Technical SectionPoint feature label placement for multi-page maps on small-screen devices☆
Graphical abstract
Introduction
In the last decade, digital maps on mobile devices have become an integral part of our daily lives. Due to their ubiquitous availability they have become an indispensable tool, e.g., for navigation or exploration within a city. We consider a user following the information-seeking mantra by Shneiderman [1] as illustrated in Fig. 1: The user starts with an overview, zooms and filters information, and is provided with details on demand. We address the need for interfaces and methods that provide access to large and detailed geographic data sets without requiring the user to zoom to an extremely large scale. In particular, we consider the user obtaining a rough overview of a city, then zooming and panning the map to a specific area of interest. From a large set of points of interest, he or she queries for features of a certain category, e.g., restaurants. These are displayed as labels, such as small text or icons. We propose an approach that can handle high information densities without producing overlapping labels and cluttered maps. Instead of showing all labels contained in the area of interest at the same time, we distribute them over multiple pages (see Fig. 2). A user can navigate through the individual pages, each of which contains a non-overlapping subset of the labels. We call a distribution of labels over a certain number of pages a multi-page labeling.
Other labeling strategies reduce clutter by limiting the presented set of labels or shifting the labels outside the map. However, we see the following problems with these strategies. Only showing a subset of labels can lead to the omission of information relevant to the user. Hence, interactive map operations such as zooming or panning are often used for further exploration. However, repetitively zooming in and out and panning the current view of the map easily causes a user to lose orientation and the overall context of the map. Furthermore, placing the labels outside the map results in less screen space being available for displaying the map. Additionally, connecting lines are necessary for unambiguous label-feature associations, which can obscure other map features (e.g., roads) and can be difficult to follow. Instead, we propose the concept of multi-page maps which are a special type of zoomless maps. This generally refers to maps that provide specialized interaction techniques allowing a user to access all information without the need for zooming [2], [3].
Preceding research on zoomless maps focuses on the labeling of temporally fixed map sections which are considered separately. Hence, when a user pans the map, two successive map sections are not labeled consistently. This easily leads to distracting effects, such as labels repeatedly appearing and disappearing. In this paper, we tackle this problem and propose a method with two phases; a pre-processing phase and a query phase. In the pre-processing phase, we use an optimization approach to pre-compute labelings on the level of a whole city. We call such a solution a global labeling. The query phase is used for the on-demand retrieval of individual labelings at a more local level that can be accessed by a user in real time. We refer to these solutions as local labelings. Fig. 3 illustrates a global labeling and two exemplary local labelings marked with Ⓐ and Ⓑ. By querying local labelings from pre-computed global labelings, we ensure that a certain label is placed on the same page for all map sections in which it is contained. Thus, when a user pans the map, a certain label is placed on the same page in two successive map sections, helping the user to keep track while exploring information.
We emphasize that we only need to compute a global labeling once. Instead of requiring a time-consuming computation for each map section, we can rapidly extract local labelings in the query phase. Assuming a server-client communication, our strategy relieves the client in terms of computational overhead. Further, local labelings can be extracted for variable screen sizes and formats without the necessity of re-computing the underlying global labeling. Fig. 4 shows exemplary local labelings both on a smartphone and a smartwatch. To get an impression of the look-and-feel of our approach, we provide an illustrative demo video in the Appendix. Using a practical setting of a user searching for an ice cream parlor on his or her smartphone, we compare our strategy to a common labeling approach and highlight the benefits of our approach.
For the computation of global labelings in the pre-processing phase, we introduce multiple criteria that each labeling should satisfy. We set up a multi-criteria model that both aims at a small number of pages and distributes labels of important features as close to the front of the sequence of pages as possible. Thus, our model provides a user with quick access to relevant information. However, as a high label density can easily lead to an information overload [4] and a reduced map readability [5], we additionally aim at a balanced spatial distribution of labels on pages. This not only reduces information overload, but additionally avoids sparsely labeled pages. We refer to the optimization problem of finding a multi-page labeling satisfying all our criteria as MapCoverageLabeling.
We solve MapCoverageLabeling using an integer linear programming (ILP) formulation. In our experiments on real-world data, we show that large instances of MapCoverageLabeling with more than 5000 features can be solved within a few minutes. We determine a sweet spot that describes a suitable balance between our considered optimization criteria. Using different parameter settings, we compute global labelings and sample local labelings of different sizes, i.e., with different numbers of labels. We compare the extracted local labelings to reference labelings that are optimal regarding our optimization criteria. We show that the local labelings provide a comparable quality with respect to the optimization criteria. Further, we show that a majority of all labels can be placed on only few front pages.
