Different people have different prior knowledge of a certain environment, which makes it hard to provide a geo-related service (e.g. navigation) that fits all. In order to provide a service adapted to individual's spatial knowledge, a model of individual's prior knowledge of the environment is needed. The growing popularity of location aware social media provides a unique opportunity to study individual's spatial knowledge. With Foursquare being one of the most popular location-based social media, this thesis focuses on modeling individual's familiarity of places by using Foursquare data. To achieve this overall goal, two objectives are set. The first objective is to identify individual's meaningful places. To derive meaningful places from the check-ins on Foursquare, which represent the frequency of visits to a place, an appropriate clustering algorithm is required. A comparison of four existing clustering algorithms (SLINK, K-means, DBSCAN and EM Algorithm for Gaussian Mixture Model) was then conducted. DBSCAN turned out to have the best overall performance. To attain the second objective, which is to model individual's familiarity of places, information indicating the affected responses of an individual, e.g., the text descriptions and photos along with a check-in, was added to weight each check-in and eventually to measure the familiarity. To evaluate the modeling framework, an online survey was carried out. The results demonstrated the possibility to model individual's familiarity of places using social media.