SENTIMENT ANALYSIS: AN ENHANCEMENT OF ONTOLOGICAL-BASED USING HYBRID MACHINE LEARNING TECHNIQUES

With the fast development of World Wide Web 2.0 has resulted in huge number of reviews where the consumers share their opinion about a variety of products in the websites, forum and social media such as Twitter and Instagram. For the organizations, they have to analyze customer's behavior to find new market trends and insights. Sentiment analysis concept used to extract the positive, negative or neutral sentiment of the features from the unstructured data of product reviews. In this paper, we explore the techniques and tools used to enhance the ontology-based approach. Combination of ontology-based on Formal Concept Analysis (FCA) which a process of obtaining a formal ontology or a concept hierarchy from a group of objects with their properties and K-Nearest Neighbor (KNN) to classify the reviews. We believe with these techniques, we are able to view the strength and weakness of the product in more detail where the feature selection process will more be systematic and will result in the highest feature set.