Summarizing, this paper provides the following scientific contribution. For labeling small-screen maps, we introduce a two-phase strategy that distributes labels on multiple pages. Our approach facilitates the exploration of information without the need for zooming by swiping or clicking from page to page. Contrary to previous concepts of zoomless maps, we provide a consistent label-page assignment when a user pans the map. This requires in particular a careful design of the first phase of the method, which deals with pre-computing a data structure from which local labelings can be queried in the second phase.
Following, we provide a review of related work covering both map labeling and algorithmic research on vertex coloring (Section 2). Subsequently, we introduce our optimization criteria and propose a formal model (Section 3). Using this model, we define MapCoverageLabeling. We tackle this optimization problem by means of mathematical programming and present an ILP formulation in Section 4. Further, we explain our two-phase approach in Section 5. Using real-world data from four cities in the USA and Canada, we evaluate our approach based on experiments with different parameter settings (Section 6). Finally, we summarize our key contributions and give a brief outlook on future research (Section 7).
Section snippets
Related work
In the following, we address both fundamental and recent literature related to our research. We divide the related work into two main topics. In Section 2.1 we focus on research that relates to map labeling. In Section 2.2, we address the algorithmic contribution of our paper and shortly discuss the state of the art of vertex coloring.
Model
In order to set up a formal model, we first discuss suitable optimization criteria for distributing labels on pages. One important aspect that we take into account is the number of pages. In order to provide a user quick access to all labeled information, the number of interactions (e.g., swipes) necessary for browsing through the pages should be kept small. In many use cases, it is also preferable to consider feature priority and place labels of important features as close to the front of the
Mathematical programming
Generally, multi-page labeling asks for an assignment of labels to pages such that no two labels placed on the same page are in conflict with each other. This problem can be considered as a vertex coloring problem. We can transfer a set of labels into a graph that contains a vertex for each label and an edge for each label-label conflict. Distributing the labels on pages without introducing graphical conflicts then corresponds to a -coloring of the graph. Even for graphs representing the
A Two‐phase approach
Solving MapCoverageLabeling for each frame separately can be a time-consuming task that is not suitable for real-time applications. Further, it would not guarantee a consistent label-page assignment. Hence, we pre-compute global labelings in a pre-processing phase and extract local labelings in a query phase. Following, we describe both phases in more detail.
Experiments
In the following, we discuss our evaluation experiments and results to analyze the multi-page labelings obtained with our two-phase approach. First, we describe our experimental setup in Section 6.1. Then, we evaluate and discuss our results in Sections 6.2–6.4.
Conclusion and Outlook
In this paper, we follow the visual information-seeking mantra by Shneiderman [1]: overview first, zoom and filter, then details-on-demand. We propose a strategy for internally placing labels on small-screen maps that resolves label overlaps by distributing the labels on multiple pages. Our concept is neither limited to representing only a subset of labels, nor to shifting the labels outside the map. When exploring a particular map area, a user does not have to repeatedly zoom in and out, thus
CRediT authorship contribution statement
Sven Gedicke: Conceptualization, Methodology, Software, Investigation, Writing – original draft, Writing – review & editing, Visualization. Adalat Jabrayilov: Conceptualization, Methodology, Writing – original draft. Benjamin Niedermann: Conceptualization, Methodology, Writing – original draft. Petra Mutzel: Conceptualization, Methodology, Supervision, Funding acquisition. Jan-Henrik Haunert: Conceptualization, Methodology, Supervision, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
Partially funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC-2047/1 – 390685813; and by DFG [grant number 5451/6-1].
References (88)
- et al.
Placing text labels on maps and diagrams
Graphics Gems IV
(1994) - et al.
An empirical study of algorithms for point-feature label placement
ACM Trans Graph (TOG)
(1995) - et al.
A unified model and algorithms for temporal map labeling
Algorithmica
(2020) - et al.
Beyond maximum independent set: an extended integer programming formulation for point labeling
ISPRS Int J Geo-Inf
(2017) - et al.
Exactly solving the maximum weight independent set problem on large real-world graphs
2019 Proceedings of the 21st workshop on algorithm engineering and experiments (ALENEX)
(2019) - et al.
A comprehensive multi-criteria model for high cartographic quality point-feature label placement
Cartogr: Int J Geogr Inf Geovis
(2014) - et al.
