Visualization of geospatial metadata for selecting geographic datasets

Abstract

The importance of geospatial metadata that describe geographic data resources has grown along the increase of supply and use of geographic data. Discovery services have been developed on geospatial metadata systems, but supporting tools for users evaluating the suitability of geographic datasets for an intended use are lacking. This dissertation studies how different visual representation forms of metadata could support users in selecting suitable geographic datasets.

Maps are a traditional representation form of geographic data, and therefore, maps representing samples of datasets were considered in this research. In addition, multivariate visualization methods were of interest, as they enable users to study and compare multiple characteristics of multiple datasets at a time. These visual representation forms were tested with professional users of geographic data in two stages. First, static displays were used to test the design concepts. Successful concepts were implemented in an interactive prototype that included sample maps, a parallel coordinate plot and star glyphs as multivariate visualization methods, and textual metadata. A user test was carried out with this prototype. Users' thinking protocol was applied in both tests.

The findings of the research suggest that sample maps, when displayed individually, provide little concrete support to users of geospatial metadata. However, comparison of sample maps of different datasets stimulates users of geospatial metadata in questioning the data quality and searching for detailed descriptions of datasets.

A parallel coordinate plot proved to be a valuable method to users who need to compare datasets by aspatial metadata elements. Already six alternative datasets in the prototype test indicated its usefulness, and the value of this method increases when the number of alternative datasets rises. The test results suggest that a parallel coordinate plot is easily adoptable to occasional users of geospatial metadata. However, the thinking aloud of the subjects confirmed that users need support for understanding the concepts and special terminology used in geospatial metadata.

Although the textual form of metadata alone does not support comparison and evaluation of geographic datasets, it proved to be an essential form in the selection process when confirming or checking details of metadata. As a whole, use patterns of the different representation forms during the process varied between the subjects. The process itself was iterative: the subjects refined the evaluation criteria while gaining insight of metadata.

The research results strongly indicate that tools for visual representations of geospatial metadata are useful in metadata services. The different representation forms should provide linked views to metadata in order to support users' thinking. The proposed visualization of metadata includes both presentational and exploratory characteristics.