Evaluating the Performance of Graph-Based Recommendation Systems

Today, recommendation systems are considered a main component of social media platforms and many other online websites. Recommendation systems can be defined as tools that aim to introduce and suggest products to users. The suggestion process depends on many factors, such as user behavior and product similarity. In recent years, many research papers have discussed recommendation systems and introduced new solutions and methods to build them.

On the other hand, in the last few years, data representation has also become an important issue. Because of the massive increase in data, new methods to represent data have been introduced and adopted, such as graph-based data representation. In this work, the efficiency of employing graph-based databases in building recommendation systems was evaluated and compared to traditional approaches.. Specifically, Amazon Product Reviews dataset was used to build a recommendation system using traditional methods. This data was then transformed to a graph format and used to generate recommendations. Metrics such as accuracy, recall, and precision were adopted to determine the efficiency and accuracy of the results, as will be discussed later.