View article

Nearest neighbor search, a problem that asks for a nearest point in the database given a query, is a problem in areas of computer science such as information retrieval, pattern recognition, image, and text processing. There is still ongoing research toward the improvement of the performance. Despite that, there is no comparison between methods proposed. We have no idea whether a method could work in high dimension or not, or how about the time and space complexity, or what about the results returned by specific method. In this paper, I compare six of them here, they are Exhaustive, Vantage Point, Random Projection Matrix, Random Projection Tree (RP Tree), Random Ball Cover (RBC), and Locality-Sensitive Hashing (LSH). This paper consists of comparisons between each method mentioned so that readers could see some characteristics of these methods to solve nearest neighbor search problem. In this paper, I use two distance metrics, that is Euclidean and Cosine Similarity, to compare the performance of each method.