期刊名称：International Journal of Advanced Computer Science and Applications(IJACSA)
印刷版ISSN：2158-107X
电子版ISSN：2156-5570
出版年度：2020
卷号：11
期号：4
DOI：10.14569/IJACSA.2020.0110483
出版社：Science and Information Society (SAI)
摘要：In a vector set, to find similarities we will compute distances from the querying vector to all other vectors. On a large vector set, computing too many distances as above takes a lot of time. So we need to find a way to compute less distance and the MTree structure is the technique we need. The MTree structure is a technique of indexing vector sets based on a defined distance. We can solve effectively the problems of finding similarities by using the MTree structure. However, the MTree structure is built on one computer so the indexing power is limited. Today, large vector sets, not fit in one computer, are more and more. The MTree structure failed to index these large vector sets. Therefore, in this work, we present a novel indexing method, extended from the MTree structure, that can index large vector sets. Besides, we also perform experiments to prove the performance of this novel method.
关键词：MTree; DMTree; spark; distributed k-NN query; distributed range query
