An Efficient Similarity Search Approach to Incremental Multidimensional Data in Presence of Obstacles

Abstract

In data mining field similarity search has always been a crucial task. A similarity search finds the data points from the same data set space that matches the given query sequence exactly or differs slightly, and is done for whole sequence matching or partial sequence matching. In data sets the existence of obstacle information greatly affects the performance of similarity search in terms of efficiency and effectiveness. Thus, in this paper we present an efficient approach to similarity search based on dynamic selection of input features or attributes in presence of obstacles in respect to better running time and accuracy, with the incremental multidimensional data set. The results show that performance of the similarity search is highly dependent on data size. Thus, our approach can improve the data analysis of financial market, engineering and scientific databases, and telecom industry, providing better performance of classification, clustering, machine learning, and medical diagnosis.