Sinirsel işbirlikçi filtreleme tabanlı çok kriterli otel tavsiye sistemi

Nowadays, users are offered numerous hotel alternatives available online to meet their accommodation demands. Although this provides many advantages regarding accommodation planning and the opportunity to evaluate and decide on many alternatives online, it also presents various difficulties. Due to the abundance of information available, the process of deciding on a hotel might be demanding for users, resulting in spending an excessive amount of time and potentially making poor choices. To assist users in making better decisions, online booking platforms need to embrace innovative solutions to manage this information overload. The existence and increase of different platforms allow users to choose and evaluate, but at the same time, due to the constant increase in information, it has evolved to a point that can negatively affect the decision-making process. Instead of evaluating hotels based only on their overall score, including different criteria in the recommendation system will support the personalized recommendation process. This thesis presents a multi-criteria neural collaborative filtering hotel recommendation system to address this need. The proposed model utilizes user and hotel identifiers as inputs to estimate multi-criteria ratings, then estimates the overall rating based on the predicted ratings to generate recommendations. The single-criterion model, which does not use multi-criteria ratings, and a conventional multi-criteria recommender system, which does not predict multi-criteria ratings but instead uses them as input, were both used to test the proposed model. These tests demonstrated that the proposed model outperforms other models, showcasing the superiority of multi-criteria decision-making over single rating systems and its significant impact on overall ratings.