Different types of semantic information can be used in different ways. One aspect of this was semantic data which included organisational information about an institution and its research. This kind of semantic data had been used in previous work by the project partners from Oxford University, for instance to disambiguate names of researchers. Semantic information of this type is important and was used in project developments. However, it was semantic information relating to knowledge domains that was found to be of particular relevance. This arose out of one of the primary motivations for attempting to enhance existing tools. Although these were effective, they still were restricted by relying on the use of common terminology to detect matches. A key requirement to be considered therefore was whether connections could be found at higher conceptual levels. Dealing with this issue requires using structured knowledge representations, such as taxonomies and ontologies, which provide ways of representing more complex connections. How relevant ontologies could be derived thus became in turn a problem which needed to be addressed.
Using methods such as these, a concept map representation of relationships between topics and associated researchers could be created. The visualisation for this used a powerful open source tool, VUE, which also allowed additional semantic information to be included as necessary. An application was then written to automatically convert this to a suitable formal ontological representation, for instance using the SKOS (Simple Knowledge Organisation System) formalism, creating an ontological representation of the collective expertise of a network or community. Shown here is an example of part of the mapping for the SPIRES community.
How to quantify the connections in a structural knowledge representation was one of the questions that arose, and semantic distance techniques that allowed this to be done were investigated and implemented. Although these techniques could find connections between research which was already taking place, a question that arose was whether it was possible to find ways of identifying completely new areas for research. To do this the project experimented with techniques using ontologies as generative frameworks which specified the syntax and semantics of a “language” of research topics. The semantic tools and techniques developed were used with ConnectApp to find connections and as part of the wider integrated tools with the visualisation environment.