Leverage Google Maps API to Perform K-Dominant Region Query
A popular search on the Internet is to retrieve the top-k spatial objects (e.g., restaurants) whose location and description are most relevant to a reference point and a set of keywords. When an object is returned, we say the object is k-dominant on the reference point with the given set of keywords. Our research is interested in leveraging existing search engines such as Google Maps to process k-Dominant Region (k-DR) queries. Given a spatial object and a set of keywords, a k-DR query returns the region which contains all points where the object is k-dominant with the keywords. Naïve solutions such as performing top-k queries on all possible reference points may not work, because search engines usually allow users to perform a limit number of queries every day. Our research has developed an approach that can process k-DR queries by having Google Maps to perform exhaustive searches. Our approach is based on the monotonicity properties of geo-proximity and relevance of the textual description to the search keywords. These characteristics make it possible for us to estimate the ranking function used in a search engin and then use the function to process k-DR queries. We have implemented the proposed approach on Google Maps. In this talk, we will present the implementation and examine the performance results.
Committee: Ying Cai (major professor), Simanta Mitra, Jin Tian