion by analyzing texts in social media becomes a hot topic, which enables companies or government organizations to take prevention strategies in time if needing. In this paper, we propose a multivariate Gaussian model based abnormal emotion detection method. In multivariate Gaussian model, whether a user has abnormal emotion is determined by the joint probability density. The distribution test shows that the negative, positive, and neutral emotions of a user follow a normal distribution, while the surprise and anger emotions do not follow. The emotions of texts from social media posted by a group follow a “power law distribution”, while the individual users do not. The abnormal emotion can be detected by multivariate Gaussian about 84.60%.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%