Transdisciplinary approach to archaeological investigations in a Semantic Web perspective

1.Introduction

Archaeological investigations have been relying more and more on reflexive methodologies [15]. Nowadays, making sense of archaeological investigations starts its journey in the excavation site and continues up to museum curatorial practices, accompanied by labels in exhibitions and records in digital repositories and archives. In fact, though interpretations still rely upon the expertise of the excavation team [28], the trend is to carry reflexivity to its extreme through the video recordings of initial sense making during the excavation and producing daily reports by using web-based interfaces, up to filling the data base entries for the excavation. This documentation, which can also be accessed later, reveals much of the background to the interpretations. The audiences, as well as other scientists, can query the data and evaluate conclusions.

The other methodological issue that characterizes the current conduction of an archaeological investigation is the contribution of archaeometry, acknowledged by many archaeologists as an essential and integral part of archaeology. Archaeometry involves the development and application of natural scientific methods and concepts to the solution of cultural-historical questions. Although applications of natural sciences in archaeology have actually a long tradition (e.g., “the quantitative analysis of Roman coins in 1799 by Martin Heinrich Klaproth in Berlin”), archaeometry is archaeology by ultimate aim, but natural science by approach. It includes all the disciplines that may contribute to archaeology (e.g., physics, chemistry, biological sciences, anthropology, geological sciences), by measuring and evaluating facts and interpretations [1,33].

However, as archaeology, with the growing contribution of archaeometry, becomes fragmented into specialized areas of knowledge, challenges to achieve an integrated interpretation increase. The individual archaeologist interfaces with the recording structure, which supports access to reflection and dialogue with all the members of the project; additionally, the challenge is to realize a holistic view of the data, with interpretations about findings, stratigraphic units, or sites to be developed in broad contexts, satisfying historical and natural scientific constraints [6,41]. Although problems derived from “faultlines between field and laboratory staff or from the practical separation of ever more complex forms and types of data” [3] have been acknowledged in digital integration, the adoption of digital technologies and methods in the field (such as GIS and 3D visualization on tablets) has led to a maturing and expansion of the reflexive objectives.

In a number of cultural heritage areas, digital data curation (or DDC) has emerged as a viable workflow for the management of the related digital assets during their entire lifecycle [42]. It consists of “actively managing data […] with the aim of supporting reproducibility of results, reuse of, and adding value to that data, managing it from its point of creation until it is determined not to be useful, and ensuring its long-term accessibility and preservation, authenticity and integrity” (Digital Curation Center – DCC11). In archaeological investigations, the digital assets can be more or less formal descriptions of artifacts and of the excavation context (stratigraphic units and preliminary interpretations), curated by archaeologists, or measurements of some physical parameters that reveal some hidden property, resulting from some archaeometric investigation [18]. Data recording sheets enable the recording of excavation outcomes in archaeological databases; however, the interpretation (e.g., the classification of some artifact or the estimation of some chronology) proceeds in incremental phases and, also given the contribution of archaeometric methods, can be subject to revisions. The research goes through a truly transdisciplinary endeavor, where research questions arise through the collaboration and peer-to-peer cross-fertilization of several disciplines [27]. At the same time, datasets are increasingly available online: projects such as, e.g., the Digital Archaeological Record,22 the catalogue section of the Central Institute of Cataloguing and Documentation of the Italian Ministry of Cultural Heritage,33 and the Archaeology Data Service44 make a number of archeological data available for quantitative testing and processing, and these data are reused by other researchers in novel ways (see, e.g., [37]).

However, most datasets are actually isolated from one another; some researcher also reports no connection to grey literature (the so-called unpublished excavation reports), and there is a demand on semantic interoperability between differing database structures and terminology [34]. Semantic interoperability is also called to overcome some of the limits that have been raised for IT applications in archaeology, which, while appointed to bring some data-driven theory-neutrality to archaeological investigations, have been appraised as “unrealized ‘great expectation”’ [26].

In this scenario, the Semantic Web approach has been invoked to support the sharing of data, particularly for the transdisciplinary endeavors [19]. In recent years, some projects have provided access to collections of archaeological data through the integration of knowledge organization systems/services (KOSs),55 conceptual frameworks such as the Dublin Core Metadata Initiative (DCMI),66 the CIDOC-CRM conceptual reference model.77 Project ARIADNE (Advanced Research Infrastructure for Archaeological Dataset Networking in Europe) relies on these ontological tools and models to enable the sharing and re-use of about two million archaeological datasets.88

However, according to our knowledge, the representation of the archaeometric processes as well as a modern and transdisciplinary conception of the archaeological endeavor at large have not found their way through the Semantic Web endeavors. This paper presents a conceptual model and ontology for supporting this transdisciplinary conception of the archaeological investigations, at the crossroad of many archaeometric disciplines, contributing to its reflexive methodology in the context of an encompassing digital curation of the data. In recent work, we have proposed an ontology-based approach for the encoding of the semantic knowledge underlying the archaeological forms to be filled for the documentation of the excavation and the interpretation phases [21], related to ongoing EU project “Beyond Archaeology” (BeArchaeo99), which consists in an archaeological excavation, the consequent interdisciplinary archaeometric analyses of the site and the excavated materials, the interpretation of the findings, and the dissemination of the results through physical and virtual exhibitions. Here, we address the overall ontological approach, which specializes the CRMarchaeo model [10]1010 and the CRMsci model [11],1111 of the CIDOC-CRM family.

