SPATIOTEMPORAL PATTERN QUERIES HANDLING WITH UNKNOWN DATA ESTIMATION

• Main Author: Alif Arifin, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/50238
• Institution: Institut Teknologi Bandung

Movement data are generally recorded in discrete form and represent only samples of movement; thus, there is missing data between two sequential samples, which is called unknown data. Spatiotemporal pattern (STP) query enables a query to contain more than one predicates with temporal constraints among them. The existing implementation does not handle unknown data, so that we cannot know the predicate value between two sequential samples. This final project is perfecting the existing implementation by estimating unknown data using interpolation. The STP predicate expresses the STP query that describes the pattern as a set of time-dependent predicates—can be obtained by lifted predicate operation—which satisfies the temporal constraint. We focus on the lifted predicate operation. The lifted predicate operation accepts 4 inputs, particularly a predicate, two entities—spatial or moving object--, and interpolation type. Interpolation is used to estimate unknown data; therefore, the trajectory of the moving object is formed. The trajectory is operated to get the trajectory when the predicate is fulfilled then transformed into time-dependent predicates. The existing implementation is successfully perfected so that we can handle the STP query with unknown data estimation in PostgreSQL. The predicate lifted operation has been able to detect the predicate, which is fulfilled in the middle of an interval. The execution time of the tool is at least 1.25 times slower than the existing implementation. Based on the evaluation, it can be concluded that the extension has succeeded in detecting predicate on unknown data.