MMap Plasticity
Christian Kray
Institute for Geoinformatics (ifgi), University of Münster, [email protected]://orcid.org/0000-0002-4199-8976
Auriol Degbelo
Institute for Geography, University of Osnabrück, [email protected]://orcid.org/0000-0001-5087-8776
Abstract
With the arrival of digital maps, the ubiquity of maps has increased sharply and new map func-tionalities have become available such as changing the scale on the fly or displaying/hiding layers.Users can now interact with maps on multiple devices (e.g. smartphones, desktop computers,large-scale displays, head-mounted displays) using different means of interaction such as touch,voice or gestures. However, ensuring map functionalities and good user experience across thesedevices and modalities frequently entails dedicated development efforts for each combination. Inthis paper, we argue that introducing an abstract representation of what a map contains andaffords can unlock new opportunities. For this purpose, we propose the concept of map plasticity ,the capability of a map-based system to support different contexts of use while preserving usabil-ity and functionality. Based on this definition, we discuss core components and an example. Wealso propose a research agenda for realising map plasticity and its benefits.
Human-centered computing: HCI theory, concepts and mod-els
Keywords and phrases map semantics, map interaction, map plasticity, ubiquitous computing
Maps have become ubiquitous nowadays, and they have a broad range of application areasincluding science, city planning, journalism and storytelling, tourism, maritime navigationand aviation. This diversity of scenarios is being served by specific maps, which neverthelessare all based on the same fundamental concept. While many definitions of what a mapactually is exist (as recently discussed in [7]), maps are frequently defined as a symbolised,visual representation of the environment. Technological advances have opened up newpossibilities for map making and map use, notably mass production, changing scale on thefly and displaying (or hiding) layers. Yet, these advances come with new challenges for thescience and practice of map design. For instance, users can now interact with maps onmultiple devices (e.g. smartwatches, smartphones, desktop computers, large-scale displays,head-mounted displays). They then expect a similar (enjoyable) experience across all ofthem as well as the ready availability of map functionalities they require to solve their taskat hand. Cross-device computing (thoroughly reviewed in [2]) opens up further opportunitiesin application domains such as education, data exploration, health and collaboration, but itremains unclear how map-based systems can embrace these without dedicated developmenteffort for different device types. How to keep map experiences across devices consistent andhow to realise smooth cross-device map uses, is still not fully understood in GI Science,Cartographic and Human-Computer Interaction research. In this article, we thereforeintroduce the concept of map plasticity to advance discussions of map-based systems which a r X i v : . [ c s . H C ] S e p :2 Map Plasticity support different contexts of use while preserving usability and functionality. We outline thebasic concept, provide some examples for what it can do and outline a research agenda forrealising it. Maintaining consistent map experiences across different contexts of use requires first andforemost a better understanding of the two key concepts of ‘map’ and ‘context’. Below, wethus briefly present relevant research on these two axes, along with the key idea behindinterface plasticity, which is the concept that the idea of map plasticity is based on.As discussed by Peuquet [9], a map can be viewed conceptually as an image or a geometricstructure . Considering maps as images has resulted in the application of principles fromcognitive and perceptual psychology to map-based research; considering them as geometricstructures is evidenced in the application of tools from mathematical subfields (e.g. topology,graph theory) to map-based research. In the digital context, viewing map as images has ledto the raster data model, while viewing them as geometric structures led to the vector datamodel. Another perspective, prominent in Tomlin’s map algebra [13], is to view maps as a setof layers , each of which supports a set of operations (e.g. computation of the distance from apoint on the layer to another location). Functions of map algebra are most fully developedfor datasets in raster format. A less well-known way of viewing maps is to conceptualisethem as a set of assertions which can be extracted by looking at it [11]. This view has theadvantage that it enables an expressive description of facts depicted by the map (e.g. in alanguage such as Resource Description Framework), and their subsequent retrieval usingsemantic technologies. That is, it facilitates map retrieval beyond keywords extracted fromtheir metadata, and makes content-based querying of maps first-class citizens.A discussion of ‘context’ in relation to map-based design was presented recently in theresearch agenda outlined by Griffin et