Label placement preferences for digital maps

Digital map applications produce maps on the fly, using label placement algorithms to optimize the map layout for various factors. Previous work has made great strides in quickly generating layout placement algorithms that minimize collisions or reduce clutter, but little has been done to explore optimizing layouts while considering the visual preferences of map users. This dataset was collected as part of a study to better understand people’s visual preferences for various label positions in map layouts and to support the development of new label placement algorithms that can generate map layouts that consider those preferences. This dataset of label placement preferences includes two parts. The first part contains 956 binary choice responses that asked participants to choose their favorite of two maps. Labels on each map differed in either their alignment to the point of interest or distance from it. The dataset also includes 1912 ranked choice responses where participants were asked to rank 3 maps that differed in both alignment and distance. The alignment and distance positions considered are the same across both parts of the study.


Objective
Digital maps generate new map displays as users pan and scale around the map.Label placement algorithms are an important part of this process, automatically placing labels at locations near their respective areas of interest.Label placement algorithms are generally optimized to avoid collisions with other labels and reduce clutter [1], but do not account for visual preferences of map users.As a result, digital map designs often do not meet cartographer standards or align with the map users' preferences or expectations [2].
Some previous work has explored learning and representing preferences in maps as a way to recommend content based on location [3].Other work has used mouse movements to determine user interest and hide content a user would not be interested in [4].Researchers have also used Bayesian optimization of human performance data to refine the design of map tooltips [5].To our knowledge, previous work has not studied people's explicit preferences for label placements on digital maps.
The dataset presented here was collected as part of a study to better understand people's visual preferences for point labels in map layouts.In particular, we consider their preferences involving two attributes, alignment and distance, as these are commonly manipulated by label placement algorithms when generating layouts that avoid collisions.We examined user preferences for a label's alignment with a point independently of its distance from a point.The different label positions were considered on different backgrounds and in the context of different route label densities.We also collected data on how people ranked maps where labels differed in both alignment and distance.The resulting data was used to design a preference model that shows how different users traded off between conflicting attributes [6].It is being used to support ongoing work to design an interactive machine learning tool for generating map views [7].Other researchers may be interested in using it to inform new map layout designs or to design and test new preference models and algorithms.

Data description
A total of 478 students from Louisiana State University (LSU) participated in the study for course credit.The study was approved by the LSU IRB.Part I of the study was a within subjects design with four independent variables: route density (low, medium, high), alignment (top left, direct left, and bottom left), distance (close, medium, and far), and background (grid map and radial map).This resulted in three separated 3 × 2 repeated measures designs: presented were route density by background, alignment by background and distance by background.The dependent variables included preference (binary choice) and preference strength (scale from 0 to 100).
Part II surveyed the participants' preferences using a preference ranking (1-3) of maps that represented various tradeoffs between alignment and distance.Participants performed their rankings on two separate label placement sets.A change from Part I to Part II included increasing the number of map backgrounds used to increase the number of data points collected for each label placement set from 2 to 4. This resulted in 8 total tradeoff rankings per individual (4 in Label Set 1 and 4 in Label Set 2), providing additional data for model fitting.A label set consisted of three maps, each representing a different tradeoff between alignment and distance.Between the two label placement sets, six possible combinations were surveyed, including (top left, medium), (direct left, far) and (bottom left, close) in Label Set 1 and (top left, far), (direct left, close), (bottom left, medium) in Label Set 2. An in-depth discussion of the study design can be found online at: https:// osf.io/ fhbjt/ wiki/ Metho ds/.
The results of the study indicated that people exhibit consistent visual preferences about where labels should be placed in relation to points of interest on a map (see Figs. 1 and 2).In Part II, when both alignment and distance were manipulated, people were more consistent in choosing positions that prioritized alignment over distance (see Fig. 3).
When considering the results of both Parts I and II, it was found that in situations where the most popular distance position (medium) was paired with the least popular alignment (bottom left), few participants chose to rank (bottom left, medium) highly.Instead, most people preferred either (top left, far) and (left, close).This indicates that more people gave alignment greater weight than distance when ranking between positions that change in both alignment and distance.
The provided dataset provided from this study includes the preference and preference strength responses from Part I of the study [8], as well as the rankings from Part II of the study [9].A Jupyter notebook showing some example Python code for using the datasets is also provided [10].

Fig. 1 •
Fig.1 Summary of Alignment Preference Results.A shows the proportions of participant choices between Left and Bottom Left.B participant choices between Left and Top Left.C shows the proportion of participant choices between Top Left and Bottom Left.Most participants preferred the Left alignment condition (66.4% vs Bottom Left, 57.7% vs Top Left), followed by Top Left (42.3% vs Bottom Left)

Fig. 2
Fig. 2 Summary of Distance Preference Results.A Shows the proportions of participant choices between Medium and Close.B participant choices between Far and Close.C shows the proportion of participant choices between Medium and Far.Most participants preferred labels in the Medium distance condition (74.3% vs Close, 69.8% vs Far), followed by Far (63.6% vs Close)

Fig. 3
Fig. 3 Summary of rankings from Part II.A Shows the percentage of participants that ranked each item as 1st, 2nd or 3rd in Label Set 1. B Shows the same for Label Set 2. C Shows the number of participants that ranked each item in 1st place from Label Set 1 (left column) and Label Set 2 (right column)

Table 1
Overview of data files/data sets Label Name of data file/data set File types (file extension) Data repository and identifier (DOI or accession number)