Survey of Fast Nearest Neighbor Search

A spatial query takes a location and given keywords as arguments and returns objects that are ranked according to both spatial proximity and text relevance relative to the query. Spatial queries like nearest neighbor retrieval and range search, occupy only conditions on geometric properties of object. For finding objects which are satisfying a spatial predicate and predicate on their associated texts, novel form of queries are called by many applications. Consider situation of retrieving a nearest neighbor query will call for all nearest restaurant whose menu list contains butter chicken, biryani, pulav, without calling all the restaurants nearest to it. At present, IR2-tree is the best suited solution for such queries. Efficiency of IR2-tree badly is impacted because of some drawbacks in it. The solution for overcoming this problem should be searched. Spatial inverted index is the access method which will be solution for this problem. Spatial inverted index extends the conventional inverted index to handle the multidimensional data. To deal with nearest neighbor queries with keywords, it has algorithms which will handle with those queries. Spatial inverted index do better than the IR2-tree by using a feature of orders of magnitude, in query response time appreciably.