An Opinion Mining Approach to Handle Perspectivism and Ambiguity: Moving Toward Neutrosophic Logic

One main incentive for the utilization of opinion mining (OM) in social media is the impossibility of manually analyzing millions of opinions. However, applying OM for decision-making requires imitating human brain reasoning for more pragmatic results. Analyzing opinions with human-like intelligence is vital for avoiding misleading results. This occurs when the process can properly excogitate social influence and environmental uncertainty. In this study, an OM model for Twitter is proposed to handle perspectivism and opinion ambiguity. For perspectivism, social network analysis (SNA) is conducted, where users are ranked and then weighted using the UCINET tool and neural networks. An uncertainty classifier is used to integrate users’ influence levels with the polarity scores of their texts, providing a new polarity score that can reflect the real-world reasoning of opinions. The polarity scores needed for integration are done using the lexicon resource TextBlob. In the proposed model, three uncertainty classifiers are tested: type1 fuzzy logic (T1-FL), type2 fuzzy logic (T2-FL), and neutrosophic logic (NL). A comparative analysis of the methods shows the ability of NL to deal with the uncertainty existing in the data more accurately, proving the benefit of NL in improving the power of OM in social media.