Multidimensional Management of Geospatial Data Quality Information for its Dynamic Use Within GIS

Metadata should help users to assess the quality (fitness for use) of geospatial data, thus reducing the risk of data misuse. However, metadata presents limitations and remain largely unused. There still exists a need to provide information to users about data quality in a more meaningful way. This research aims to dynamically communicate quality information to the users in a rapid and intuitive way in order to reduce user meta-uncertainty related to geospatial data quality, and then reduce the risks of data misuses. Such a solution requires a data model able to support heterogeneous data quality information at different levels of analysis. Using a multidimensional database approach, this paper proposes a conceptual framework named the Quality Information Management Model (QIMM) relying on quality dimensions and measures. This allows a user to easily and rapidly navigate into the quality information using a Spatial On-Line Analytical Processing (SOLAP) client-tied to its GIS application. QIMM potential is illustrated by examples, and then a prototype and ways to communicate data quality to users are explored. Introduction The context in which geospatial data is used has changed significantly during the past decade. Users now have easier access to geospatial data and GIS applications, especially through the internet. As they were formerly almost restricted to geospatial experts, it is now frequent that users with a limited expertise in the geospatial domain use geospatial data and GIS applications. Although this is a positive evolution in general, one problem has emerged: today’s typical geospatial data users have less knowledge in the geographical information domain (Agumya and Hunter, 1997; Aalders and Morrison, 1998; Curry, 1998). Consequently, their knowledge about risks related to the use of geospatial data is limited (Goodchild, 1995; Agumya and Hunter, 1997; Curry, 1998; Elshaw Thrall and Thrall, 1999). In that sense, Goodchild (1995) argues that “GIS is its own worst enemy: by inviting people to find new uses for data, it also invites them to be irresponsible in their use”. This sometimes leads to faulty Multidimensional Management of Geospatial Data Quality Information for its Dynamic Use Within GIS Rodolphe Devillers, Yvan Bédard, and Robert Jeansoulin decisions based on these data, possibly having significant social, political or economical consequences which is being discussed in the literature (Beard, 1989; Monmonier, 1994; Curry, 1998; Agumya and Hunter, 2002; Gervais, 2004). In order to reduce the risks of misuse, geospatial data producers spend a lot of resources documenting their datasets to inform the users on dataset specifications and quality. Among these documents, metadata (i.e., data about data) provide information on several aspects of the datasets, such as, data producer identification, spatial reference systems, lineage, definition of features or attributes, and data quality (FGDC, 2000; ISOTC/211, 2003). However, metadata are defined in the literature as producer-oriented offering only limited benefits for the users who want to assess the fitness of the data for their use (Frank, 1998; Harvey, 1998). In fact, experience shows that metadata do not reach their information goal for non-expert users and are also difficult to understand by many expert users (Timpf et al., 1996; Frank, 1998; Harvey, 1998). Understanding and reaching conclusions that could be used legally, for example, about the quality of geospatial data rapidly becomes an unmanageable task when one wants to take into consideration the spatial, temporal, thematic, acquisition, and other heterogeneities found in a dataset. Consequently, metadata related to data quality usually remain unused by non-expert as well as by experts, even with the best datasets, leaving users in a state of ignorance about the characteristics of the geospatial dataset being used. As demonstrated by Gervais (2004), an increasing number of geospatial data is intended for general public and must follow legal requirements related to a mass-product category. Metadata, as currently provided or defined within international and national standards, do not reach these obligations, especially concerning the requirements of providing easily understandable information as well as information about potential risks of misuse. According to Gervais (2004), there is a need for a computerized instruction manual that would reduce the risks of misuse by providing to the users of geospatial data information that is easier to understand. Several authors highlighted the need to design such a tool, sometimes identified as “Quality-aware GIS,” “Quality GIS,” or “Erroraware GIS” that would dynamically take quality information into consideration during data manipulation (visualization, queries, and updates) in order to prevent the user from “illogical operations” (Unwin, 1995; Hunter and Reinke, 2000; Duckham and McCreadie, 2002; Qiu and Hunter, 2002). PHOTOGRAMMETRIC ENGINEER ING & REMOTE SENS ING Feb r ua r y 2005 205 R. Devillers and Y. Bédard are with the Centre de Recherche en Géomatique (CRG), Université Laval, Québec, G1K 7P4, Canada, ([email protected]; yvan.bedard@ scg.ulaval.ca). R. Jeansoulin is with the Laboratoire des Sciences de l’Information et des Systèmes (LSIS), CMI, Université de Provence, 39 Rue Joliot Curie, 13453 Marseille Cedex 13, France ([email protected]). Photogrammetric Engineering & Remote Sensing Vol. 71, No. 2, February 2005, pp. 205–215. 0099-1112/05/7102–0205/$3.00/0 © 2005 American Society for Photogrammetry and Remote Sensing 04-073.qxd 1/11/05 1:39 PM Page 205