Sliding labels for dynamic point labeling
(2011) - et al.
Consistent dynamic map labeling with fairness and importance
Comput Aided Geom Des
(2020) - et al.
Temporally stable boundary labeling for interactive and non-interactive dynamic scenes
Comput Graph
(2020) - et al.
Navigation techniques for small-screen devices: an evaluation on maps and web pages
Int J Hum-Comput Stud
(2008)
Stripemaps: improving map-based pedestrian navigation for smartwatches
Proceedings of the human-computer interaction with mobile devices and services (MobileHCI’15)
Off-screen landmarks on mobile devices: levels of measurement and the perception of distance on resized icons
KI-Künstliche Intelligenz
Implementation and evaluation of a fast area feature labeling method using auxiliary lines
ISPRS Int J Geo-Inf
A note on the complexity of the chromatic number problem
Inf Process Lett
On some properties of linear complexes
Matematicheskii sbornik
An exact approach for the vertex coloring problem
Discr Optim
Polyhedral studies of vertex coloring problems: the asymmetric representatives formulation
ArXiv e-prints
A branch-and-cut algorithm for graph coloring
Discr Appl Math
Cliques, holes and the vertex coloring polytope
Inf Process Lett
SmartWatch design fundamentals
The eyes have it: a task by data type taxonomy for information visualizations
Proceedings of the Visual Languages (VL’96)
Multi-page labeling of small-screen maps with a graph-coloring approach
Advances in Cartography and GIScience of the International Cartographic Association (ICA)
Zoomless maps: external labeling methods for the interactive exploration of dense point sets at a fixed map scale
IEEE Trans Vis Comput Graph
The concept of information overload: a review of literature from organization science, accounting, marketing, MIS, and related disciplines
Inf Soc
Measuring the influence of map label density on perceived complexity: a user study using eye tracking
Cartogr Geogr Inf Sci
Positioning names on maps
Am Cartogr
A program for automatic name placement
Proceedings of the sixth international symposium on automated cartography
Computer technology and cartographic change
The computer in contemporary cartography
The logic of automated map lettering
Cartogr J
An algorithm for automatic name placement around point data
Am Cartogr
Integer programming applied to the map label placement problem
Cartogr: Int J Geogr Inf Geovis
A packing problem with applications to lettering of maps
Proceedings of the seventh annual ACM symposium on computation geometry (SoCG’91)
A combinatorial framework for map labeling
Proceedings of the sixth International Symposium on Graph Drawing (GD’98)
A unified approach to labeling graphical features
Proceedings of the 14th annual symposium on computational geometry
Optimal packing and covering in the plane are np-complete
Inf Process Lett
Practical results using simulated annealing for point feature label placement
Cartogr Geogr Inf Syst
Improving label placement quality by considering basemap detail with a raster-based approach
GeoInformatica
Tabu search heuristic for point-feature cartographic label placement
GeoInformatica
Point labeling with sliding labels
Comput Geom Theory Appl
A new mathematical model and a lagrangean decomposition for the point-feature cartographic label placement problem
Comput Oper Res
Towards unambiguous map labeling-integer programming approach and heuristic algorithm
Expert Syst Appl
Analytical estimation of map readability
ISPRS Int J Geo-Inf
Towards an evaluation of quality for names placement methods
Int J Geogr Inf Sci
Placing text and icon labels simultaneously: a real-time method
Cartogr Geogr Inf Sci
Cited by (8)
Editorial Note
2021, Computers and Graphics (Pergamon)Citation Excerpt :The restricted screen space requires new visualization techniques as established strategies, such as maximizing the number of placed labels, quickly omit information. A paper by Gedicke et al. [2] proposes an approach that distributes all labels contained in a temporarily fixed map section on multiple pages. Their approach provides consistent label-page assignments so that labels do not appear and disappear when the user pans the map.
Points-of-Interest Tag Cloud Centered on the User's Geographic Location
2024, Journal of Geo-Information ScienceLBS Tag Cloud: A Centralized Tag Cloud for Visualization of Points of Interest in Location-Based Services
2023, ISPRS International Journal of Geo-InformationVisual human-computer interactions for intelligent vehicles
2023, Handbook of Human-Machine SystemsAn Empirical Study on Interfaces for Presenting Large Sets of Point Features in Mobile Maps
2023, Cartographic JournalAutomating the external placement of symbols for point features in situation maps for emergency response
2023, Cartography and Geographic Information Science
- ☆
This paper was recommended for publication by Luiz Henrique de Figueiredo.