al. [6]. They stated that “we continue to lack effective,consistent strategies for describing context and implementing our understanding of it tosolve design problems”, and proposed a model of context with four dimensions: map, user,activity, and environment. Griffin et al. took an emergent view on context, that is, distinctmap use contexts emerge from individual map use situations. Finally, they proposed 15research directions revolving around map design outcomes that are effective for varying mapuse situations. Opportunity transferable between devices,users, or activities) is the focus of the current paper. The discussion of what exactly shouldbe transferred is arguably dependent on what view of a map is adopted. For example, wemight want to transfer some visual appeal (map as image), topological consistency (map asgeometry), distance values between two points (map as layer) or even some very specific factsdepicted on the map (map as set of assertions). While fascinating, this discussion is left forfuture work for the moment. The article concentrates solely on sketching some ideas on howthe problem of design transferability can be approached on a more general, abstract level.Interface plasticity has been defined as a property of adaptation to different contexts[3], where context includes aspects such as device properties, means of interaction and thephysical environment in which the interaction takes place. Realising interface plasticitycould benefit both developers and users of interactive systems. From the developers’ pointof view, productivity is the main gain, as they can focus on implementing functionalities,and disregard platform-specific (or context-specific) requirements. From the users’ point ofview, consistency of experience across contexts is the main benefit. The context (e.g. device) ray & Degbelo X:3 becomes transparent to the user, as the experience remains similar despite variations in thesituations of use. A useful approach to realise interface plasticity is model-driven engineering(see [4]), that is, having a model , which represents a thing, and a meta-model , which setsthe rules for producing models. In general, no technique so far can be said to have fullyrealised interface (or interactive system) plasticity for all circumstances, but some existingtools/techniques arguably enable plasticity to some degree. For instance, Apache Cordovaenables developers to write one code base, and build apps for several mobile platformsthrough its cross-platform workflow . Since the final apps on the native mobile platforms canbe considered similar, Apache Cordova can be said to realise mobile app plasticity to someextent. In the context of web maps, responsive web design enables the appropriate renderingof maps on a variety of devices and window or screen sizes, and can thus be considered atechnique, which enables plasticity to some extent. Due to the specific properties of mapssuch as depicting a certain part of the physical world, simply changing the location andsize of the interface element showing the map is not sufficient. For example, moving from adesktop to a mobile device might make the map so small that the relevant entities depictedon the desktop version are no longer visible on the smartphone version. There are many moreaspects of maps that could be adapted to different contexts (e.g. scale, visualisation, contentselection). To address these, the next section introduces the concept of map plasticity. As discussed in the previous section, interface plasticity is the general concept that underpinstechnologies such as responsive web design, which enables users of interactive system tointeract with a system in different contexts. Maps can be conceived of as being a specialisedkind of user interface, i.e. one that facilitates working with spatial information. Maps affordspecific actions such as pan or zoom, which can be triggered in different ways, e.g. by clickinga mouse button over a map location or performing a gesture. Another particularity of mapsis that they usually represent (spatial) phenomena from the physical world. This differsfrom the more general case of user interface elements such as menus, which do not usuallyhave this inherent link to the physical world. Based on these considerations, it makes senseto consider what plasticity means when applied to maps and how to achieve it as it wouldpotentially enable a broader and more effective adaptation of maps to different scenarios.
As outlined above, maps are particular types of interfaces and their specific properties shouldbe considered in a corresponding definition. Based on these considerations we define map plasticity as the capability of a map-based system to support different contextsof use while preserving usability and functionality of the map.There are several terms in that definition that require further unpicking. By map-basedsystem we mean any system that provides users with a map-like representation (usuallycorresponding to a part of the physical world) as integral part of its interface. Examples forsuch systems include geographic information systems (GIS), pedestrian navigation systemsor typical web sites for finding a hotel.
