Line Simplification for Efficient Approximate Join Queries On Big Geospatial Data
Fatima Al-Hammadi, Haya Almadhloum Alsuwaidi, Shooq Abdelrahman Alzarooni, Isam Mashhour Al Jawarneh · 2024
Line simplification algorithms are often used to render high-resolution geographic features at appropriate resolutions when applied to polygons. They are generalization techniques in which selective vertices are removed from a line feature to eliminate details whilst preserving the line’s basic shape. In this paper, two different line simplification algorithms (Douglas-Peucker and Visvalingam-Whyatt) are used in conjunction with spatial join of geo-referenced mobility and air quality data, to reduce the size of the polygon files. A filter-and-refinement dimensionality reduction-based approach is then used to join the data. This framework allows for an optimized spatial join on an integrated schema through a state-of-the-art filter-and-refine based approach. The reduced files can then be used in geospatial related data science tasks such as DBSCAN, clustering, and regression at lower computational costs. Our experimental results show that incorporating a reduction approach such as line simplification before performing the spatial join can significantly reduce the computational cost and improve the performance with the number of vertices reduced by 94% after simplification and accuracy, MAPE is minimized with a low score of 0.048 and 0.049 for DP-Map shaper and VW-Map shaper respectively.