Extending an association map to handle large data sets

Abstract

This paper presents the Association Map-Large (AM-L), an interactive visualization of entity associations. AM-L is an extension of a previously reported AM interface that has been enhanced with search and interaction features for supporting larger data sets. We report on a user study with thirty two participants, which assesses user performance and experience with AM-L versus a tabular representation of the same data in the context of an enterprise system. Participants with varying levels of experience were given both simple and complicated tasks to complete with each system. Results indicate greater enjoyment and lower levels of mental effort when using AM-L, as well as less time spent on average when performing tasks. Accuracy results in terms of correctness indicate a learning curve, with overall performance worse with AM-L on the first two simple questions and first complex question, but then as well or better on subsequent questions. Given that the AM-L interface is unlike any with which the users had prior experience, it is not surprising that some exposure to the interface, such as training, would be helpful prior to use. Suggestions from the participants will inform future enhancements to the interface, which will be validated with further studies.