Modified Indexing Algorithm based on Priority Queue in Metric Space for MVP Tree

Abstract

The Internet of Things (IoT) algorithms process huge amounts of heterogeneous data in real-time. One of the most computationally intensive tasks using cloud technologies is the task of clustering and classifying data. The authors propose to develop an approach to data classification within the “Query by Similarity” paradigm, which uses the technology of data indexing based on Metric Access Methods (MAM). To improve the performance of data indexing, this paper proposes a similar nearest neighbor search method combining a multiple vantage point tree (MVP) and improved algorithms for processing the priority queue of nodes. The following two algorithms for processing the priority queue of nodes were developed: 1) algorithm for all kinds of points-queries, which makes it possible to take into account parent nodes of all higher levels; 2) algorithm for grouped based on clustering of points-queries by reusing previously obtained search results. Experimental results confirm the effectiveness of the proposed approaches and algorithms.