A Fine-Grained Ontology-Based Sentiment Aggregation Approach

Sentiment analysis techniques are widely used to capture the voice of customers about different products/services. Aspect or feature-based sentiment detection tools as one of the sentiment analyses' types are developed to find the customers' opinions about various features of a product. However, as a product may contain many features, presenting the final obtained results to the users is a challenge. Even though this issue is addressed in the literature by developing different sentiment aggregation methods, their results are mostly presented at the basic-level features of a product. This may cause in losing customers' opinion about at minor sub-features. However, as the performance of a basic feature is dependent on those of its different sub-features, we propose an approach which aggregates the extracted results at a fine-grained level features using a product ontology tree. We interpret the polarity of each feature as a satisfaction score which can help managers in investigating the weaknesses of their products even at minor levels in a more informed way.