Sentiment analysis and emotion recognition: Evolving the paradigm of communication within data classification

The process of sentiment analysis and emotion recognition (SAER) entails using artificial intelligence components and algorithms to extract emotions and sentiments from online texts, such as tweets. The information extracted can then be used by marketing, customer support and public relations teams to foster positive consumer attitudes. Advances in this discipline, however, are being hindered by two significant obstacles. First, although ‘emotion’ and ‘sentiment’ are distinct entities that require distinct analysis, there is no agreed definition to distinguish between the two. Secondly, the nature of language within the electronic medium has evolved to include much more than textual statements, including (but not limited to) acronyms, emojis and other visuals, such as video (in its many forms). As visual communication lacks universal interpretation, this can lead to erroneous analysis and conclusions, even where there is a differentiation between emotion and sentiment. This paper uses examples and case studies to explain the theoretical basis of the problem. It also offers conceptual direction regarding how to make SAER more accurate.