The increasing number of new students and workers migrating to other cities has led to a growing need for temporary housing, particularly boarding houses. This condition highlights the need for a recommender system that can help users find accommodations that match their preferences, especially regarding the facilities offered. Although similar studies have been conducted in the restaurant and hotel sectors, existing approaches have not fully captured user preferences contextually, such as facility descriptions in boarding houses. This study proposes a boarding house recommender system based on content-based filtering, utilizing Word2Vec vector representations and Cosine Similarity. Data were collected through web scraping from the Mamikos platform. The Word2Vec model was trained to form a semantic representation of the available facilities, which was then used to calculate the similarity between boarding house units. System evaluation was carried out using precision, recall, and F1-score metrics. The evaluation results show that this approach produces relevant recommenders, achieving a precision of 0.9636, recall of 0.7763, and F1-score of 0.8600. Compared to the TF-IDF-based method, the purposed system demonstrates a more balanced and contextual performance. These findings indicate that Word2Vec is effective in capturing the semantic meaning of boarding facilities and excels in providing accurate recommenders even without relying on large amounts of rating data.