There are two different but related search tasks in the digital
domain. The first is exemplified by a search in a large digital
library. The user is only aware of a few items and does not (and
probably will not) grasp the full nature of the
corpus
(just as a library patron
doesn't know the full set of books the library has). The second type
of search task is more constrained, and answers the user's need to
find things in their own information space (i.e. his/her own
bookshelf, rather than the library). The bookshelf/personal model has
two unique properties that differentiate it from the library: a) the
user has direct control of what goes on the bookshelf, and b) the user
has direct control of the bookshelf's organization.
Until recently this second type of search task has not been a singificant problem. Users were restricted to limited disk space and limited bandwidth, and could search their corpus manually. Both factors contributed to limitations in the amount of information a user would ever be exposed to. Most systems were therefore designed to deal with the large corpora present only in library type systems. This class of library information systems can be further categorized by the intended audience. One class deals with the very general class of users (i.e. the Internet user base). The other deals with a very small number of specialized users (i.e. librarians or researchers). Unfortunately, neither approach completely satisfies, nor takes advantage of, information specific to a single user.
Today, we are rapidly approaching the end of the limits to personal disk space and bandwidth. Information is making its way quickly to -- and residing permanently on -- the user's desktop. While in the past personal information retrieval has been only an abstract problem, it has become a real issue today. In this thesis we will concentrate on finding information on the personal level. For our purposes, personal information corresponds to things that a user possesses in some form. The information can exist as physical paper, as digital bits, in the user's memory, or any combination of these forms.
What kind of questions might we ask about this type of information? Some examples are:
These questions are a small sample of what a user may ask when trying
to make sense of, organize, or find something in, their personal
information space. Can we apply previous methods of approaching the
general search problem to the personal situation? The short answer
is: not efficiently or effectively. A more detailed answer requires
an understanding of the tools currently available to us.
Section 
will describe a model for querying an
information system for object retrieval in a collection of data. In
section we develop categories for the different
components in this model. We then extend the model in
section for hybrid-search. Finally, in
section we will summarize the goals of this thesis.