An Ontology for Legacy Data on Ancient Ceramics of the Plain of Catania?

Abstract

Digital representation and organization of legacy data plays a crucial role in the di usion, use, and understanding of data stored in old publications, archives, and museums. An interesting case study comes from data of potteries discovered in ancient rural territories of Eastern Sicily, as the majority of legacy data for this research area exists in the form of old maps and paper catalogues: to make these datasets available at a global level, innovative digital technologies are needed. The Semantic Web o ers well established methodologies and tools to semantically model application domains and to integrate data, making them global entities available on the Web. In this contribution, we present OntoCeramic 2.0, an OWL 2 (Web Ontology Language 2) ontology storing archaeological data from the plain of Catania regarding ancient potteries, and whose taxonomy re nes and extends OntoCeramic 1.0, an ontology for the classi cation of ancient ceramics de ned in a previous work by some of the authors. OntoCeramic 2.0, constructed according to the standard CIDOC Conceptual Reference Model (CRM), represents and integrates new survey and legacy data on ancient pottery stored in the archives of Heritage Superintendence of Syracuse and Catania, in the Regional Technical O ce of Sicily, and in the State Archives of Palermo and Catania.

-

Archaeological studies carried out in Sicily for over a century report on forgotten cities, necropolises, monuments, artefact scatters, and other landscape features. Resulting data are still limited in quantity and variable in quality: this problem is particularly prevalent in the countryside. Because of that, legacy data nowadays available for the plain of Catania (Sicily) are of basilar importance for archaeologists. This holds, in particular, if such data are represented and organized in a digital way, globally accessible on-line, and easily veri able. ? We gratefully acknowledge support by \Universita degli Studi di Catania, Piano della ricerca 2016/2018 Linea di intervento 2".

Currently, a MySQL database containing legacy data on potteries is being developed. A rst version of it is described in [ 5 ] and is part of the Ru.N.S. project (Rural Networks in Sicily). Analysis of potteries is fundamental for scholars since it helps in providing a clear image of rural population trends in ancient times, and in reconstructing the organization of the agrarian territory in the Hellenistic and Roman ages [ 18 ]. The study area considered in this contribution is located in the western portion of the plain of Catania, the area between the Simeto river to the north and the Margi river to the south. With an extension of 540 km2, the area forms a perfect case study due to the number of excavations and survey projects carried out by the Heritage Superintendence of Catania and the Chair of Ancient Topography (University of Catania) over the last few decades. An overview on excavations and survey projects in the area can be found in [ 4, 21 ].

Relational databases, however, even though well-assessed tools for organizing and querying information, do not support global data and exible integration mechanisms with other sources. Moreover, they su er from limited modelling and reasoning capabilities.

Semantic Web [ 3 ] o ers methodologies, languages, and tools for knowledge representation systems in which data are published, accessed, and integrated with information from other sources at a global level, thus allowing coherence and dissemination of knowledge. Moreover, the presence of dedicated automated reasoning systems permits to verify the consistency of the model, to query the dataset, and to infer implicit knowledge present both in the taxonomy and in the data. The de nition of a speci c domain is widely called an ontology.

Recently, capabilities of ontologies have been understood and appreciated by archaeologists [ 16, 17 ]. Some projects have been started regarding speci c kinds of archaeological nds such as ancient manuscripts [ 12 ] and epigraphs [ 13 ]. In [ 9 ] we presented OntoCeramic 1.0,3 an OWL (Web Ontology Language) ontology for cataloguing and classifying ceramics, originated by a synergic e ort between computer scientists an archaeologists to address the problem of e ciently automatize the task of correctly cataloguing ceramics and to make such knowledge easily accessible and usable by researchers of the eld.

In this contribution, we present OntoCeramic 2.0,4 an ontology storing data of ancient ceramics discovered on the western side of the plain of Catania in Sicily and collected by the Ru.N.S. project. OntoCeramic 2.0 models principal features of potteries such as ceramic class, shape, type, dough, and chronological periods of production of the nds. The ontology is completely mapped in the CIDOC Conceptual Reference Model (CRM) [ 11 ], the international standard of exchange of cultural heritage knowledge. This makes it exible and fully embeddable with ontologies conceived for di erent application domains.