Contexts of use refer to technological componentsinvolved in the presentation of the map as well as those facilitating interaction with the map. https://cordova.apache.org/docs/en/latest/guide/overview/ (last accessed: April 2, 2019). :4 Map Plasticity For example, the size and resolution of a display or properties of a gesture-recognition systemwould be subsumed under this concept. General properties of the underlying system (e.g.computational power) can be considered in this category as well. In addition, the actual usersof the system and their properties as well as the properties of the situation within which theyuse the system are included under this heading. Perceptual aspects such as colour perceptionor current stress level are examples for user-related aspects. The degree of crowdedness atthe time and location where the map is used would qualify as a property of situation of use.Another key element of our definition is the preservation of usability and functionality ofa map . In the original definition of interface plasticity [12], preserving usability entails thatusers of an interactive system are able to perform the actions they require to complete thetask they are pursuing even when physical characteristics of a system change. For example,users should be able to complete a purchase with an online retailer regardless of whether theyaccess the corresponding payment system at their desktop using a keyboard and a mouse oron their mobile using voice input. This general notion of usability preservation also appliesto maps but needs to be extended to account for their particularities. More specifically,this includes consideration of graphical [1] and other variables [8] that are used to generatethe map as well as layers, scale, and the viewport of a map. All these aspects need to betranslated in some way for a plastic map to preserve its usability and functionality.Finally, it is worth mentioning that recent research in Human-Computer Interaction (HCI)attempts to draw boundaries between the concepts of usability and user experience, usingthe former to refer to pragmatic properties of an interactive system (e.g. effectiveness), andthe latter to refer to hedonic properties (e.g. fun). Nonetheless, the term ‘usability’ is kepthere in the definition, because used in the original definition of interface plasticity [12]. Mostimportantly, as discussed in [14], in the early days of HCI, usability was used ‘to encompassalmost any aspect of HCI’. Thus, map plasticity as presented here, does not exclude thepreservation of hedonic or other properties (e.g. aesthetics) of maps, as one moves from onecontext to another. The next subsections outline how this move could be achieved.
The realisation of map plasticity poses similar challenges to realising general interface plasticity.While it adds further considerations (i.e. aspects specific to maps), it also reduces complexityby focusing on a single user interface element (maps) rather than all. Figure 1 providestwo perspectives on how map plasticity could be realised. The left side of Figure 1 mirrorsThevenin et al.’s original framework [12] and adapts it to map plasticity. The frameworkcombines several models covering key aspects to derive a concrete/physical interface. In thefigure, models that can largely be kept as in the original framework, are depicted in blue.Models that need to be considerably adapted are shown in orange.Like in the original framework, the user task model thus formally describes the goalsand activities of a user, whereas the system task model encapsulates how these work taskscan be realised by the (map-based) system. The platform model also mirrors the definitionfor general interface plasticity, i.e. it describes the physical characteristics and resources ofthe target platform. The other components of our proposed framework deviate from theoriginal concept. The interactors model consists of interactors that are available for specific,concrete instances of a map. They respond to events produced by users (or other componentsof the system) and translate them to operations that be executed by the system in orderto perform user tasks. The environmental model originally encapsulated general aspectsof the context of use (e.g. objects or people nearby at the time of use of the system), buttakes on an important additional role in our proposed model for map plasticity. Since maps ray & Degbelo X:5
Figure 1
Left: Components and their interplay to achieve map plasticity (adapted from [12]);right: information flow and mappings usually depict a specific part of the physical world, this part needs to be represented aswell, i.e. as part of the environmental model or a dedicated model encoding properties ofthe environment depicted in the map. In Figure 1(left), we depicted the latter option viaa second box behind the environmental model. This second model is closely linked to firstenvironmental model. For example, a nearby object present in the original environmentalmodel often can correspond to a real-world entity that is also part of the other environmentalmodel since it is of relevance for the map currently being shown.The core element of our framework, the abstract map model is also strongly connected tothe environmental model that encapsulates information about the part of the physical worldthat is of relevance to the map. The abstract map model describes the actual content of amap on an abstract level, independent of how it is rendered/presented to the user and ofhow a user interacts with it. Consequently, from a single abstract map model it is possibleto generate any kind of specific map-based interface (e.g. for different devices, interactors orenvironments) – provided the corresponding models are available as well as a mechanism toderive specific maps using them. The abstract map model needs to take into consideration thetasks and goals specified in the user and system task model to ensure that users can achievetheir goals using the information contained in it. The concrete/physical map corresponds tothe actual (physical) instantiation