A GML Compression Approach Based on On-line Semantic Clustering

Geography Markup Language (GML) has become a de facto international encoding standard for exchanging geospatial data among heterogeneous Geographic Information Systems (GIS). Whereas, structurally redundant tags and textual data representation usually inflate the sizes of GML documents substantially, which makes the storage and transport costly. In this paper, we propose an effective compression approach based on on-line semantic clustering of GML documents. The approach deals with a GML document under compression on the fly via separating data from structures, clustering data based on the semantic similarities exploited from tags and texts, dictionary-encoding structures and delta-encoding geometric coordinate data before the general text compression on back end. We conduct extensive experiments on real GML documents to evaluate the performance of the proposed approach. Results show that our approach outperforms the most popular general text compressor gzip, the acknowledged best XML compressor XMill, and the first and up to now the only GML compressor GPress in compression ratio.