Databases Reference
In-Depth Information
The area of “semantic desktop” research also creates web-like data that is relatively discon-
nected from the wider web. The various projects in this area are attempting to use semantic web
technology for desktop-based systems. They typically mine the user's files, email, address book, etc.
and put the information into RDF triples. This can then be explored using ontology-based tools or
used to augment desktop applications using plug-ins. For example, when viewing a file, one might
be shown links to the person who created the file, and through that email form that person.
This web of personal data is often called a personal ontology. It will start with standard classes
such as date, person, project, and event and may allow the user to add classes of their own, for example,
sub-classing 'events' to include 'raves.' Most importantly, the ontology's classes are populated with
the user's own data.
The personal ontology may also include links to external resources; for example, countries may
link through a GeoNames URI to DBpedia entries for that country. However, the links are one way
and other people would not normally be allowed to look inside each other's personal ontologies.
Clearly, there are times when it would be useful to be able to connect one's information, for
example, in a commercial organisation or even at home to link information on friend's addresses.
However, there are technical and user interface barriers. At the technical level, some of the classes
that might be found in a personal ontology, such as 'Family' is egocentric; Tiziana's 'Family' is not the
same as Alan's 'Family.' Mapping between ontologies is a difficult problem for knowledge engineers,
let alone ordinary users. At an interaction level, one needs to be able specify what can and cannot be
shared. For tabular data, it is relatively easy to allow access to a particular collection or for hierarchies