The paper is organized as follows. In the next section, we report on the related work about the digital approach to archaeological data, with particular reference to their semantic organization. Then, we introduce the general context of the digital data curation and BeArchaeo, a DDC-born archaeological project. The core of the paper is the description of a comprehensive approach to the conceptualization of the archaeological and archaeometric domains, at the base of a transdisciplinary approach to archaeological investigations. Running examples are taken from the BeArchaeo project, carried on with a semantic organization of the data in support of the coordination of all the tasks, from the excavation planning to the final exhibition of the results.

2.Related work

Archaeological projects go digital in all their phases: data collection, curation, and visualization (see, e.g. [20,36], among others), analysis (e.g., GIS [7]), exhibition (starting from the virtual archeological reconstructions of the 1990s [2,32] and addressing general public outreach and participation [35]).

A particular mention goes to the pioneering Çatalhöyük project, concerning a Neolithic settlement in Turkey, carried out with the goal of maintaining the data as long as possible. The Çatalhöyük Database and the Çatalhöyük Image Collection Database1212 make the documentation of the Çatalhöyük excavation site available. Custom platforms allow for the search of data uploaded during every excavation season and then made available through the Çatalhöyük Living Archive, which tells about two decades of excavations and analyses.

Project ARIADNE provides an event-centric ontological representation of the archaeological excavation relying on CRMarchaeo and CRMdig ontologies [23]. However, the legacy of the ARIADNE project, which currently continues with ARIADNEplus, is to be a web of interlinked archaeological datasets that comply with the Linked Open Data principles. The effort required to project partners is to convert and work with data in the (not always familiar) Semantic Web formats. In fact, a large amount of digital data demand for the coherence of the recorded information, as contributed by each intervening discipline.

However, even across projects within single institutions, the global picture is a “rather disparate grouping, or ‘archipelago’, of diverse, specialized, but rather isolated and independent information systems and databases” [9]; limits concern sharing and standardization of data [8]. Also a survey made within the ARIADNEPlus project1313 reports that researchers are not very aware of the issues of data sharing and Linked Data. Linked Open Data are also advocated to encourage the dissemination and the linking of archaeological datasets [17]. The motto “data sharing as publication” promotes an initiative to publish data and resources from archaeology after review by an editorial board and to integrate data through some (simple) ontological model. Integration and sharing of data through the instantiation of acknowledged ontologies support the major challenge archaeologists have to face, namely data reuse [12]. Kansa and Kansa get to promote a general “data literacy” for archaeologists, who should care personally for their own data, through direct management and communication [16].

There have been some semantic approaches, especially in the context of the reflexive methodologies, hence requiring some knowledge to interconnect objects, events, and people, historical context and excavation process [4]. CIDOC-CRM ontology has been employed to deal with interpretations as events that occur from the excavation process and can occur later again, when initial interpretations are revised or integrated, in the context of the long running Çatalhöyük project [22]. In this case, CIDOC-CRM worked as the backbone for a digital counterpart of a more conventional print report, emphasizing the need for time-consuming data cleansing with typical archaeological datasets. One of the most relevant takeaways of the analysis was the need for a publishing platform, where the complex and massive content could be inserted and accessed through user-friendly interfaces.

An indirect use of CIDOC-CRM data model is through the Arches platform [24], on which a number of projects are based: for example, the two projects, namely EAMENA (Endangered Archaeology in the Middle East and North Africa)1414 and ASOR (American Schools of Oriental Research) Cultural Heritage Initiatives for Syria and Iraq,1515 which record archaeological sites and landscapes that are under threat or damaged across the Middle East and North Africa, with goals of documentation, sharing information, and planning responses. Arches manages six resource types: heritage resources (such as archaeological sites or buildings), heritage resource groups (e.g. urban districts), actors (e.g. persons or organizations), historical events (e.g. floods or epidemics), activities (e.g. investigations), and information resources (e.g. media files). The data model of Arches builds on CIDOC-CRM and other interoperability standards, such as the Open Geospatial Consortium (OGC) with its encoding standards (e.g., Earth Observation GeoJson) and system integration interfaces (e.g., WMS – Web Map Service), which ensure compatibility with GIS applications (e.g., ArcGIS and Google Earth), common browsers, and online map services. Also, Arches includes modules for vocabulary management, such as Getty Art and Architecture Thesaurus.1616

3.Digital data curation and the BeArchaeo project

Digital data curation consists of the coordination of the representation and management of the digital assets related to cultural heritage, i.e. tasks as selection, processing, preservation, maintenance, collection, and archiving of the digital, with possible added value for subsequent exploitation [42]. The notion of digital data curation has been revised and updated several times, with a recent focus on motivations and big data [30]. To systematize goals and practices of digital data curation, a number of models have appeared in the literature from many institutions, such as, e.g., Digital Curation Center Curation Lifecycle Model [14] and I2S2 Idealized Scientific Research Activity Lifecycle Model [29]. Here we describe the digital data curation through an abstract representation of the tasks, adapted from [18].

