CAA Conference

Archaeologists have long clamoured for better access to each other's data. The detailed and feircely contested structure of excavation recording is entirely predicated on the notion that other people will use not only our results, but any and every observation down to the Munsell colour of the topsoil. But examples of reworking data remain scanty, and the experience of working with other people's archives tends towards premature ageing. Many people have argued that this is because of the technological failures of paper and hybrid paper/digital archives and look forward to the day when we can cross corelate inclusions within primary pit fills right across Europe. The challenge, or course, is that while we can't practically answer detailed questions, our brains shy away from asking them. This paper will look at the process of questioning multiple data sets as it exists today to see which areas could do with better support. When are researchers asking questions of multiple sites? At what stages in ther careers and at what stages of a project? What kinds of information are they looking for and at what level of synthesis? Where are they when they are asking these questions, and to what extent are their questions spatially bound? How to they formulate, record and pursue these questions? How formal and clear are their search procedures? And perhaps most importantly who is asking questions of multiple sites?

Çatalhöyük is one of the most important and iconic sites currently under excavation. The data derived from the site has already spanned decades of ongoing work, and has been the focus of innovative theoretical, methodological and technological approaches. Efforts have also been made to, as much as possible, fully digitise and publish the excavation data for use and re-use in the future, including digitising the individual unit data (similar to an individual context, as used in the single context recording tradition), some of which can be accessed online through the www.catalhoyuk.com website. This includes information about the complex stratigraphy found throughout the site, along with comprehensive information associated with each unit. The nature of single context recording allows spatio-temporal relationships to be re-organised and recombined in ways compatible with the way data is structured using the RDF triple format, which is used by the Semantic Web. It is also compatible with the CRM-EH, which is a domain ontology for archaeology, based on the single context recording tradition.One of the most popular reasons for using Semantic Web principles and technologies is as a means to create interoperability between heterogeneous datasets, and successful exemplars of this now exist. In contrast, the data from Çatalhöyük is housed in a bespoke database to facilitate the nuanced intra-site analysis necessary for understanding the site's complexity. As such, there are other features of the Semantic Web, which may help to further unlock the spatio-temporal relationships in the stratigraphy of the site, which are not reliant on interoperability, and have seen fewer exemplars. This is largely owing to both the recent advent of the technology, and the structure and querying of graph data being unfamiliar to most archaeologists. Some of the complexity that might be better understood at the site, includes defining a 'deposit lifespan' for a series of units, which would be based on developing a system of coding time within a stratigraphic matrix. It may be possible to use the querying abilities of SPARQL (and potentially GeoSPARQL) to ask more complex questions of the stratigraphic data, and to allow the 'data haystack' model of graph data to allow easier combining and re-combining of 'deposit lifespans' for analysis under a variety of criteria. This exploration will be carried out with an eye towards how the results might be subsequently visualised. In addition, there may be potential for the use of Semantic Web 'inference' to allow the creation of new data which can be understood from the existing data, at a greater degree than might be possible using more traditional relational data structures.This paper will explore the potential research questions, which may be answerable using Semantic Web principles and technologies, using a subset of data from the Çatalhöyük excavations. It is hoped that this exercise will result in new exemplars of the usefulness of the Semantic Web to archaeology, and the potential for a new and more fluid understanding of the spatio-temporal nature of the occupation at Çatalhöyük.

