Author: Dante Godolja
Supervisor: Julia Neidhardt, Co-Supervisor: Thomas E. Kolb
Abstract
Content-based restaurant recommender systems use features such as cuisine type, price range, and location to suggest dining options to users. By analyzing the content of restaurants, these systems can generate recommendations. Current research explores ways to improve their effectiveness. In this thesis we explore different ideas on how to build a robust recommender system. Such ideas include text and image processing. For image processing we explore suggesting restaurants based on image color similarity or feature extraction using pre-trained image models. For text processing we explore TF-IDF as a baseline and the state-of-the-art model SBERT. These ideas are then used in a practical case. Results show that with proper preprocessing, TF-IDF can achieve similar scores to SBERT and depending on the scenario even outperform it. However SBERT still provides more novel recommendations than TF-IDF. Depending on the scenario, both models can also be used to make meaningful restaurant recommendations.