3 https://github.com/dfsantamaria/OntoCeramic-1.0/blob/master/

OntoCeramic1.owl

4 https://github.com/dfsantamaria/OntoCeramic-2.0/blob/master/

OntoCeramic2.owl

Preliminaries 2.1

Semantic Web and Web Ontologies

Semantic Web is a vision of the World Wide Web in which machine-readable data enable software agents to access, extract, integrate, manipulate, and query information on behalf of human agents, and thus to gain a deeper knowledge of the domain. To achieve such goals at global level, information must carry an explicit meaning and must be modelled by appropriate languages endowed with formally de ned semantics supporting automated reasoning procedures. For this purpose, the Word Wide Web Consortium (W3C) identi es the Web Ontology Language (OWL), a family of knowledge representation languages relying on Description Logics (DLs) [ 2 ], as the standard for representing ontologies. We recall that an ontology [ 15, 19 ] is a formal description of the domain of interest carried out by combining three basic syntactic categories: entities, expressions, and axioms, which constitute the logical part of ontologies, namely what ontologies can express and the type of inferences that can be drawn. Entities represent primitive terms of an ontology and are identi ed in a unique way. They are individuals (actors), object- and data-properties (actions), and classes (sets of actors with common features).

OWL5, currently in version 2.1, is based on the idea of triples, which are ways to connect two individuals or and individual and a data-value. In order to provide a formal description of the domain, OWL 2 triples can be organized in two main categories: axioms and expressions. Axioms are constructed by applying OWL 2 primitives to entities, thus forming complex descriptions, whereas expressions describe what is true in the domain. For example, one can combine an axiom for equivalent classes with an expression of class union such as to de ne the class Late Antiquity as the period corresponding to the union of the periods represented by the classes Ostrogothic Age and Late Imperial Age.

OWL 2 admits three main types of expressions: object-property expressions, data-property expressions, and class expressions. Object-property expressions represent binary relationships among individuals, whereas data-property expressions represent binary relationships among individuals and data type values. Class expressions represent sets of individuals sharing common characteristics. Such individuals are said to be instances of the respective class expressions. Class expressions are constructed recursively by using classes, properties, and class expressions, and by applying restrictions on property expressions. For a detailed explanation of axioms and expressions introduced in OWL 2, the reader is referred to [ 1, 14 ].

5 https://www.w3.org/TR/owl2-overview/.

2.2

OntoCeramic 1.0 OntoCeramic 1.0 is an OWL 2 ontology presented in [ 9 ] for cataloguing and classifying ancient potteries, designed with the purpose of e ciently addressing signi cant problems concerning knowledge management about potteries such as the classi cation by shape, type, and class, and the analysis of nds by their components and discovery places. The ontology has been designed on ICCD6 (Istituto Centrale per il Catalogo e la Documentazione) data sheets taking into account relevant papers in the eld [ 10, 16 ]. It contains more than 90 classes, 33 object-properties, 20 data-properties, and 13 SWRL rules permitting several reasoning tasks on the knowledge base in a short time. The expressive power of the language underlying Ontoceramic 1.0 has been studied in [ 6,22 ]. In particular, in [ 22 ], an OWL 2 pro le representing OntoCeramic 1.0 has been constructed from a decidable fragment of set theory, and it has been proved that the computational complexity of the consistency problem for its knowledge bases is NP-complete. 2.3

