This paper reports the results of two studies examining userexpectations of robust annotation positioning in modifieddocuments. The authors present an interesting case for one of theproblems encountered by users as we move our world ever deeper intothe electronic management and revision of documents.

The authors clearly define the terms needed to better understand theannotation positioning problem. The first issue they address is howdigital annotations work. The second issue is how documents aremodified. Starting with a discussion of the way paper basedannotations work, the authors then relate it to the digital annotationarena. They introduce the concept of an anchor and of a contextrelated to a content area, both of which are needed to create a robustannotation. They then discuss the issue of modification, and present amodification classification scheme describing three types ofmodifications: deletes, rewords, and moves. They describe what itmeans when an annotation is “orphaned” (when an online documentchanges and the link to the annotation is lost).

The authors include a good summary of the current research andchallenges in the area, including reporting on several studies ofusers using an electronic annotation system. They acknowledge many ofthe commercial systems already performing electronic annotations, anddescribe how they work. They provide a good set of references. Theirstudy extends the work of others further by delving into peoples’expectations in dealing with annotations, using those expectations tofurther fine-tune the annotation algorithms.

The article contains several interesting and excellent tables thatfurther explain the authors’ results and findings. Two trends seem toemerge from their studies: the greater the percentage of theannotation anchor text found, the more satisfied people are; and themore drastic the modifications to the anchor text, the less satisfiedpeople were when the annotation anchor was found. This second findingwas a little surprising. It seems that with significant changes, participants were all right if the annotation was orphaned. Keywords,proper names, and quotations are important factors to figure into theannotation algorithm.