Ancient texts, like ancient objects, may be regarded as sets of linked and linkable information. Yet, despite many conceptual similarities, there has been little examination of how the use of computational methods of marking-up and linking primary manuscripts can be used to inform the mark-up of primary material culture, and vice versa. Like archaeological contexts, discrete and philologically significant sections of manuscripts require skill both to identify and to record and define. Links between related information pervade archaeology: The Harris Matrix describes links between contexts, and the stratigraphic sequences between them, and database management systems have long been used to link information about artefacts and features across sites, and to enable cross-searching. More recently, approaches such as the CIDOC CRM and Semantic Web have been used to link defined entities of archaeological or cultural heritage information identified by URIs and described and linked using controlled standards such as RDF. This paper will examine the use of such standards and methods in archaeology, and focus on their transference to defining and linking related units of text in original manuscripts. Our case study is primary textual material from the Sharing Ancient WisdomS (SAWS) project. Comprising three international teams from the UK, Sweden and Austria, the aim of SAWS is to present and analyse the tradition of wisdom literatures in Greek, Arabic and other languages, which present complex challenges for linking. Throughout antiquity and the Middle Ages, anthologies of extracts from larger texts, containing wise or useful sayings (gnomologia) were created and circulated widely, since few complete texts were available in manuscript form. Focusing on original manuscripts of gnomological texts (not editions), SAWS uses a bespoke TEI XML schema to mark up individual segments identified by editorial experts, which are then linked to parallel segments in other traditions. Parallels include (for example) translations of individual sayings, derivations from a saying in one tradition to another, references to the same subject across traditions, authorship of sayings, and so on. Segments are then linked according to their significant properties, described according to an ontology that extends the CIDOC CRM, and linked using RDF. It is possible to regard such segments as artefacts, albeit of textual, rather than material, nature; and the parallels with archaeological information and sequencing are thus significant. While it would not be useful to impose a straight metaphor of 'textual artefacts' on this material, the theory that they are connected in complex ways owes much to material culture, and the latter's language provides clues for deeper interrogation: what typologies and common attributes can be applied to segments, how do these evolve over time, can one class of gnomic saying be demonstrated to have evolved in response to (and/or under the influence of) another. This paper will provide concrete examples from SAWS to demonstrate how the combination of XML, RDF and CIDOC are being employed; and thus delve deeper than existing secondary literature approaches to archaeological text mining.

The Archaeology Data Service (ADS) has a mandate to provide a digital repository for outputs from research funded by AHRC, NERC, English Heritage and other bodies. Archaeology has seen increasing use of the Web in recent years for data dissemination, and the ADS holds a wide range of datasets from archaeological excavations. However datasets and applications are currently fragmented and isolated. Different terminology and data organisation hinders search and comparison across datasets. Because of these impediments, archaeological data is rarely reused and re-examined in light of evolving research questions and interpretations. The STAR project addressed these concerns by developing semantic and natural language processing techniques to link digital archive databases and the associated grey literature, via an overarching framework (the CIDOC Conceptual Reference Model, extended for archaeological purposes by English Heritage). STELLAR aimed to generalise and extend the data extraction tools produced by STAR to facilitate their adoption by third party data providers. The extracted data is represented in standard formats to allow the datasets to be cross searched and linked by a variety of Semantic Web tools, following a linked data approach. As a result, the ADS has begun to ingest some of its excavation data into a triple store and expose it as linked data. This paper will briefly discuss the STAR and STELLAR projects to provide context to the ADS linked data. It will also outline the technologies used to develop the ADS triple store and linked data output. In particular, it will discuss the more practical details of creating our triple store, populating it with excavation data, and finally publishing it as linked data. This will hopefully provide some guidance for other interested parties who may want to set up something similar. Finally, with the ADS linked data as a concrete example, an overview of the suspected future directions will be outlined. In particular how we can enrich the existing and forthcoming linked data with both archaeological and non-archaeological data in addition to how the ADS linked data can enrich other data sets. In this sense we hope to identify some of the immediate and potential archaeological questions that can be asked and answered from our linked data.

