Emotion-corpus guided lexicons for sentiment analysis on Twitter.
Bandhakavi, Anil; Wiratunga, Nirmalie; Massie, Stewart; Deepak, P.
Research in Psychology have proposed frameworks that map emotion concepts with sentiment concepts. In this paper we study this mapping from a computational modelling perspective with a view to establish the role of an emotion-rich corpus for lexicon-based sentiment analysis. We propose two different methods which harness an emotion-labelled corpus of tweets to learn world-level numerical quantification of sentiment strengths over a positive to negative spectrum. The proposed methods model the emotion corpus using a generative unigram mixture model (UMM), combined with the emotion-sentiment mapping proposed in Psychology  for automated generation of sentiment lexicons. Sentiment analsysis experiments on benchmark Twitter data sets confirm the equality of our proposed lexicons. Further a comparative analysis with standard sentiment lexicons suggest that the proposed lexicons lead to a significantly better performance in both sentimentclassification and sentiment intensity prediction tasks.
|Start Date||Dec 13, 2016|
|Publication Date||Nov 5, 2016|
|Publisher||Springer (part of Springer Nature)|
|Institution Citation||BANDHAKAVI, A., WIRATUNGA, N. and MASSIE, S. 2016. Emotion-corpus guided lexicons for sentiment analysis on Twitter. In Bramer, M. and Petridis, M. (eds.) 2016. Research and development in intelligent systems XXXIII: incorporating applications and innovations in intelligent systems XXIV: proceedings of the 36th SGAI nternational conference on innovative techniques and applications of artificial intelligence (SGAI 2016), 13-15 December 2016, Cambridge, UK. Cham: Springer [online], pages 71-86. Available from: https://doi.org/10.10007/978-3-319-47175-4_5|
|Keywords||Knowledge discoverty; Data mining; Speech; Natural language interfaces; Machine learning; Ontologies semantic web|
BANDHAKAVI 2016 Emotion-corpus guided lexicons