Fig. 1.

Abstract representation of the digital data curation model.

3.2.Application of the model to a BeArchaeo example

Now we illustrate this model of digital data curation with an example that is related to some digital data generated from an archaeological finding during the BeArchaeo DDC-born archaeological project. The project carries out an archaeological excavation and the related archaeometric analyses of the Tobiotsuka Kofun, located in Soja city in Okayama Prefecture of Japan. Together with other Kofun burial mounds and the related archaeological material in ancient Kibi and Izumo areas, researchers aim to develop a transdisciplinary vision in studying the archaeological site and other archaeological materials now stored in museums and laboratories, in Japan.1717

The project activities and outcomes are accessible to the general public through engaging media communication along the project development. In this section, we apply the proposed digital data curation operational framework for ongoing activities of the archaeological discoveries, scientific interpretations and the related database.

Fig. 2.

Digital data curation model applied to the archaeological finding SH1 in the BeArchaeo project.

Figure 2 instantiates the general model above on one operational workflow addressing the digital data originated since the discovery of the archeological finding named SH1, undergoing a specific investigation path, at the current stage of development. As we have seen above, interpretations are recorded in some digital format and then revised or updated, also encoding other formats, going formally when possible.

The conceptualization of the knowledge in the BeArchaeo project is driven by the design principle of recording the archaeological/archaeometric activities and the collected data that occur both on the archaeological site and in the lab. The data are recorded in a database filled by the scientists in order to be employed in interpretation processes and exhibition organization. The goal of the digital data curation is to support the scientific research on the composition of the findings and to examine their relation with the question of their similarities and differences. In this specific example, the research question is to find the provenance of a set of similar potteries through a comparison of the component materials, including elemental composition, morphological features, presence, typology and composition of inclusions such as minerals or rock fragments.

The digital data curation workflow starts as soon as SH1, an archeological finding fragment, has been found. In particular, Fig. 2 addresses a measurement carried out in the lab, where scientists acquired images of the fragment by Scanning Electron Microscopy (SEM), coupled with Energy Dispersive Spectroscopy (EDS). The process generates raw data (a magnification is shown in the figure, jpeg file format). The task of data modeling and processing enriches raw data with metadata that reveal a feature of the asset at some level (e.g., the possible presence of a surface coating). Elemental maps of a portion of the sample, which are visible in the figure, highlight that the coating is depleted in Al2O3; later, it may suggest an enrichment in iron compounds, which would indicate that a coating was actually present. Such information derives from the combination of different scientific tests and different expertises. In a digitally-born project, the need to harmonize the procedures strongly supports the synergistic interaction. An example, which we can use for sake of simplicity, can refer to the archaeological question of defining if an archaeological finding (e.g., a pottery fragment) may share a common origin with other fragments that have been found in other archaeological sites. The question can be faced, as a first instance, by determining the elemental composition of the fragment. Presently, it has been determined by induced coupled plasma optical emission spectroscopy (ICP-OES). Raw data must guarantee interoperability and reuse; then, the acquisition step must guarantee that all the information on measuring conditions and procedures is recorded (as also stated in [25]). The processing and modeling step produces the information on the quantitative elemental composition of the sample, ensuring a high-quality base for data interpretation.

In the interpretation step, we can compare the elemental chemical composition of the fragment with the compositions of other fragments, so that the hypothesis of a common manufacture can be discarded or supported, respectively. In the latter case, we can go on with building the multidisciplinary knowledge by including, in the decision process, further items from the investigations with other scientific techniques (such as optical microscopy or mineralogical/petrographical data) which can lead to discard/support the interpretation made with elemental analyses data. A single operation of data acquisition plus processing and modeling can be included in many interpretation processes, supporting reflectivity and fertilizing interdisciplinarity. The intermediate and the final data are stored into the repository, currently a Google drive shared folder (to evolve into a more effective data repository connected to the database), through the tasks of Data curation and preservation. Moreover, the interpretation, in the format of Microsoft Powerpoint slides, is also selected and stored, as part of the Data documentation and archiving task, into the BeArchaeo Archive, namely a MySQL database, underlying an Omeka-S installation, which also works as centralized database for the coordination of digital data curation. The model will also be enriched with further metadata (e.g., the digital image also receives the identifier of the physical fragment). The database schema design as well as the organization of the Google drive folders are based on the proposed semantic model worked out after the conceptualization phase, to ease the problems of interoperability and connection between the archeological and the archaeometric data.