CIDOC CRM

The CIDOC Conceptual Reference Model (CRM) is the international standard for the controlled exchange of cultural heritage information since 2006. It provides a general speci cation which can be adopted in any cultural heritage context to construct a Semantic Web-based information system, to serve as a guide for good practices of conceptual modelling, and to improve information sharing. There are several institutions that successfully implement CIDOC such as galleries, libraries, museums, archives, as well as any other cultural environment based on cultural heritage data that publishes and shares its information in the Semantic Web formats. The CIDOC core covers several general aspects of cultural information, such as material and immaterial entities, events, space, and time. Such general concepts can be specialized, contextualized, and integrated in order to address practical aspects of cultural heritage issues. It models several notions, such as participation, appellation, parthood and structure, material and immaterial stu s, location, assessment and identi cation, motivation, and so on. 3

The Ontology OntoCeramic 2.0 In this section, we illustrate the ontology OntoCeramic 2.0 which re nes and extends OntoCeramic 1.0 to model and reason on the survey and legacy data of the plain of Catania collected in the ambit of the Ru.N.S. project. Speci cally, in OntoCeramic 2.0 we re ned the de nition and enriched the classi cation of (a) fabric and pottery types (also called ceramic class and type, respectively), which help in determining the production site of archaeological nds, (b) the shape of the nd, which helps in determining the pottery type, and (c) the pottery sizes, usually determined by measuring the external diameter of the rim in millimeters.

6 http://www.iccd.beniculturali.it

In OntoCeramic 2.0 we also consider further features: (d) the pottery chronological context with respect to the Sicilian historical periods, and (e) populating the ontology with data collected from the Ru.N.S. Finds catalog dataset. The latter is an Excel le consisting of 4384 rows, each containing basilar information on the archaeological nds discovered during the recognition phase of the considered territories (western edges of the plain of Catania), namely identi cation code, class, shape, type, conservation state, dimensions, and free-standing text descriptions. The task of populating the ontology has been performed exploiting the Protege plug-in Cell e,7 which allows one to parse Excel les and to map Excel entries to OWL triples.

OntoCeramic 2.0 has been de ned according to the standard CIDOC CRM and it uses the LinkedGeoData8 ontology for describing locations and for identifying the discovery place of nds. It consists of more than 220 classes, 40 object-properties, 20 data-properties, and 9000 individuals, excluding entities imported by CIDOC CRM and LinkedGeoData.

The rest of this section is devoted to the description of the ontology OntoCeramic 2.0. We rst list the main classes of the ontology and their characteristics, and then describe the general structure of the taxonomy.

- Archaeological Find : this class collects individuals representing archaeological nds. It is de ned as subclass of the CIDOC class E22 Man-Made Object. - Ceramic Class: is the root of a class hierarchy describing ceramic classes to which a nd may belong to. - Facies: is a subclass of Ceramic Class of particular interest because it models, together with its subclasses, all the ceramic classes in the Sicilian context. - Shape: is de ned as a subclass of the CIDOC class E26 Physical Feature and describes the shape of nds. One of its notable subclasses is the class Undistinguished Shape, introduced to model ambiguous shapes. - ArchaeologicalType: describes the type of nds by specifying the shape and, when available, the class. Among its subclasses, a relevant one is the class Undistinguished Type modelling ambiguous types. - Decoration: is de ned as a subclass of the CIDOC class E26 Physical-Features and describes the decoration of archaeological nds. - Description: contains a free-text description concerning nds and is a subclass of the CIDOC class E73 Information Object. - Dimension: de nes the size of nds and is a subclass of the CIDOC class

E54 Dimension. - Dough: describes the elements used to compose the dough of nds and is a subclass of the CIDOC class E26 Physical-Features. - Conservation State: reports on the physical conditions of nds at their discovery time and is de ned as a subclass of the CIDOC class E14 Condition State. - Sicilian Period : is the root of a class hierarchy that models Sicilian historical periods and is de ned as a subclass of the CIDOC class E4 Period.

7 https://github.com/protegeproject/cellfie-plugin

8 http://linkedgeodata.org/

The classes Undistinguished Shape and Undistinguished Type have been introduced to represent nds that have not clear shapes and types, respectively. Fig. 1 partially illustrates the class hierarchy with root Undistinguished Shape and shows how to model the shape of an archaeological nd when one is uncertain on the fact that it has the shape of a mortar or of a basin. The Mortar-Basin class contains individuals that may belong either to the class Mortar or to the class Basin. The object-property identi edAs relates such \hybrid" individuals with instances of the class Mortar or with instances of the class Basin, which clearly de ne di erent shape classes.

