This seems more like a computer science question to me. It can be thought of as a k-nearest-neighbour problem. The exact details of his problem could be expres

This seems more like a computer science question to me. It can be thought of as a k-nearest-neighbour problem. The exact details of his problem could be expressed as a metric. What he is asking for is naively trivial; compare the target to every object and see which ones are closest. If you want decent performance with large datasets, however, you would be better off using a more sophisticated approach. See http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm for more details.