Finally, in order to make the knowledge available to the archaeologists on the field, a BeArchaeo project website, based on the mentioned installation of the Content Management System (CMS) Omeka-S, is available. The recording of the archaeological findings and forms as templates are made possible through a web-publishing platform that allows for the import of semantic properties defined in a RDF file, the definition of customized vocabularies, and the construction of templates for the instantiation of filling forms [21].

Related to these concerns and potential interpretations, the database design of BeArchaeo project provides the information structure to all the digital curation phases of the project. In this case, it provides a repository while creating the archive of the archaeological findings with the related media. Media and metadata are stored in the BeArchaeo database as Archaeological Finding form, interfaced by an Omeka-S based web platform, in order to support the archaeologist’s work in recording the excavation and interpretation activities.

4.Transdisciplinary conceptualization of the archaeological/archaeometric investigations

Given the digital data curation schema, which involves a conceptualization addressing several disciplines, we have developed the BeArchaeo ontology, with the design principle to capture the connections between the archaeological and the archaeometric realms, respectively. Transdisciplinarity is mediated by the formal ontology, with research questions arising from the collaboration between the disciplines [38]. The BeArchaeo ontology pivots on the description of the objects, and merges the general archaeological and archaeometric entities with the fields of the catalogue records [21]. Design patterns, for connecting these knowledge domains, are not available (to the best of our knowledge). The result is an application ontology that merges three types of knowledge: the archaeological knowledge (lower left part of Fig. 3), the archaeometric knowledge (lower right part of Fig. 3), and the catalogue record knowledge (upper part of Fig. 3).

Figure 3 provides an overview of a sample encoding. Going left to right: the stratigraphic unit “SU 202” (content of the title field of the catalogue record for this unit) is the source of the archaeological finding “AF 59” (content of the title field of the catalogue record for this finding); the type of the finding is “Sue (ceramics style)”, as selected from the Getty-AAT thesaurus and “sekki”, as selected from the BeArchaeo thesaurus; the finding body1818 has undergone some chemical test for calcium oxide (CaO, a measurement activity), which has produced a result in wt% value. A data evaluation process assigns some dimension, namely an attribute for the body predominant composition (“Calcareous”).

Fig. 3.

Modeling of the archaeological finding “AF 59”, exemplifying archaeological and archaeometric knowledge, respectively, and the corresponding fields in the archaeological finding record. The rectangles in grey or black are the individuals; the white rectangles are the classes; object properties link individuals, datatype properties link individuals and strings; the three elements in Courier font are the strings that are actually written in the final form interface.

The realization of the BeArchaeo ontology relies on the CIDOC-CRM reference model family. The pyramidal CIDOC-CRM family of models (Fig. 4, right1919) extends the general documentation model (entities identified with prefix cidoc-crm) through specialized thematic models for the needs of projects and organizations. In particular, CRMdig is a model for provenance metadata, CRMgeo is a model for spatio-temporal entities. Of particular interest for the archaeological and the archaeometric endeavors, we address the CRMsci and the CRMarchaeo models, respectively. We plan to deal with an ontological model of provenance in the future; currently, we have encoded provenance in the notes of the investigation processes (see Fig. 8).

Fig. 4.

Major relationships between BeArchaeo and CIDOC-CRM family.

In Fig. 4, we can see the overall picture. The ontological module of the classes are identified through the prefixes CRMsci, CRMarchaeo, and cidoc-crm, respectively; BeArchaeo classes have no prefix. The figure illustrates the major relationships between BeArchaeo ontology and the CRMsci and CRMarchaeo reference models, as well as the references to the two archaeological thesauri BeArchaeo-AFT (Archaeological Finding Thesaurus), for a taxonomy of Japanese history materials, built within the project, and Getty-AAT (Art and Architecture Thesaurus). The major classes are bearchaeo/ Archaeological_Finding and CRMarchaeo/A8_Stratigraphic_Unit, which describe the objects that tangibly connect all the tasks related to an archaeological investigation (a stratigraphic unit is the source of some archaeological finding or at least of some inclusion, a fragment of some material that is relevant for the investigation). They are connected with the related catalogue records (bearchaeo/AF_Catalogue_Record and bearchaeo/SU_Catalogue_Record), which describe the respective objects. Class bearchaeo/Archaeological_Finding specializes class cidoc-crm/E18_Physical_Thing and has a type, which refers to the specialized vocabularies, Getty-AAT and BeArchaeo-AFT.

