This framework analyzes the digital map images in offline processing to extract the enclosed information and construct an attributed map graph. The user input is processed online and the search is performed according to a graph-matching algorithm,7,8 which allows both full and partial matching of the sketched query. This framework represents a user-centric approach. Users who have visited an unknown part of a city typically can’t remember street names and numbers or specific information about POIs, and have only descriptive information about a desired destination, which could be expressed through geographical (sketch), semantic (POI combinations, such as a parking lot near a restaurant), and limitation attributes (a specific city area). Our scheme is designed to support these query types.