Automatic Metadata Generation for Geospatial Resource Discovery

Introduction

The concept of metadata is hardly new—the most common definition of the term metadata is “data about data,” with the first references to this term appearing in the context of geographic information, in ANZLIC (1996) and Kildow (1996). If we look for the origins of the term metadata, we will find its roots in the Greek word “μετα,” “beyond” and the word “data,” the plural of the Latin term datum-i, “piece of information” (RAE, 2011¹). Therefore, the meaning of the word may be explained as “beyond data.” However, according to Howe (1993), the term metadata did not appear in print until 1973, despite having been coined by Lack Myers in the 1960s in order to describe sets of data and products. In the literature related to this subject we find a good number of authors who provide the interpretation and scope of the practical and theoretical meaning of the term. Among these, we find Caplan (1995), Milstead and Feldman (1999), Ercegovac (1999), Sheldon (2001), and Steinacker et al. (2001), Swick (2002), and Duval et al. (2002), or Woodley et al. (2003). Summing up the contributions of all these authors, we may define the term eclectically as the structured set of data that describe other data and whose purpose is to improve our knowledge of the described information and help us answer such questions as ‘what,’ ‘who,’ ‘where,’ ‘when,’ ‘how much,’ and ‘how.’ They may also be described as those autonomous products that, linked to the data, allow us to keep an inventory of these, enabling its publication and reference value through the catalogues kept in SDI and, finally, allowing for the reutilization of data. The importance of metadata has been recognized by entities such as the EU’s INSPIRE² Directive, and also by the endorsements of the GSDI³ initiative.

Moreover, Caplan (1995) acknowledges that the concept of metadata is used to avoid the prejudices developed by professionals in the field of information, who are closer than most to the world of libraries: computer technicians, software designers, and system engineers. Finally, metadata are used to describe the context, the quality, the condition or the characteristics of the data (Milstead & Feldman, 1999; Howe, 2003) in such a way that users can discover and understand their data sets, particularly in the context of Geographic Information (GI). For Zeigler et al. (2006), metadata is “a hierarchical concept in which metadata are a descriptive abstraction above the data it describes.”

Various experts are in favour of assigning the task of metadata creation to the owners of the geospatial datasets (geodata), in the belief that these owners are best suited to provide information about their data (Greenberg, 2004; Kolodney & Beard, 1996). In practice, metadata creation has occupied a secondary role within organizations, having been created after its production. For this reason, some organizations have considered the creation of metadata as an additional cost (Najar, 2006). This fact has been criticized by several studies; for example, in the CGIAR-CSI (2004) study we find the following statement: “The creation of metadata to novel data producers might seem burdensome, but the long term advantages are far superior to the disadvantages of the initial burden of implementing a Metadata policy within an organization. The initial expense of documenting data clearly outweighs the potential costs of duplicated or redundant data generation.”