Empirical tests using search engine optimisation techniques to compare the effectiveness of two metadata vocabularies for geospatial data discovery on the Web

Abstract

With the increasing use of geographical information science and technology in a variety of knowledge domains and disciplines, the need to discover and access suitable geospatial data is imperative. To date, spatial data infrastructures (SDI) implemented at local, national, regional or international levels provide frameworks through which geospatial data is disseminated and shared between providers and consumers. Described as the most visible part of SDIs, geoportals are online web platforms for searching and discovering geospatial data and associated metadata. However, besides being known only to people in geoinformation communities, current geoportals as entry points to SDIs present some technological issues that make it difficult for web search engines to discover and index geospatial data catalogued in (or registered) with a geoportal. This hinders the visibility of geoportals and ultimately the discovery of geospatial data. Most geoportal implementations make use of web catalogue services for the registration and discovery of geospatial resources, making the catalogue opaque to web search engines. In this study, metadata about geospatial resources was published on the Web in the form of HTML web pages. Empirical tests with search engine optimisation techniques compared Dublin Core to Schema.org in order to evaluate web search engines' retrieval effectiveness. The process of constructing HTML pages was guided by a chain of mappings from a taxonomy of search terms to the ISO 19115:2003, Geographic Information – Metadata standard, to the Dublin Core metadata standard and finally to the Schema.org vocabulary. The taxonomy of search terms was constructed from a qualitative and quantitative analysis of search terms employed by users who were instructed to search for geospatial data online using web search engines of their choice. This taxonomy helped to understand the kind of terms that users employ when searching for geospatial data on the Web. The terms in the taxonomy were used to describe geospatial resources (metadata) in the HTML pages. Two sets of HTML web pages were registered with the Bing and Google web search engines. One set was marked-up with the Dublin Core vocabulary, and a second set was marked-up with the Schema.org vocabulary. The number of HTML pages retrieved and the average position of pages on the search results pages of the two web search engines were analysed using descriptive and inferential statistical techniques. Results show that metadata about geospatial resources contained in HTML web pages marked-up with Dublin Core and Schema.org vocabularies are discoverable on both search engines (Bing and Google). Overall, the results showed that Google was more effective in retrieving these HTML web pages as compared to Bing. HTML web pages marked-up with Schema.org vocabulary were effectively retrieved as compared to HTML web pages marked-up with Dublin Core. Furthermore, the test statistics results were significant in most of the tests performed. The user study resulting in a taxonomy of search terms presents an approach for understanding search terms when searching for geospatial data on the Web. The mapping from the taxonomy to ISO 19115:2003, Geographic Information – Metadata, and from Dublin Core to Schema.org vocabulary as discussed and proposed in this research demonstrates a process for embedding and encoding geospatial metadata in HTML web pages for optimum web search engine retrieval effectiveness of geospatial data.