Information visualization deals with the interpretation of data at different layers of abstraction; from the raw data, through statistical characterization/aggregation, pattern recognition and data mining techniques, to human comprehension and annotation. I will present techniques that can allow all of these layers of interpretation to live in the same chart or representation, yet be instantly and easily distinguished. In short, each layer is drawn with a distinctive "visual vocabulary" a term which has no rigorous definition. But I will also mention some of the basis in early vision and cognitive psychology that makes developing distinct visual vocabularies more understandable than blind guessing or vague esthetic feel. Most important, I will discuss the need to not just distinguish different layers, but visually tie them to the meaning a viewer is expected to extract; thus "background noise" or contextual information might recede, chosen subsets might be more distinct, statistical abstraction and ranges might hover, and human annotation might look like a sketched overlay.
Copyright. All Rights Reserved.