Emotion-based Analysis of Reviews using Knowledge Graph

This study introduces a novel approach to sentiment analysis, combining the finesse of Knowledge Graphs with the unique perspective offered by the SenticNet and the Hourglass Model of Emotions. We aim to interpret, quantify, and categorize emotional responses in various reviews and offer, at the same time, a more nuanced understanding of user emotions than traditional sentiment analysis methods. Our process begins with extracting emotional indicators from reviews by linking their descriptive words with states of different emotions and further expressing them using terms and phrases defined in the Hourglass Model of Emotions. All the data is incorporated into a Knowledge Graph, where each review is connected to all its related aspects, description words, and their synonyms and emotions. We enhance the results with linguistic terms describing emotions of identified aspects to learn more about the justification behind the determined sentiments of the reviews.