Special section in honor of Andrew U. Frank

1. Introduction

Andrew Frank, who retired in September 2016 from the chair of geoinformation position at the Vienna University of Technology that he had held for a quarter century, has had anextraordinary influence on our field. His curiosity and wide spanning intellectual interests enriched the field’s foundations and propelled it in many ways. To name just one example, he conceived and shaped the Conference on Spatial Information Theory, COSIT (Frank et al. 1992, Frank and Campari 1993, Frank and Kuhn 1995), inspired by the success of the NATO Advanced Study Institute on Cognitive and Linguistic Aspects of Geographic Space that David Mark and Andrew Frank convened in 1991 (Mark and Frank 1991). This special section of IJGIS in his honor provides a multidisciplinary reflection on his career, including views from computer science, ontology, economics, law, cognitive science and linguistics, indicating the breadth of Andrew Frank’s interests and contributions.

2. Areas of contributions

Andrew Frank’s research interests are exceptionally broad and his contributions to our field fall into many distinct areas, ranging from spatial databases (including access methods) and spatial data models through cognitive and linguistic aspects to ontology and economics of geographic information. Moreover, he has pioneered new research directions that subsequently became mainstream. While we cannot do justice to the breadth and depth of his work in a special section, much less in the editorial for it, we attempt to highlight the main areas in the following paragraphs, giving more weight to his earlier work, as it may be less familiar today.

During his doctoral studies, Andrew Frank already had a paper on Database Management Systems (DBMS) for cadastral (a.k.a. land information) systems (Frank 1981) application accepted at the Conference on Very Large Data Bases, as well as a paper on spatial query languages (Frank 1982) at the ACM Computer Graphics Conference (SIGGRAPH). During that period, Frank was the intellectual primus inter pares in the research group of the late Rudolf Conzett, a visionary who shaped the scientific style and rigor of his students and had major impact on Geographic Information Systems (GIS) research and development in Europe and beyond. He entrusted Andrew Frank with the research design and with the prototypical implementation of a radically new information system architecture for parcel-based geographic data. With the help of his team colleagues, Andrew Frank designed and implemented a network DBMS to handle cadastral data and support legal, administrative, and planning processes based on them. This database system (Pascal Network Database) came with higher-level graph-based data models (one of the first formalizations of graph models for spatial data and their semantics), a graphical query language (MapQuery) and a sketch-based construction language (Human Interface to Least Squares), as well as an access method (FieldTree). Many of these innovative developments influenced other research groups and key industries worldwide. Frank ended up playing no small role in making spatial data models (and object-oriented design) widely acknowledged specializations of non-standard database systems. He continued to push for research on spatial databases in the successful bid for NCGIA (the National Center for Geographic Information and Analysis) a few years later and, together with Terry Smith, led an NCGIA Research Initiative on the Architecture of Very Large Spatial Databases. Although not formally a co-editor of the proceedings of the first Symposium on Large Spatial Databases, he was the mastermind behind the creation of that conference series. A driver of Andrew Frank’s early research agenda was his recognition that GIS require alternative models to those based on standard coordinate geometry, emphasizing, for example, the need for explicit topology. His seminal paper on cell complexes became a critical precursor of the early standards on simple features in OGC and ISO.

Through a series of advanced research meetings and projects at the end of the 1980s and into the early 1990s, Andrew Frank expanded the scope of his research (and that of many colleagues) from spatial to spatio-temporal data modeling. With this expansion, he then became interested in the broader ontological questions underlying GIS: what sorts of things and events in the world do the spatial and temporal data represent?

On the geometry side, as a complement to the formalization of binary topological relations, Andrew Frank saw that a formalization of cardinal direction relations could play an important role in spatial query languages and spatial reasoning. The development of the calculi for point-to-point directions that follow a half-plane or a cone model brought together a series of his skills. The concise specification in an algebraic form in combination with a prototype implementation that demonstrated that specifications were correct and enabled immediate automated inferences blended themes from Artificial Intelligence and software engineering to address a fundamental spatial problem. The integration of qualitative distances into the direction specifications established both more interesting and complex inferences, but also highlighted the difference in reasoning when a single spatial concept is employed vs. multiple spatial concepts in concert.

Andrew Frank’s interest in different kinds of errors, their causes, and how to live with errors has its roots in the measurement sciences, error propagation, and adjustment computations. Under the umbrella of indeterminate boundaries, Andrew Frank integrated also his interests in economic aspects with the topic of spatial uncertainty modeling. His interest in databases and query languages naturally evolved into a focus on cognitive aspects of space and spatial information.

3. The papers in this section

The first paper in this section, ‘Andrew U. Frank’s impact on research in land administration,’ is written by Gerhard Navratil. Among all the authors of this section, Navratil is the only one who was Frank’s doctoral student. Navratil’s work on land administration and cadastral systems, in particular the Austrian cadaster, has been inspired by an original interest in data quality. His interest is in the formal instruments of a cadastral system but behind may be the search for accurate representations of land rights, restrictions, and responsibilities. Accordingly, his review of Frank’s work in this area focuses on Frank’s approach by algebraic modeling that was not only fundamental, but also inspiring for many, since it allowed a formal and testable specification.

The second paper, ‘Formal representation of qualitative direction’, is written by Jasper van de Veen, Christian Freksa, and Diedrich Wolter. Freksa and Frank are connected by a common interest in formal spatial reasoning methods, in particular on directions, where both have published landmark research on complementary topics – Freksa on temporal reasoning as well as relative directions (Freksa 1992a, 1992b), and Frank on cardinal directions (Frank 1996). They have then collaborated in the International Quality Network on Spatial Cognition and introduced generations of researchers in artificial intelligence to cognitively motivated formal thinking about qualitative spatial relationships and reasoning. The paper proposes a taxonomy of spatial direction representations from different research strands in the field.

The third paper, ‘The world, the computer, and the mind: how Andrew Frank helped make human language and cognition cornerstones of geographic information science’, is written by Daniel Montello and David Mark. Montello is a psychologist working in a geography department and interested in behavioral and cognitive geography (e.g. Montello 1993, Lovelace et al. 1999). Mark is a human geographer with roots in physical geography and a strong interest in language and ontology (e.g. Mark and Frank 1991, Egenhofer and Mark 1995, Smith and Mark 2001). Two NCGIA research initiatives were led jointly by David Mark and Andrew Frank: one on user interfaces (Mark 1995), and the other one on languages of spatial relations (Mark and Frank 1992). This collaboration also led to the Las Navas workshops cited above, which have recently been revisited (Frank et al. 2013). The paper in this section, by Montello and Mark, looks at Frank’s fundamental contributions to a cognitive geographic information science.