of a map that users can directly interact with, e.g. a visualmap displayed on a touchscreen. The concrete map is the result of a systematic process(e.g. constraint resolution/propagation, reasoning, planning) that takes into account theplatform model, the interactors model, environment model(s) and the abstract map model.The outcome is an interactive artefact that is adapted to the context of use and preservesthe map functionality users need to achieve their goal.An alternative perspective of map plasticity is depicted in Figure 1. Here we have focusedon information flow and mappings with the underlying assumption that a map-based systemenables a user to interact with spatial information. This information usually resides in a spatial data repository (e.g. a database), from which we need to derive an abstract map (usingthe framework discussed above). From this abstract map, we can generate a concrete map asoutlined above, which users can then perceive. Users interact with the concrete map (usinga specific interactor) and this interaction then needs to be translated to an equivalent actionon the abstract map (also using a specific interactor). For example, the tap on a touchscreenshowing map would need to be translated to a selection of a specific entity contained in theabstract map. In order to fully realise the intention of the user, this action on the abstractmap then needs to trigger a corresponding manipulation of the spatial data repository. Forthe selection task, for example, this might correspond to setting a binary selection flag totrue in a database. Consequently, a different perspective on map plasticity is to perceive it :6 Map Plasticity as a staged mapping process, where (inter)actions on different representations are mappedsystematically to ensure consistency while enabling users to easily achieve their goal. Thefollowing section provides a short illustrative example how this could unfold in practice.
As an example, let us consider a simple restaurant recommendation service. In this case theuser task model would describe the user goal (finding a restaurant) and activities (searchingfor restaurants, accessing information about them). The system task model would encodehow this can be achieved using the system (e.g. via specific queries issued to a database).The abstract map model would describe the entities of interest, i.e. restaurants in the regionof interest as well as auxiliary information of relevance such as street networks or publictransport links. The environment model would encode the situational context (e.g. currentlocation of user or co-present persons) and detailed information about the region of interest(e.g. full geometric details about any entities contained in the region). The platform modelwould encode information about the device through which users interact with the concretemap. For example, the computational, interactive and display properties of a smartphone orhead-mounted display. For each means of interaction, there would be an interactor descriptionin the interactor model. This then allows for automatically generating an adapted map-basedinterface for a smartphone and a head-mounted display. In the former case, a standard mapwith touch interaction could be generated. The touch interactor translates concrete touchesthe users performs on the device to actions on the abstract map. In the latter case, the mapcould be realised as an augmented reality overlay, e.g. graphical markers that indicate thelocation of restaurants in the environment. Users could interact with these, for example, bylooking at them for an extended period of time. The gaze interactor would then translatethis to the same action on the abstract map as in the case of smartphone. Other contexts ofuse could be realised similarly.
While the idea and short example in the previous sections provide a glimpse of what mapplasticity might be and enable, this is clearly at an early conceptual stage. Much furtherwork is need to realise and explore the concept. In terms of future work, the different modelsof the framework need to specified in detail. While this work may borrow from work inHCI and GI Science (cf. e.g. [10]), particularly the formalisation of abstract maps as wellas representational/interaction mapping constitute enticing opportunities for breaking newgrounds. Based on such formalisations, it will then be possible to investigate mechanismsthat connect all components of the framework and are capable to generate concrete mapsfrom them. The authors of the original interface plasticity concept suggested constraint-basedapproaches for this purpose though there is no inherent property that would preclude theexploration of alternative approaches. Finally, it would of course also make sense to evaluatethe overall concept of map plasticity along several dimensions. This includes determining theformal properties and limitations as well as carrying out user studies to assess whether theapproach is successful in supporting different contexts of use while preserving usability andfunctionality of the map. If we manage to realise map plasticity as envisioned in this paper, itmay lay the foundation for a new understanding of what constitutes a map and enable muchmore flexible map-based systems. It could be a key enabler for intelligent geovisualisations[5], which is based on the vision of adapting maps to different input datasets while beingcapable of monitoring and maximising understanding on the user’s side. ray & Degbelo X:7
In this paper, we outlined the concept of map plasticity, based on Thevenin et al.’s concept ofinterface plasticity [12]. The proposed framework has the potential to enable the adaptationof maps to different contexts of use, and thereby to reduce development effort, increaseconsistency and facilitate intelligent geovisualisations. We described the core components ofour framework and their interaction while also outlining the underlying information flow.Following a short example, we sketched a research agenda for realising map plasticity, whichwe intend to pursue in the future.