CRMarchaeo reference model takes inspiration from Harris’ model [13], which accounts for the stratified arrangement of an archaeological excavation. The excavation model includes the description of the dichotomy between the (natural or human) phenomena that produced the stratification (centered around the class CRMarchaeo/A1_Excavation_Process_Unit) and the units that are the outcome of the generation/modification process (centered around the class CRMarchaeo/A8_Stratigraphic_Unit). Stratigraphic units contain some remains, classified as physical objects (centered around the class cidoc-crm/E18_Physical_Thing of the core ontology). Stratifications and their contents are analyzed and interpreted to determine the relative chronological order of the strata, then the classification and functionality of the objects therein, up to the high-level reconstruction of the beliefs and behaviors of some group of people in the past in that place. A stratigraphic unit, produced by some genesis process (CRMarchaeo/A4_Stratigraphic_Genesis), can also be modified by a bearchaeo/A5_Stratigraphic_Modification, of which formation process types, acknowledged by the official excavation recording forms, are a specific vocabulary.

Archaeological findings, as physical things, can be the object of a task CRMsci/S19_Encounter_event (an archaeologist encounters a finding in a stratigraphic unit). Physical things are a subclass of observable entities (class CRMsci/S15_Observable_Entity), which can be observed (specifically measured), producing values (any cidoc-crm/E1_CRM_entity) for some property type (class CRMsci/S9_Property_Type). The data collected can be evaluated (class CRMsci/S6_Data_Evaluation) for the assignment of some dimension (property CRMsci/O10_assigned_dimension) to the archaeological finding (check the description of the digital data curation for the example SH1 above).

5.The BeArchaeo ontology

The conceptualization described above has been enriched with specialized vocabularies for supporting the digital data curation process of an archaeological investigation. As observed through the example in Fig. 3, the development of the BeArchaeo ontology comprises three modules, the archaeological knowledge, the archaeometric knowledge, and the catalogue record knowledge, with connections to standard ontologies and the inclusion of non-ontological resources. In particular, the third module concerns the form through which the first two modules are recorded for the digital data curation process. In the rest of this section, we address the major decisions for the ontology modeling process and then we provide an overview of the classes and properties of the BeArchaeo ontology.

5.2.Overview of BeArchaeo classes and properties

Now we provide an overview of the archaeological and archaeometric modules; the classes and properties of the catalogue record module, sketched in Fig. 3 reflect the entities presented here and are accessible through the web platform interface implemented for the scientists to insert their data during the excavation and the laboratory work (Fig. 11).

Fig. 5.

Conceptual model of the stratigraphic unit knowledge (including references to thesauri and vocabularies (with list of terms)).

Fig. 6.

Conceptual model of the archaeological finding.

5.2.1.The archaeological module

In Figs 5 and 6 there are the classes, vocabularies, and properties concerning the description of the stratigraphic unit and the archaeological finding, respectively. Going clockwise, a stratigraphic unit has inclusions (i.e., entities that are contained in the stratum), which are of some type, that can be generic or specific, and have a frequency of occurrence in the unit, qualitatively valued as rare, medium, or frequent. Inclusions have types that are taken from partially overlapping vocabularies, based on the practical experience of the archaeologists (these may change and should be aligned with the types included in the thesauri for the archaeological findings). Some informal properties, noted as free text, are the state of preservation of the unit and the measurements taken during the excavation, with a particular concern for Elevation. The distinguishing criterion determines how this unit has been identified: the terms that concern this attribute are three (Color, Composition and Compaction) and there are other three properties that possibly specify the actual values for such attributes (namely 6-valued soil/matrix term for composition, 5-valued term for compaction, and a free string for color). Color, in the relationship with archaeometrists (specifically, the soil scientists) can be recorded with the encoding provided by the well-known Munsell color system, in use in pedological studies.2424 Finally, the formation process concerns a specialization of the processes that are responsible for the creation and modification of the stratigraphic unit, with a frequent term vocabulary, which can be further augmented with free text insertion. The properties in the center of the figure specialize the stratigraphic relation property (CRMarchaeo/AP13_has_stratigraphic_relation):

• – sameStratumAs, for two stratigraphic units that are claimed to belong to the same stratum of soil interrupted by some intervening unit;2525

• – isBoundTo, for a stratigraphic unit that is a limit for another one;

• – abuts/isAbuttedTo, for a stratigraphic unit that edges another one;

• – cuts/isCutBy, for a stratigraphic unit that introduces a discontinuity into another one;

• – covers/isCoveredBy, for a stratigraphic unit that covers (stands over) another one;

• – fills/isFilledBy, for a stratigraphic unit that has filled a cut (see above);

Also, there are two temporal relations, laterThan and earlierThan, resulting from the interpretation of the stratigraphy. The latter terms, which originate from the terminology reported in the institutional records of the excavation recording, shall be later aligned with some general temporal ontology.

An archaeological finding (Fig. 6) can be part of another archaeological finding (frequent is the case of fragments to be composed afterwards) and is sourced by some stratigraphic unit as well as museum collection or other places. This variety of sources concerns the goals of the BeArchaeo project (and many other projects), because of the employment of the ontology into the design of the final exhibition. The archaeological finding has a reference type and some component material. Types refer to terms in the previously mentioned Getty-AAT thesaurus and the BeArchaeo-AFT thesaurus, the latter encoding knowledge from an authoritative Japanese reference [40]. Also the component material has a type (referred again in Getty-AAT) and the information about the administrative location. Finally, an archaeological finding is marked with its chronology, currently limited to a free text insertion, together with its motivation, but with the idea of providing an encoding in the terms of a time ontology, with possibly many alignments, depending on the disciplinary traditions in both archaeology and archaeometry.

