Real estate recommendation systems present a significant challenge due to the multifaceted nature of buyer decision-making. This growing field necessitates the development of a holistic recommendation approach. This study proposes the Est-Model, which captures customer behavior related to real estate purchase decisions. The Est-Model leverages a relational knowledge model to organize property features. This model is then integrated with a database containing user profiles and real estate information. To address the cold-start problem for both items and users, the study introduces a novel real estate recommendation methodology. Utilizing a session-based recommendation framework, the system effectively manages sequential and contextual data to predict user preferences for future interactions. Empirical evaluations demonstrate the superiority of this approach over existing methodologies within the top-n recommendation paradigm. These findings are validated through experiments conducted on a real estate dataset.