Efficient Dictionary and Grid-Based Framework for Answering Durable k-Nearest Neighbor Queries on Time Series Data

With the massive amount of data generated, a data engineering method is needed to process the data to obtain new information, which can later be used for further understanding and business decisions. One of the data processing methods is the k-nearest neighbor (k-NN) query. The k-NN queries aim to identify k-objects with the shortest distance based on object references over the network. However, a critical limitation of the k-NN query lies in its inability to process a query when an object has proximity durability to a reference object at a particular time interval. This limitation arises since the k-NN query does not consider the durability and time interval variables in its calculation algorithm. This article discusses the problem of time interval-based durable k-NN queries on time series data by introducing the algorithmic framework and employing dictionary and grid data structure to process these queries. The proposed algorithm, which is a dictionary and grid-based durable k-NN queries method, incorporates parallel computing techniques to efficiently search for a collection of k objects that exhibit durability in their proximity to a reference object within a predefined time interval. The effectiveness and efficiency of the devised algorithms were assessed through extensive testing using both real-world and synthetic datasets.