Hybrid-Based Recommender System Utilizing Ontology for Semantic Modeling in E-Commerce

Recommendation algorithms improve the functionality of online shopping platforms by assisting buyers in selecting the best products depending on their interests. In essence, recommender engines are a component of an online personalized strategy that enhances user experience via dynamically loading different types of content into emails, apps, and websites. This paper focuses on investigating and putting into practice data filtering techniques that are commonly used in recommender systems. These techniques include semantic-based filtering, hybrid filtering, collaborative filtering, content-based filtering, graph-based filtering and ontology-based filtering. In order to understand these methods and how they operate, a brief discussion of the relevant state-of-the-art research will be conducted. Next, this paper will then dive deeper into various semantic-based recommendation techniques to build an ontology for modeling semantic data and relationships. Ontology is simple to expand because the relationships and concept can easily be matched and added to the existing corpus. Furthermore, ontology also offers the ability to represent all data types, including structured, unstructured, or semi-structured, thus, facilitating more seamless data integration. Additionally, a dashboard will be created for the end-user and administration to visualize the descriptive analytics of the buying pattern. The evaluation metrics employed are root mean squared error (RMSE) and mean absolute error (MAE). Then, the results will be presented through data visualization techniques, including graphs, charts, and some advanced visualization tools. Last but not least, a dashboard will also be created in this project for the end-user and administration to visualize the descriptive analytics of the buying pattern.