References Jacques Bertin, William J Berg, and Howard Wainer.
Semiology of graphics: diagrams,networks, maps , volume 1. University of Wisconsin press Madison, 1983. Frederik Brudy, Christian Holz, Roman Rädle, Chi-Jui Wu, Steven Houben, Clemens Klok-mose, and Nicolai Marquardt. Cross-device taxonomy: survey, opportunities and challengesof interactions spanning across multiple devices. In
Proc. of CHI 2019 . ACM Press, 2019. Gaëlle Calvary, Joëlle Coutaz, David Thevenin, Quentin Limbourg, Laurent Bouillon, and
Jean Vanderdonckt. A unifying reference framework for multi-target user interfaces.
Inter- acting with computers , 15(3):289–308, 2003. Joëlle Coutaz. User interface plasticity: model driven engineering to the limit! In NoiSukaviriya, Jean Vanderdonckt, and Michael Harrison, editors,
Proceedings of EICS ’10 ,pages 1–8. ACM Press, 2010. Auriol Degbelo and Christian Kray. Intelligent geovisualizations for open government data(vision paper). In Farnoush Banaei-Kashani, Erik G. Hoel, Ralf Hartmut Güting, RobertoTamassia, and Li Xiong, editors,
Proc. of SIGSPATIAL International Conference on Ad-vances in Geographic Information Systems , pages 77–80. ACM Press, 2018. Amy L. Griffin, Travis White, Carolyn Fish, Beate Tomio, Haosheng Huang, Claudia RobbiSluter, João Vitor Meza Bravo, Sara I. Fabrikant, Susanne Bleisch, Melissa Yamada, andPéricles Picanço. Designing across map use contexts: a research agenda.
InternationalJournal of Cartography , 3(sup1):90–114, oct 2017. doi:10.1080/23729333.2017.1315988 . Menno-Jan Kraak and Sara Irina Fabrikant. Of maps, cartography and the geography of theInternational Cartographic Association.
International Journal of Cartography , 3(sup1):9–31, oct 2017. doi:10.1080/23729333.2017.1288535 . John B Krygier. Sound and geographic visualization. In
Modern cartography series ,volume 2, pages 149–166. Elsevier, 1994. Donna J. Peuquet. Representations of geographic space: toward a conceptual synthesis.
Annals of the Association of American Geographers , 78(3):375–394, sep 1988. doi:10.1111/j.1467-8306.1988.tb00214.x . Robert E Roth. Cartographic interaction primitives: Framework and synthesis.
The Car-tographic Journal , 49(4):376–395, 2012. Simon Scheider, Jim Jones, Alber Sánchez, and Carsten Keßler. Encoding and queryinghistoric map content. In J Huerta, S Schade, and C Granell, editors,
Proc. of AGILE 2014 ,pages 251–273. Springer International Publishing, 2014. David Thevenin and Joëlle Coutaz. Plasticity of user interfaces: Framework and researchagenda. In
Interact , volume 99, pages 110–117, 1999. Dana C. Tomlin. Map algebra: one perspective.
Landscape and Urban Planning , 30(1-2):3–12, 1994. doi:10.1016/0169-2046(94)90063-9 . Noam Tractinsky. The usability construct: A dead end?
Human–Computer Interaction ,33(2):131–177, mar 2018. doi:10.1080/07370024.2017.1298038doi:10.1080/07370024.2017.1298038