Contextual data plays a vital role in designing effective Chatbots that offer personalized interactions to users, particularly in tourism industry. Success of such systems relies heavily on the accurate modeling of contextual data, encompassing user behavior, location, time, weather, and other dimensions. However, the dynamic, incomplete, and noisy nature of contextual data, coupled with diverse range of resources used for its capture, poses challenges to achieve precise modeling. Motivated by the above challenges, this review paper aims to explore various approaches to contextual modeling, examine their strengths and limitations, and delve into the multidimensional aspects of the context within the tourism domain, specifically to focus on the architecture of contextual systems, notably proactive Chatbots. Moreover, this research paper underscores the significance of a comprehensive architecture in context-aware systems and proposes future research directions to address the challenges associated with contextual modeling.