Semantic Tools for Archaeological Resources

Increasingly within archaeology, the Web is used for making results and findings available as datasets. However Google and other web search engines are ill equipped to retrieve information from these richly structured databases. Important archaeological results and reports are also appearing as 'grey literature' on the Web, before or instead of traditional publication. Typically these are not indexed or made available for searching other than as ordinary web documents. It is difficult using conventional search engines to link across these datasets or to search them using terminology other than that employed by the authors. Different people use different words for the same concept or may employ slightly different concepts and this is a barrier to widening scholarly access.

This research is in collaboration with English Heritage (EH). The aim is to construct a Web Demonstrator to investigate and evaluate novel computer techniques for searching across archaeological databases and linking them to grey literature reports. This will open up these datasets for wider scholarly inquiry than immediate subject specialists. English Heritage is a collaborator in the project and contributes staff time and various datasets from the Raunds project, which covers a large area in Northamptonshire, focusing mainly on Roman material (with some Iron Age).

Building upon previous work at Glamorgan (the FACET Project), the research makes use of knowledge organisation vocabularies, such as classifications, thesauri and ontologies, which can be used to structure and connect different databases together. These will be combined with computer-based linguistic techniques to make links to grey literature reports that have not been indexed with controlled keywords from any knowledge organisation system. The research investigates the potential of a high level ontology, the Conceptual Reference Model (CRM), to bridge the terminology employed by very different databases and reports. The CRM covers cultural heritage generally and is envisaged as a 'semantic glue' mediating between different sources and types of information.

EH staff are known for work in digital archiving. However, the existing situation is one of fragmented datasets and applications, employing different terminology systems. Even simply expressed queries are currently difficult to answer, due to lack of tools for cross database searching. There is a need for an integrative framework and EH have built a core ontology based on the CRM. The intention is that a common ontology will provide greater semantic depth and potential for cross-domain searching by researchers within and beyond the archaeological sector. However work to date has focused on modelling. The proposed research will investigate the potential of combining this CRM-based ontology with query expansion techniques from the FACET project to assist archaeological inquiry.

Many archaeologists are familiar with detailed searches within single project datasets but currently such databases exist as isolated universes. A key aim is to show archaeologists (who may not be computer specialists) new possibilities for broadening such searches, linking datasets and grey literature. These outcomes have relevance beyond the immediate EH datasets to archaeology researchers more generally and beyond that to cultural heritage scholars attempting cross-domain search.

Evaluation will include a comparison of the Demonstrator with currently available functionality, taking account of cost benefit and utility issues. Two workshops, with EH and wider archaeological users, will assist evaluation. The first workshop will evaluate a pilot search system and feed into iterative design of the interface and demonstrator. The second workshop (held at the Archaeology Data Service) will evaluate the Demonstrator, as well as reviewing the outcomes and discussing further exploitation to digital archives and grey literature more generally.