A Systems View of Annotations

Motivated by our design of an annotation system (Notable), we have developed a framework for annotation systems, based on dimensions of the platform used to create the annotations, the platform used to read them, the annotations themselves, the targets of the annotations, and the correspondence between annotation and target. We demonstrate the framework by using it to compare 10 different annotation systems, and to find unexplored possibilities in the design space. INTRODUCTION The notion of ‘annotation’ refers to many different practices. It refers to highlighting words in a text, placing marks or words in a margin, attaching PostItTM notes to objects, publishing an annotated edition of an important work, etc. The term ‘annotation’ has also been used to describe more formal activities, such as assigning metadata to a literary work. As Marshall points out, even a photograph of a sky tucked into a meteorology textbook might be called an annotation [3]. With the growth of digital media, the document tradition is moving forward while the annotation tradition is being left behind. This paper sheds light on the space of annotation systems, in order to help designers of annotation systems see their way forward. The move towards electronic documents only makes it more difficult to say precisely what is meant by ‘annotation,’ and therefore to understand what is expected of a system that provides an annotation facility. Some writers argue that appearance in situ is a necessary characteristic of any successful annotation system [6]. Others argue that the location of an annotation merely affects the style of the annotation. For example, studies of voice annotations have shown that independent annotations such as comments on a paper left on voice mail tend to be higher-level comments than those written directly on a paper [1]. There is room for a range of digital annotation systems that support both in situ and independent annotations. At the former federal penitentiary Alcatraz, visitors are given devices that play audio annotations specific to various landmarks within the facility. While this is far from a traditional annotation platform, it gives a hint about how much technology changes our view of concepts like annotations. This particular system highlights a subtle point: the system used to author the annotation does not need to be the same as the one used to render the annotation (the technology used at Alcatraz, tape players, does not even support recording). We have developed a broad definition of annotation that encompasses these examples and many more: Definition of annotation: An annotation is a commentary on an object that: • the annotator intends to be separable from the object itself • the reader interprets to be separable from the object itself With this definition objects, as well as documents, can be annotated (imagine a sticker on a computer monitor that says “broken”). There are three roles involved in the practice of annotation: the author created the object that is being annotated; the annotator is the one commenting on the object; and the reader is the agent making sense of the commentary. The same person often fills two or more of these roles, and sometimes a role is filled by a group of people. When a student highlights a textbook for later study, she is both annotator and (while studying) reader. When a group of people pass around marked-up drafts of a paper they are jointly writing, they fulfill all three roles. This paper takes a broad, systems view of annotations. The primary contribution is a framework for comparing annotation systems, both paper-based and electronic. The analysis is motivated by an annotation system we have developed (Notable) that separates the annotation “platform” from the object being annotated, and uses search to bridge the gap between the annotation and the annotation target.