** A spatial similarity ranking framework for spatial metadata retrieval
Recently there has been a trend to facilitate access to large government repositories of spatial data 1. In addition, there are an increasing number of non-government spatial providers who provide access to large quantities of spatial data for little or no cost. These factors have promoted research into improving metadata standards and metadata retrieval.
Unfortunately, the current methodology has mainly focused on improving the textual description and subsequent textual query matching retrieval frameworks for spatial metadata. Consequently, the current metadata standards only include spatial extent and spatial reference information. This has resulted in the current standards having limited spatial querying functionality.
Spatial data by definition is spatial aware, therefore, the spatial component should be exploited as much as possible to allow more complex spatial querying of spatial metadata. This paper presents an extension to the existing ISO metadata schema for Geographic Information and details a framework that ranks the spatial similarity between a query and spatial metadata. The framework utilises an object frequency (of) and inverse spatial frequency (isf) which is incorporated into the spatial metadata schema. This novel methodology is based on the tf-idf method used for text-based information retrieval. A Bayesian inference retrieval engine is utilised to rank the similarity between spatial query and metadata.
This contribution will allow complex spatial queries of spatial metadata. The spatial metadata will be ranked by spatial similarity which will improve the performance and efficiency of the spatial retrieval process.
1. For example, the Australian Spatial Data Infrastructure framework and ANZLIC metadata project aim to improve access to Australia’s spatial repositories.