This paper introduces a novel approach for extracting customer preferences from hotel reviews, in order to provide personalized recommendations. We defined a finite set of aspects that we deemed important for occupants. These aspects are then quantified and represented as a vector for each user, capturing their unique preferences. The proposed collaborative filtering system utilizes these vectors to recommend hotels that align closely with the preferences of individual users. By considering the similarity between user vectors, we can effectively match users with hotels that are likely to meet their expectations. This approach offers a more nuanced understanding of customer preferences compared to traditional rating-based systems.