5.2.2.The archaeometric module

Archaeometry is a vast endeavor. As far as we know, this is the first attempt to model the archaeometric investigation in a digitally-born archaeological project. We want to keep record, in the digital data, of the decisions made during the analysis (going from acquisition to processing and interpretation) and to relate the archaeometry-based interpretation with the evaluations, data, and interpretations conveyed by the archaeologists. The focus of the project is on the documentation and dissemination of the results; in the future, we plan to also address consistency and inference between the disciplines participating into the endeavor, with the semantic web encoding.

The current development of the BeArchaeo archaeometric module implements a trade-off between a wide appraisal of the archaeometric domain, with its processes and data formats, and the needs of the BeArchaeo project, which addresses a restricted set of archaeometric investigations in detail. However, the alignment of the archaeometric module with the CRMsci standard model and the richness of the multidisciplinary team working on the project provide us a wide scope. Now, we first address the conceptualization of the archaeometric model; then, we give an insight on the ontological model; finally, we illustrate two paradigmatic examples.

5.2.3.Conceptualization of the archaeometric model

The goal of the conceptualization phase for the archaeometric module is to provide a coherent and cohesive structure for all the archaeometric investigations, which work in a transdisciplinary setting, mutually influencing one another. The several disciplines specialize the CRMsci reference model through the specific processes and the corresponding digital data formats. The disciplinary researchers have been asked to speculate on the procedures and results concerning the stratigraphic units and the archaeological findings, in order to single out the concepts that are related to their disciplinary contribution to the overall investigation. Each monodisciplinary team has thus deeply reflected on their own procedures, data formats, and knowledge contributions. After that, the broad group of researchers have discussed the links that could have been set among the diverse monodisciplinary outcomes, in order to enhance the overall knowledge in a transdisciplinary perspective. So, they carefully selected the entities supporting the inferential processes from data, in order to include them into the conceptual model. Finally, they tackled the challenge of conceptual modelling according to a common formal structure based on core CIDOC-CRM and CRMsci models.

Figure 7 shows a portion of the upper level structure of the measurements that occur in the archaeometric domain, when dealing with the archaeological findings. BeArchaeo archaeometric measurements are all subclasses of CRMsci/S21_Measurement; classes are distinguished by the object measured (archaeological finding or stratigraphic unit), the measurement technique (e.g., Polarized Light Microscope, Thermoluminescence, Archaeomagnetism, Metabarcoding of microbial taxonomic diversity), and the material addressed (e.g., pottery, glass, organic remains). Specialized vocabularies identify the observed property types and, for each measurement, the observed values. Measurements are typed and also connected to some entry in the Getty AAT thesaurus (if this exists). For example, Fig. 8 shows an instance of a measurement class concerning the X-ray Fluorescence Spectrometry (XRF), applied to the Archaeological finding “BA18”. XRF has a type in the Getty AAT (300224161).

Fig. 7.

Overall model of the BeArchaeo archaeometry ontology.

Fig. 8.

Instance of the XRF measurement acquisition.

All measurements rely on a number of factors, such as environmental conditions, the actual device, with its settings and calibrations, precision, and scale. Following the indications provided by the CRMsci reference, this information is reported in a note, currently a string datum, connected through the cidoc-crm/P3_has_note property. Figure 8 reports the note for the XRF measurement, consisting of, e.g., the instrument that made the measurement, the voltage utilized, the beam size, and the number of acquisitions that have been done. As noticed, measurements address the acquisitions in the digital data curation pipeline, producing the so-called raw data (Fig. 1). So, we include such information into the catalogue record designed for the object. The same considerations hold for the processed data, where algorithms and software libraries are determinant for the achievement of the results. We are aware that a note is not the best solution for these relevant metadata and the connection to data provenance ontologies, such as CRMdig or PROV-O, is to be deployed in the near future.

We have currently developed classes and properties for archaeometric analyses such as: Polarized Light Microscopy, elemental chemical analysis by X-ray fluorescence (XRF) and induced coupled plasma optical emission spectroscopy (ICP-OES), molecular chemical analyses by Raman spectroscopy and Diffuse Reflectance Spectroscopy, Thermoluminescence dating, Archaeomagnetism, Soil morphological assessment, Radiography, Tomography, and Metabarcoding of microbial taxonomic diversity. In each case, we have developed specific vocabularies, geared to the project specificity. The alignment with external, comprehensive resources is planned for the near future.

