Neo4j-Driven Content-Based Recommender System

With the increase in the applications of recommender systems and the need to make its suggestions transparent and more accurate to actual item relations and user interests, the urge to have a more qualified recommender system has increased. On top of that, the applications of knowledge graphs have proved their superiority and advancements in the recent technological era. Integrating graphs with recommender systems would enhance the scalability, flexibility, and ability to yield more accurate and transparent recommendations. The representation of entities and their relations through a graph demonstrates the many-to-many relationships that can be extracted from it. Those techniques can be utilized to build an efficient and accurate model that meets the needs of the customers and the business. This paper utilizes the Neo4j native graph database to build a content-based recommender system applied to the Amazon cell phone reviews dataset. Along with incorporating some data preprocessing and natural language processing techniques to create more robust features that can represent relations more accurately.