The increasing availability of rich and complex data in a variety of scientific domains poses a pressing need for tools to enable scientists to rapidly make sense of and gather insights from data. One proposed solution is to design visual query systems (VQSs) that allow scientists to search for desired patterns in their datasets. While many existing VQSs promise to accelerate exploratory data analysis by facilitating this search, they are unfortunately not widely used in practice. Through a year-long collaboration with scientists in three distinct domains---astronomy, genetics, and material science---we study the impact of various features within VQSs that can aid rapid visual data analysis, and how VQSs fit into a scientists' analysis workflow. Our findings offer design guidelines for improving the usability and adoption of next-generation VQSs, paving the way for VQSs to be applied to a variety of scientific domains.
View on arXiv