Building rules for effective legend design upon gestalt laws

Abstract

Maps are widely used in a variety of fields, such as research, engineering, economy, politics, and education. However, many maps are criticized for their inefficiency and/or difficulty in understanding (interpretation), especially those designed by GIS users who have not received proper cartographic training. Efforts have been made by researchers on the development of rule-based systems to assist GIS users to design effective maps. However the results are not very satisfactory, mainly due to the difficulty in acquiring proper knowledge and the lack of proper rules. This project aims to develop one type of rules for cartographic design, more specifically rules for the design of effective map legend. From literature, it has been found that no systematic investigations into effective legend design have been carried out although some aspects of legend design have been touched such as design of legend symbols, evaluation of some feature arrangement and design of interactive legend. As a consequence, this project concentrates on a systematic investigation into the building of cartographic rules essential for the effective arrangement (i.e. spacing, alignment, and grouping) of legend features (i.e. symbols + descriptions) upon Gestalt laws a group of famous visual perception principles. The objectives of this project are then, (1) to build sets of cartographic rules essential for arrangement of legend features upon Gestalt laws, (2) to prioritize these rules, and (3) to examine the applicability of these rules to interactive legends. To achieve these objectives, the first strategy is to make full use of the Gestalt laws for the development of rules for legend design. As a result, four of the thirteen Gestalt laws, i.e. similarity, proximity, continuity and common region, have been utilized for the development of spacing, alignment and grouping rules. The second strategy is to evaluate the effectiveness and efficiency of the legends designed based these rules developed in this project, by using experimental testing. To conduct such tests, (a) a web-based system is developed to enable all tests on line; (b) accuracy and efficiency of visual search for a series of features are employed as benchmark; (c) statistical tests, i.e. F-test and t-test, were applied to analyse the results in order to examine whether or not there is any possible significant improvement. The first set of cartographic rules for legend design developed in this project is spacing rules. A total of nine rules have been built upon similarity and proximity laws. Such rules are used to guide the consideration of spacing between letters, between words, between a symbol and its description, between features, and between groups of features. That is, these rules specify the proper use of spacing to enforce the effective visual grouping of symbols and text in the legend. For example, Spacing Rule (g) states: "the spacing between a symbol and its description must be smaller than the spacing between adjacent lines of features (in row or column)". This rule makes the symbol and its description grouped together as a group instead of two adjacent neighbouring symbols in different rows in a map legend. Evaluation results of the nine spacing rules indicate that (1) the spacing rules do make the efficiency of legend improved significantly; and (2) spacing rules (a), (h), (i) and (g) are more important than the other spacing rules (b), (d), (c), (e) and (f). The second set of cartographic rules for legend design developed in this project is alignment rules. A total of three rules have been built upon the law of continuity. Such rules are used to guide the consideration of alignment between a symbol and its description, between features, and between groups of features. These rules specify the proper determination of the alignment to enforce the effective visual grouping of symbols and text in the legend. For instance, Alignment Rule (a) states: "a symbol and its description must be aligned along the horizontal centre lines". This rule makes the symbol and its description visually grouped effectively. Evaluation results of the three alignment rules indicate that (1) the alignment rules do make a legend significantly more efficient; and (2) The alignment between features is more important than the alignment between symbol and description and the alignment between groups. The third set of cartographic rules for legend design is grouping rules. A total of three rules have been built upon similarity, proximity and common region laws. These rules specify the proper ways of grouping features to enforce the effective visual grouping of symbols and text in the legend. For example, Grouping Rule (a) states: "symbols to be grouped must be relatively similar". This rule makes the features visually grouped effectively. Evaluation results of the three grouping rules indicate that (1) the grouping rules do make a legend efficiency significantly improved; and (2) The grouping of features upon similarity is more important than the grouping upon common region and the grouping upon proximity. Prioritization of these rules have been made with the results of grouping, spacing, and alignment rules in order, through a theoretical analysis of the intensities of grouping (e.g. grouping features). For instance, the intensity of grouping features by the grouping rules is higher than that by the spacing rules as the effect of combined similarity, proximity and common region is stronger than the effect of proximity. Evaluation results indicate that the grouping rules are more important than the spacing rules, which in turn are more important than the alignment rules. It is expected that the spacing, alignment and grouping rules are not only applicable to the legends in static form, but also to the legends in interactive form. Therefore, the applicability of these rules to control-panel legends was studied. The three types of rules were applied one by one to the interactive legend of the GlobeLand30 China's global land cover map with 30m resolution. Evaluation results indicate that all the three types of rules are applicable to interactive legend and the improvement in efficiency of interaction is significant.