To illustrate the depth of the knowledge encoding, we show the ontology developed for modeling archaeological pottery investigation by means of morphological qualitative methods (Fig. 9), in particular Polarized Light Microscopy. Analogous ontological models have been deployed for the other archaeometric processes mentioned above; below, we also show how the several investigations converge on the evaluation for achieving an interpretation.

The model is based on the annotation structure suggested by Quinn for the investigation of pottery prepared as thin sections [31]. The transdisciplinary value of the conceptualization is that the scheme has been adjusted to match the investigations carried out by the many disciplines involved in the archaeometric investigation of pottery findings. In particular, the model fleshes out the similarities spanning the diverse disciplinary procedures, by replicating the same major structure developed for modeling the analyses of thin sections in pottery investigation to other scientific tests. It models, in particular, 1) the investigation in cross section of pottery, 2) the determination of qualitative chemical composition by XRF in glass and pottery, 3) the investigation of inclusions by Scanning Electron Microscope in glass, 4) the spectroscopic investigation of glass through Diffuse Reflectance Spectroscopy.

Fig. 9.

Investigation of archaeological pottery prepared as thin sections through polarized light microscopy.

The analysis by polarized optical microscope of the archaeological ceramics in thin section reveals the complexity of these materials (Fig. 9). They are composed of three main components (inclusions, matrix and voids), each one investigated by a section of main process (classes bearchaeo/Measurement_PLM_Inclusions, _Matrix, and _Voids). The representation of how pottery thin sections are analyzed by means of optical microscope under polarized light consists of attribute values along some dimensions (e.g., relative abundance and sizes of inclusions) and terms from specialistic vocabularies (e.g., grain size distribution, valued as unimodal, bimodal, or heterogenous, or mineral/petrographic component, with subtypes such as quartz presence or alkali feldspars presence, valued as XXXX, i.e. >50%, XXX, i.e. 50–30%, XX, i.e. 30–10%, X, i.e. <10%, D, i.e. detectable).

Finally, connected to the data interpretation of the digital data curation schema is the modeling of data evaluation (class cidoc-crm/S6_Data Evaluation) that follows acquisitions/measurement and processing. In the instantiated model reported in Fig. 10, the results from Thermoluminescence, Archaeomagnetism, X-ray powder diffraction and Scanning Electron Microscopy (on the left) are combined to infer the firing temperature of a pottery shard (namely, the sample No. 7 from Tatetsuki area). In particular, the numerical value obtained from archaeomagnetism analyses can be confirmed by the observations of other parameters (i.e. moisture content at saturation, presence/absence of calcite, porosity and sintering degree of body paste), which are obtained from other scientific techniques, initially used to obtain other type of information. They also produce data that can be exploited to cross-check knowledge in an interdisciplinary environment as each contribution independently suggests specific temperature ranges. In the next section, we see how this information is annotated by the BeArchaeo archaeometric team in the database to reflect such a transdisciplinary approach.

Fig. 10.

Evaluation of data for the assignment of a dimension.

6.Preliminary evaluation of the model in the BeArchaeo project

The digital data curation of a few findings in the BeArchaeo project forms a preliminary evaluation of the BeArchaeo ontological model. As the conceptualization and modeling of the archaeological and the archaeometric knowledge proceeds, we have developed a web platform for the form filling of the scientists, based on the catalogue record model. So, we can report on some preliminary evaluations of the approach.

6.1.Deployment of BeArchaeo ontology for the Tobiotsuka Kofun excavation

Project Beyond Archaeology (BeArchaeo) consists of the archaeological excavation, archaeometric analyses, interpretation of the findings, and eventually dissemination of the results about the Tobiotsuka Kofun (Soja city in Okayama Prefecture), and other archaeological materials of the ancient Kibi and Izumo areas now stored in museums and laboratories, in Japan. The ontology described above underlies a semantic database for the encoding and storing of the digital data concerning the documentation of the archaeological excavation and the account of metadata that arise from the archaeometric tests and interpretations.2626 In particular, the project has drawn inspiration from the forms distributed by national authorities, which have informed the classes and properties of the catalogue record module of the BeArchaeo ontology. The vocabularies addressed above have been encoded as custom vocabularies into an installation of the semantics-based Content Management System Omeka-S.2727 As seen above, the catalogue record module is connected to the archaeological and the archaeometric knowledge, and the plan is to perform inferences and consistency checking of the interpretations in the future.

The forms have been deployed as “Resource Templates”, with the fast prototyping of user interfaces for both the back-end of the system, accessible by the archaeologists and the archaeometrists, and the front-end, where supervisors and stakeholders check the development of the archive and the related findings. Also, considering the multi-cultural and multi-lingual issues of the Be-Archaeo project, knowledge interoperability between Japanese and English researchers as well as data terminology have been addressed by providing also Japanese resource templates for the Archaeological Finding and Stratigraphic Unit records, respectively (currently, in the development site2828). Also, we have uploaded rich media materials (photos and 3D models acquired from photogrammetry and scanning), that are being used for interpretation and will be the basis for the final exhibition. Figure 11 reports two images, from the back end and the front end, respectively, of the production website.2929

Fig. 11.

