Bridging the Gap between Spatial Data Sources and Mashup Applications

Abstract

Utilizing online spatial data sources to create added values has been quite common in modern Web applications. Through client-side mashup techniques, one can efficiently integrate some popular spatial data services (e.g., Google Maps) through their well-defined interfaces as well as useful tools for mashup. However, many other spatial data providers lack of resources or motivations to provide such rich data services like Google Maps. Instead, they may provide only limited service functionalities, such as static files download only. Furthermore, their data formats and interfaces are vastly heterogeneous. This introduces many more difficulties in data integration, especially for spatial vector data, to which the data accesses often require queries with spatial predicates. Moreover, they may not guarantee system performance in responding client requests. Therefore, all these create a gap between het-erogeneous spatial data sources and mashup applications. To address the problem, we envision a server-side spatial data mashup platform that can provide a unified interface with rich data access functionality on top of these heterogeneous spatial data sources. This paper presents the architecture and a proto-type of such a data mashup platform for spatial vector data specifically. In addition to the typical on-the-fly approach of mashup, the platform can also preload data from data sources with limited system capacities to provide more controllable performance. We demonstrate the effectiveness of this platform through an example web application accessing the integrated data from the platform. This paper further evaluates the system performance and shows the performance tradeoffs of deploying this server-side platform.