A self-index GML storage approach based on element coding

Geography Markup Language (GML) has been widely used in various application systems as internal geo-spatial data expressing. GML, semi-structured XML-encoding data, behaves low query performance, for full traversal is needed to locate target elements. Indexing GML by pre-sorting search data can improve query performance. Spatial index of GML is necessary as well as non-spatial index for the feature of GML, especially in location-aware query. We design element coding extending interval coding, which separates spatial nodes from non-spatial nodes, and elements from GML model tree to generate sequence of element coding. The paper proposes GML index integrating B+-tree and R-tree as entrances of non-spatial and spatial query on element coding sequence stored as key-value structure, and is also a self-index storage approach that can completely replace original GML document. The storage, index and query optimization methods are also introduced in this paper. The proposed storage mechanism proved to be effective by the result analysis of storage and query experiments.