TCDL Bulletin
Current 2006
Volume 2 Issue 2

What did the Others Say?
Probabilistic Indexing and Retrieval Models in Annotation-based
Discussions
Ingo Frommholz
University of Duisburg-Essen
D-47048 Duisburg, Germany
ingo.frommholz@uni-due.de

Abstract
Annotations, supporting different facets of Digital Libraries, can manifest themselves
in different forms and dimensions. In the form of shared, textual and nested
annotations they can be used to model discussions. Such discussions contain
additional information that can be interesting for both document and discussion
search. In my thesis I am going to develop a formal annotation model and, based
on this, discuss and evaluate probabilistic indexing and retrieval models for
annotation-based document and discussion search.

1. Motivation
There is a lot of ongoing research in the design and development of software
systems incorporating annotations. As a recent study shows [3], annotations support
different facets of Digital Libraries (like creation, management, access and
retrieval, and effective use) in several ways. Annotations can manifest themselves
in different forms and dimensions, ranging from simple text highlighting and
personal notes through (typed) links between documents up to nested and shared
annotations with which collaborative discussions about a specific topic are
realised. In the latter case, annotations and the annotated object form an
annotation thread. Such threads consist of (one or more) documents and nested
annotations discussing the documents and their content. Depending on one's
viewpoint, either a document and its annotations are equal or annotations are
What did the Others Say? Probabilistic Indexing and R... http://www.ieee-tcdl.org/Bulletin/v2n2/frommholz/f...
1 of 13 17/04/2011 14:46