Map-labelling with a multi-objective evolutionary algorithm

We present a multi-objective evolutionary algorithm approach to the map-labelling problem. Map-labelling involves placing labels for sites onto a map such that the result is easy to read and usable for navigation. However, map-users vary in their priorities and capabilities: for example, sightimpaired users need to maximise font-size, whereas other users may be willing to accept smaller labels in exchange for increased clarity of bindings of labels to sites. With a multiobjective approach, we evolve a range of labellings from which users can select according to their particular circumstances. We present results from labelling two maps, including a difficult, dense map of Newcastle County in Delaware, which clearly illustrate the advantages of the multi-objective approach. Categories and Subject Descriptors: I.2.1 [Artificial Intelligence]: Applications and Expert Systems — cartog

• Type: Conference
• Year: 2005
• Where: GECCO
• Authors: Lucas Bradstreet, Luigi Barone, R. Lyndon While
• Keywords: Evolutionary Algorithm Approach, GECCO 2005, Involves Placing Labels, Multiobjective Approach