Screenshot from the BeArchaeo resources website, concerning the archaeological finding no. 59, with the related fields and media. On the left, the back end; on the right, the front end. Elements in red are links to other elements of the documentation (e.g., Stratigraphic Unit 202) or to some external knowledge source (e.g., Getty AAT thesaurus).

During the development of the BeArchaeo project, we could observe the behavior of the archaeologists and the archaeometrists, respectively. Archaeology and archaeometry are at a different stage of development with what concerns the curation of the digital data. The archaeologists have found the model accurate, mostly because of the connection of the model to the forms that are already in use, being the latter a conceptualization effort made by national authorities; so, the alignment of the catalogue record module with the archaeological knowledge resulted to be effective. The categorization of the data inserted through the form fields and the possibilities offered by the web platform to introduce and motivate different annotations has led to discussions between the team members, with an impact on the reflectivity issues mentioned at the beginning. Again, by relying on a web platform, the several roles of the users, namely Authors, Reviewers, and Editors, have contributed to a fruitful awareness of the results of the project. The work with each archaeometric disciplinary team tackled the task of conceptualization within the small group at the beginning, focussing on the use of a specific investigation technique, and then extending it within larger disciplinary groups. The final broad discussion sessions have lead to the final procedures adopted within the whole multidisciplinary team. The modeling phase, which continuously enlarges its coverage, takes advantage of this transdisciplinary account of the data and the whole archaeometric team is gaining a great awareness of the similarities and differences of the procedures adopted within the disciplinary accounts, in a holistic perspective. The integration of the archaeometric and the archaeological knowledge, through a centralized database, has triggered an effort in the alignment between the interpretations provided by the different members of the team. In particular, the system has triggered discussions within the several disciplines of the archaeometric team and between the archaeological and the archaeometric teams, respectively.

7.Conclusion

We have presented a transdisciplinary ontology-based approach to the encoding of archaeological and archaeometric knowledge. In particular, we have setup a procedure for addressing the transdisciplinary endeavor and we developed a prototype ontology of the interconnected archaeological and the archaeometric domains, respectively. These issues are particularly relevant for the digital data curation of an archaeological investigation; we have also devised how the knowledge is linked to the form interfaces, for collecting the data as the excavation goes on, to be continued in the analysis labs, and eventually with the design of the exhibition. We have identified the major entities that are required for a reflexive methodology of archaeology, especially in its relationship with the archaeometric knowledge. The conceptual model is the outcome of several modeling sketches and subsequent discussions carried out by the members of the archaeological and the archaeometric teams, representing the several disciplines involved. The conceptualization has been developed in support of a digital data curation framework that serves the needs of an ongoing archaeological investigation.

The conceptual model and the ontology of the archaeometric knowledge serve the design and implementation of the interface forms for both archaeological and archaeometric filling, in order to enable researchers operating on the field and afterwards in the labs to load their results into the database. As far as we know, BeArchaeo is the first born-semantic project that assumes a joint archaeological/archaeometric perspective from the start. In fact, the multi-disciplinary, multi-cultural, and multi-lingual characters of Be-Archaeo raise a high demand of interoperability of knowledge and data. The alignment with CIDOC-CRM is pursued at the disciplinary level, by aligning the archaeological and the archaeometric descriptions through the CRMarchaeo and CRMsci models, where possible.

The realization of an overall approach, together with the adherence to well known standards and with an implemented workflow from the excavation design to the exhibition, can greatly contribute to the replication of the method across other projects. The BeArchaeo archaeological team is a proper representative of the “archaeological community”: the Japanese archaeologists are strictly linked to the Japanese Research Institute for the Dynamics of Civilization,3030 the Portuguese archaeologists are part of the Centro de Arqueologia de Universitade de Lisboa,3131 and the Italian archaeologists are set within the International research Institute for Archaeology and Ethnology.3232 Also, after BeArchaeo, the model is going to be adopted in further initiatives in Europe (e.g., check the networking session of the UNITA project on October 20213333).

In the next future, we continue the encoding of further archaeometric aspects and the strict connection with the archaeological interpretations, to implement some form of automatic reasoning on the data collection. As the project database will be growing in the collection of data, we are going to improve the interfaces for engaging a higher number of diverse researchers and promote the usage of the conceptual model in other archaeological/archaeometric projects. The Omeka-S frontend, which has been an immediate solution for monitoring the project initial database schema (given some previous experience with the tool), will be replaced by a customized interface, while continuing to serve as a backend to the database monitoring. We are also working on a novel repository (currently a Google drive folder) for the media supporting the archaeometric analyses and interpretations. In particular, we are currently in the phase of analyzing the requests about the possible uses of the data in the future, in order to devise the best repository solution.

Finally, we are going to evaluate the contribution of the centralized semantics-enhanced digital data curation in its impact onto the final exhibition.