Automated Schematic Mapping: 12 years on Position Paper for Schematic Mapping Workshop 2014 Silvania Avelar ETH Zurich, Geological Institute CH-8092 Zurich, Switzerland [email protected]

Abstract—This position paper addresses current computational and cartographic challenges of automated schematic mapping, taking as its basis the first research achievements on these subjects, published in 2002. Some unresolved problems are listed, which cover the following topics: contextual schematization, multi-scale information, map quality, data repository, complexity of large cities and cognition. Keywords—automated schematic mapping; wayfinding; metro maps; schematic maps

I.

INTRODUCTION

Spatial orientation and wayfinding have fascinated people through the ages [1] and this has been expressed in maps and mapping tools. Mapping provides a mechanism by which to associate data to places and draw people’s attention to certain information such as mnemonic techniques. Nowadays, digital maps are more widely used than paper maps. People almost everywhere use online applications to visualize locations and find routes. Twelve years ago, there were none of the popular online navigation tools that are now part of everyday life. With online mapping tools, people not only use maps for wayfinding tasks, but also create more maps and innovative representations for combining various data sets. Crowdsourcing changed the way our society works, with people sharing more data and effort in many different types of collaborative projects, notably also in mapping, e.g. OpenStreetMap. However, visual information on transportation routes is still lacking in many cities, just as it was twelve years ago. Automated schematic mapping remains a challenge. Ideally, the automatically generated schematic maps should have a quality comparable with good hand-drawn maps, and also be appropriate for viewing upon different devices. Wayfinding is a key design issue. The map layout should be informative, efficient, and aesthetically pleasing. These three seemingly simple requirements for a schematic map are judging from past experiences - very difficult to fulfil. The schematization of spatial information can be based on criteria related to characteristics of the data to be mapped,

map design and/or users. In addition to the vast possibilities for schematizing networks, “there is greater joy in a good approximation than in an exact solution” (Julian Schwinger, Nobel in Physics 1979). Indeed, many approaches to find approximate solutions for automated schematic maps have been proposed. However, the ultimate goal is the user’s comprehension of the automated map. Users should be able to use the schematic map to find their way quickly around the real environment. II.

PAST RESEARCH

More than 12 years separate the PhD thesis of Avelar published in 2002 [2] from the publications by various authors presented in this Schematic Mapping Workshop. The PhD thesis presented the first investigation of automated generation of schematic maps, in which computational and cartographic challenges were examined, and solutions developed and implemented. Different strategies for designing schematic maps were also analysed. The resulting schematic networks were produced by iterative techniques applied to the road network of a city. The research of Avelar focussed on: • constraint-based schematization of networks [2, 3, 4, 5] • modelling of transport network data [2, 6] • design of schematic maps for showing routes of transportation systems [2, 7, 8] • analysis of quality of schematic networks [2, 4] • assessment of users’ experience of a public transport map [9] Generating schematic maps automatically requires different sets of knowledge to be combined. They include how to schematize map elements, the modelling of navigation tasks that the schematic maps are meant to support, how users perceive and understand the environment and the mapped world, and how to apply cartographic principles to view information upon different devices. While a number of new approaches for schematizing routes have been developed over the last twelve years, less attention has been paid to the integration of the aforementioned sets of knowledge for the creation of high quality schematic maps.

Metro maps are an instantiation in this direction. Progress has been made in their automated mapping since 2002, such as [10], [11] and [12], together with the evaluation of users’ experience, e.g. [9] and [13]. Metro maps these days can be created using many computer techniques and geometries, which can mainly be traditional stylized lines, curves (e.g. [13] and [14]) and circles (e.g. [15]). III.

environment of many large cities means that their schematic representations must be continually updated. How much change can a map layout have without causing confusion to users due to an inconsistent experience? Do we really have to offer a single solution to users or can we provide map options as personalised solutions? (6) Studies in cognition and human mobility patterns. Find shortest-distance-yet-with-fewestturns routes. How many reasonable alternative routes? In which cases should routes be schematized as lines, curves, circles or concentric circles?

CHALLENGES AND OPPORTUNITIES

Based on the past work and experience, here are a few reflections upon the open challenges of the automatic generation of high quality schematic maps. Some items may be arguable, and can be further discussed and debated during the workshop. (1) Contextual schematization of networks. Road and water networks of a city are commonly considered. How to weigh and prioritise different schematization specifications to direct the schematization process, as well as to evaluate result afterwards?