Landscape and Identities: the case of the English Landscape 1500 BC - AD 1086 (EngLaID) is an ERC funded project running for five years at the University of Oxford which began during the second half of 2011. To analyse themes of change and continuity in the English landscape, the project will make use of many sources of data which cover England during the relevant period and have become available in digital form over the last two decades. This includes data collected as part of English Heritage's National Mapping Project (NMP), records registered with the Portable Antiquities Scheme (PAS), data collected from Historic Environment Records (HERs), as well as a number of datasets archived at the Archaeological Data Service (ADS). Combining these data from disparate sources presents a considerable challenge. Following a pilot project in 2009 using a small subset of data, we are turning to technology from the Semantic Web. We will be mapping data to RDF and indexing it within a triplestore; as well as providing a basis to query the collected data, this will enable us to keep track of the provenance of all of all of this collected knowledge. To the maximum extent possible, it is our intention to release the resulting dataset as linked data so that others can benefit from, and build on top of, the project's efforts. In order to structure the data, we intend to map it to the concepts defined by the CIDOC Conceptual Reference Model (CIDOC-CRM). At the time of writing, we are still early in this process, but we believe that it offers potential to provide consistency in our data which will assist in the searching and analysis There remain many challenges in delivering the technology to support this project including identifying and managing the many sites of artefacts which are represented in more than one of the available datasets; linking the work to use GIS tools to manage and visualise the geographic data with the Semantic Web tools, and representing and managing the differing degrees of precision and certainty which are present across the available data. We intend to present our design for the technology that will support the EngLaId project, and report on progress made and lessons learnt by March 2012.

This presentation reports on work undertaken to implement the CIDOC-CRM for the ArcheoInf project. ArcheoInf (www.archeoinf.de) is intended to act as a data repository, a database whichcombines and integrates primary archaeological data collected and processed during surveys orexcavations and stored in individual project databases. Although creating ArcheoInf presented anumber of difficulties (many related to terminology), the presentation focuses on an aspect of thecreation of metadata mapped to the CIDOC-CRM for export in RDF format using STELLAR. The"standard" STELLAR templates enable exports into the English Heritage version of the CIDOCCRM, but since ArcheoInf is not concerned with databases or projects using English Heritagedocumentation standards, these had to be modified.The presentation describes the process of mapping existing databases into the CIDOC-CRM andthen creating STELLAR templates for exporting the results of SQL queries into an XML/RDF format.

What do we imagine will be the archaeological remains of the Occupy Wall Street or Occupy London protest camps or events, in say 10 years, or 100, let alone 1000 or 2000 years? Will any evidence survive of where protesters occupied buildings, or indeed when the occupations began, or how long they lasted (a question yet to have an answer at the time of writing this abstract), let alone any idea of who was involved (protesters, homeless people, police, clergy, musicians, etc). While the possibility of an archaeological record of these events surviving my seem unlikely right now, it may be worth noting that buried below the current area of the St Pauls Churchyard occupation is one of the most commonly identified, and to some extent 'popular', layers for archaeologists excavating in London. This archaeological layer is commonly described as "the Boudican fire layer", a direct result of a riotous and destructive occupation and sacking of London in 60-61AD. This paper will explore some of the questions raised about how events might be recognised in the archaeology and recorded or interpreted by archaeologists. It will partly do so by considering the options, using methods currently available, for researching the connections between the site of the St Pauls Churchyard occupations and the relevant archaeological evidence for the events at the time of Boudica which have survived in the archaeological record (noting also that much of the evidence also derives from textual historical sources). Usually in current systems only data from single archaeological investigations, or multiple sites residing in a single database system, are queried or searched and even then queries do not amount to the complexity of, say, "show contexts which are burnt deposits containing finds of animal remains that are directly stratigraphically above a context that has been identified as a floor". This paper will explain how the STAR project (Semantic Technologies for Archaeological Resources) used semantic technologies, utilising an ontology driven semantic search system, to demonstrate the possibilities of searching both data and free text together, across multiple sites and including OASIS archaeological investigation reports. The techniques involved and results achieved may point to new paradigms for how archaeological research may be carried out, reported or re-used in the future. One major issue to be considered is that in order to accurately return all the possible data associated with more complex queries requires considerably more rigor and accuracy in some of the underlying recording systems, and in particular whether people are able to agree on shared meanings in their semantic terminologies. SKOS based thesauri can greatly aid the power of search techniques and enhance the capabilities of joining up data across different but related domains.