An Efficient Nearest Neighbor Search Method for Spatial Keyword Query Processing

This paper shows how to process Generic Location aware Rank Queries (GLRQs) efficiently. A GLRQ contains a spatial object, keywords, a query predicate and ranking function. The rank is calculated based on spatial location, text and attributes. The result contains k objects satisfying the user's requirements. It returns the nearest neighbor of user's current location based on specified query keyword. Instead of considering all hotels, the nearest neighbor query ask for hotels that are nearby and provide services such as gym/spa, pool, internet and also have high health and hotel ratings at the same time. We propose a framework called SKQP (Spatial Keyword Query Processing) to process these types of queries. Three data structures R-tree, Inverted files and Concept tree are used to process the location, keywords (pool, internet) and numerical attributes (health and hotel ratings) respectively. The combined structure is the backbone of SKQP framework. SKQP can process the GLRQs efficiently.