Optimizing Sentiment Analysis with Joint Implicit and Explicit Aspect Detection and Psychomotor Attribute Integration

Aspect-Based Sentiment Analysis (ABSA) employed in the current study is defined as a sub-level of analysis in which opinions at a finer-grained level entailing sentiments related to aspects of specific entities within textual data are captured. Whereas, the identification and enumeration of the explicit aspects, which have been directly referred to, is not much of a problem and the same goes for the hidden ones, which are also known as the latent aspects, is still elusive to current techniques of sentiment analysis. Furthermore, the correlation between identified aspects and psychomotor properties, including behavioral and emotional aspects, has not been studied to a satisfactory degree. This research work presents a new framework for ABSA designed to enhance the identification of implicit and explicit aspects and their relationship to psychomotor characteristics. The framework consists of three core modules: The three stages involved are, the identification of implicit aspects, identification of explicit aspects, and linking of identified aspects with the psychomotor data. The use of both topic modelling as well as attention-based deep learning methods enables the framework itself to yield vast improvements in regards to implicit aspect extraction; the study documents a precision of 85% as well as an F1-score of 82%, the latter of which is an improvement of approximately 10 % over the baseline established by previous methods. In addition, moderate to strong correlation between the identified aspects and psychomotor properties (r = 0.68) also suggests the possibility of using sentiment analysis together with behavioral data. Through these findings, it is shown that the proposed framework can serve as an extension of ABSA’s existing uses in fields like customer review information, human–computer interaction, and mental health care by offering further analysis of the emotion and psychological aspects of the user’s content generation.