The document manager, developed as part of the
aims at supporting the user in managing and making sense of a personal
document collection. To specify the requirements for this tool,
one has to understand how people work with their documents. This
is a well researched issue, and from the studies published in the
literature and interviews with students and researchers at the DFKI,
we derived the following main requirements for the document manager:
- People use spatial organization to express relationships between
(physical) documents, so the document manager should provide means
to spatially organize document representations.
- When collecting documents, users often cannot immediately decide
where to file or how to classify these documents. The tool should
facilitate easy creation of informal structures out of newly discovered
- People tend to organize their documents into hierarchical structures.
The tool should support this activity.
- People annotate documents so it is easier for them to remember
what was important about the information at a later time. Hence,
the tool should allow users to associated an annotation with each
- Annotations are also used to express more specific relationships
between documents (e.g., "Document A describes a user evaluation
of the system presented in document B'"). The document manager
should explicitly support the definition and browsing of these
kinds of relationships.
- The tool should support the user in finding documents in the
These requirements led us to the adoption of a
two-dimensional, zoomable plane as the central interface component
of the document manager similar to system like, e.g., NaviQue. Representations
of documents can be positioned and grouped freely on this plane,
which facilitates spatial and informal organization.
In addition to allowing the user to associate with
a document text files containing in-depth notes, there is also the
possibility of defining detailed semantic relationships between
documents. Defining relations is as simple as a drag-and-drop interaction.
For example, if the user wants to express that document A describes
an improved version of an algorithm originally described in document
B, the user drags and drops the representation of A on top of B,
which will not move A on the plane, but initiate a dialog in which
the user can select the relationship she wants to define.
By creating group hierarchies and defining semantic
relationships, the user will build up a semantic net of the documents
in the personal collection. Thus, each document has a semantic context:
The adjacent objects in the net and their relationships to the document.
One immediate advantage of this net for the user is the ability
to browse it, but it is also a valuable source of information about
the user's current working context.
The study was performed with the ASL
Model 504 remote eye tracker; see also
the videos recorded during