Integration of Archaeological Datasets Through the Gradual Refinement of Models
Presentation held at the CAA2014, session Ontologies and Standards for improving Interoperability of Archaeological Data :
The problem of integrating disparate archaeological datasets has received a great deal of attention lately. This is a relevant problem, and we do not dispute the advantages of being able to query and process information that is uniformly accessed from a wide array of distinct sources. However we observe that most, if not all, of the approaches to integrate archaeological datasets comprise extremely ad hoc efforts that suffer from one of the following two issues, and sometimes both. The first issue that we have observed is that, in order to integrate different datasets, a new, distinct model capable of catering for all of them is created. This results in the paradox of generating yet one more unique, non-standard view on the data. The second issue that we have observed is that, when carrying out the integration, individual datasets are mapped to a reference model such as CIDOC CRM , bridging the large gap that exists between the specific implementation (often at the database level) and the conceptual expression, which is costly and error prone.
In this paper we propose an alternative approach to archaeological dataset integration, which introduces the notion of the gradual refinement of models. In order to integrate a collection of disparate datasets, we propose that a conceptual model of each dataset is first created; these models must be extensions of a given abstract reference model such as CHARM , so that they capture the peculiarities of each dataset, but at the same time they share a common conceptual framework. Then, these particular models are examined for similarities, and more abstract models are generalised from those that are the closest. This can be repeated as necessary until, eventually, a hierarchy of models is finally obtained, and a number of views on the data are generated at different levels of abstraction, each one encompassing more datasets the more abstract it is.
The paper will illustrate this approach through the integration of three real datasets: a subset of the SIA+, the 20-year old, half-million object dataset that Incipit CSIC uses as their regular archaeological database; the «Camiño Primitivo» database, a purpose-built store also by Incipit CSIC for the cataloguing of sites and associated ethnographic elements along the Way of St. James in Spain; and the «Town and Country in Roman Essex» dataset offered by the Archaeological Data Service of the University of York, a Roman settlements database from Essex in the UK.
A particular model will be presented for each of the three datasets. Then an abstract particular model will be generalized from the two datasets belonging to Incipit CSIC, since it is assumed that, having been created within the same organisation, they should share a significant number of conceptual commonalities. Integrated querying and information processing will be demonstrated at the top level of abstraction, involving the three datasets, as well as at an intermediate level of abstraction, which involves only the two datasets at Incipit CSIC but provides much richer and more specific semantics.
- CHARM, the Cultural Heritage Abstract Reference Model: www.charminfo.org
- ConML, the conceptual modelling language in which CHARM is expressed: www.conml.org
Cesar Gonzalez-Perez & Patricia Martín-Rodilla
Institute of Heritage Sciences, Spanish National Research Council (Incipit, CSIC)