This list can continue to grow as new concepts and applications of schematic maps are developed, as well as new algorithms applied to issues not yet solved. REFERENCES [1]

(2) Consistent multi-scale information. Automatic schematization of topographic data and visualization of results at various scales, with appropriate selection of details to be viewed in the schematized data and background. Keeping topology consistent is essential, but not the shape of map elements, thus at smaller scales shapes can become more and more abstract. How much can a user zoom in and out, such that varying scales still show consistently useful information? (3) Quality issues in schematic maps. Schematic maps research should go beyond schematizing lines. Some old questions that are still open include a quality measure for schematic networks. How do we know what is a good design for a schematic map? Assess schematization indicators, category of lines and nodes, users’ experience and cognitive ability. Combine schematized elements with different geographic information, and find out what and how much information can influence the readability of a map. (4) Data repository. A repository could be set up to archive algorithms, source code and data, which can then be shared among researchers. In this way, automated schematic mapping research becomes reusable and extensible, as has been done in other disciplines, such as in biology and physics. Freeaccess, benchmark data for schematic transport maps containing topographic data and public transport routes of cities could also be made available. In addition, time schedules and other transportation information could be embedded in the data. (5) Complexity of large cities. The complex transportation systems and rapidly changing

[2]

[3]

[4] [5]

[6]

[7] [8] [9]

[10]

[11]

[12]

[13]

[14]

[15]

P. Arthur and R. Passini. Wayfinding: People, Signs, and Architecture, McGraw-Hill: New York, 1992. S. Avelar, Schematic Maps On Demand: Design, Modeling and Visualisation, PhD Thesis, Swiss Federal Institute of Technology, 2002. S. Avelar and M. Müller, “Generating topologically correct schematic maps”, in Proc. 9th Int. Symposium on Spatial data Handling, Beijing, pp.4a.28-4a.35, 2000. S. Avelar, “Convergence analysis and quality criteria for an iterative schematization of network”, Geoinformatica, 11, pp. 497-513, 2007. S. Anand, S. Avelar, J. M. Ware, and M. Jackson. “Automated schematic map production using simulated annealing and gradient descent approaches”, In Proc. of GIS Research UK Conference (GISRUK’07), pages 414–420, 2007. S. Avelar and S. Huber, “Modeling a public transport network for generation of schematic maps and location queries”, in Proc. of 20th Int. Cartographic Conference, pp.1472-1480, 2001. S. Avelar and L. Hurni. “On the design of schematic transport maps”. Cartographica, 41(3), pp. 217-228, 2006. S. Avelar, 2008. “Visualizing public transport networks: an experiment in Zurich”, Journal of Maps, pp. 134-150, 2008. S. Avelar and J. Allard, “From graphical presentation to users’ comprehension of Transantiago network map”, in Proc. of 24th Int. Cartographic Conference, 2009. S.-H. Hong, D. Merrick, and H.A.D. do Nascimento, "The Metro Map Layout Problem," Proc. Asia Pacific Symp. Information Vizualization (APVIS '04), pp. 91-100, 2004. J.M. Stott and P. Rodgers, "Metro Map Layout Using Multicriteria Optimization," Proc. Information Visualization Conf. (IV '04), pp. 355-362, 2004. M. Nöllenburg and A. Wolff, "A Mixed Integer Program for Drawing High-Quality Metro Maps," Proc. 13th Int'l Symp. Graph Drawing (GD '05), pp. 321-333, 2005. M.J. Roberts, E.J. Newton, F.F. Lagattolla, S. Hughes, M.C. Hasler, “Objective versus subjective measures of Paris Metro map usability: Investigating traditional octolinear versus all-curves schematics”, Int. J. Humand-computer Studies, 71(3), pp.363-386, 2013. M. Fink, H. Haverkort, M. Nöllenburg, M. Roberts, J. Schuhmann, and A. Wolff, “Drawing metro maps using Bézier curves”, in Proc. 20th Int. Sympos. Graph Drawing (GD’12), LNCS, W. Didimo and M. Patrignani, Eds., vol.7 704, Springer-Verlag, pp.463-474, 2013. M.J. Roberts, Underground Maps Unravelled, Explorations in Information Design, publ. by author, 2012.