The core of OntoCeramic 2.0 is depicted in Fig. 2. In bold we introduce classes (resp., properties) speci cally de ned for OntoCeramic 2.0, immediately below them we report the corresponding superclasses (resp., superproperties) from CIDOC CRM.

Instances of the class Archaeological Finds are linked to their types, shapes, and classes, by means of the object-properties hasArchaeologicalType (subproperty of P41i was classi ed by ), hasShape (subproperty of P56 bears features of ), and hasClass, respectively.

The class ArchaeologicalType is associated to the classes Ceramic Class and Shape by means of the object-properties speci edByClass and speci edByShape (subproperty of P41 classi ed ), respectively.

We have separated the notion of shape from the notion of functionality of nds, i.e., the usage nds have been originally intended for. For example, an archaeological nd may have the shape of a basin and the functionality of an holy water font. The notion of functionality is de ned by exploiting the class Functionality.

Archaeological nds are related with their functionality by means of the object-property hasFunctionality, with their conservation state by means of the object-property hasConservationState (subproperty of P34i was assessed by ), and with related free-text descriptions by means of the object-property hasDescription (subproperty of the CIDOC relation P128 carries).

Moreover, nds are related with their dimensions, modelled as instances of the class Dimension, by means of the object-property has dimension (subproperty of P43 has dimension). Since dimensions of nds can be irregular and measurement errors may occur, we introduce two subclasses of Dimension, the classes Max Dimension and Min Dimension. The object-property has value (subproperty of the CIDOC relation P90 has value) relates each dimension with its value, represented by a double. Finally, nds are related with fragments composing them by means of the object-property formed by (subproperty of the CIDOC property P46 is composed of ).

As mentioned above, OntoCeramic 2.0 is endowed with an accurate chronological modelling of the historical periods concerning the production activity of archaeological nds in the Sicilian territories. Principal historical periods are represented by means of a hierarchy of classes having as root the class Sicilian Period and whose instances are related with the individual Sicily, (instance of the class Localisation) by means of the CIDOC property P78 took place at. The class Localisation is de ned as a subclass of the LinkedGeoData class Place and of the CIDOC class E54 Place. The data-properties start date and end date link each period with its start and end dates, respectively. Each period is described by means of an OWL expression representing the time interval between its start and end dates. For instance, the period Sicilian Iron Age is de ned as the Sicilian period ranging from the year -900 to the year -476 (in absolute value) and contains as instances the sub-periods sicilian iron age 1, starting in -900 and ending in -734, and sicilian iron age 2, starting in -733 and ending in -476 (see Fig. 3). Such de nitions force DL reasoners to place individuals representing speci c sub-periods in the correct subclass of Sicilian Period. This is useful when one wishes to relate historical periods of di erent regions of the world. Historical periods, indeed, vary from a region to another, since socialcultural and environmental phenomena arise in di erent moments. For example, the Late Bronze Age in Malta, starting in -700 and ending in -500, occurs during the Iron Age in Sicily. As Fig. 3 shows, this fact is correctly deduced by the Pellet DL reasoner that places the individual malta late bronze age, modelling the Late Bronze Age in Malta, in the class Sicilian Iron Age.

Determining the ceramic class of nds not only helps in correctly dating them, but also in reconstructing the chronological information of the archaeological context [ 20 ]. For instance, the shapes of the rim and of the body of Greek black burnished wares are good chronological markers of the production activity. Hence, the task of reasoning on the relationships among ceramic classes, archaeological nds, and historical periods turns out to be crucial to recognize the production activity and to collocate nds in the correct chronological context. The production activity is modelled in OntoCeramic 2.0 by means of the class ProductionActivity (subclass of the CIDOC class E12 Production). Ceramic classes and facies are related with production activities by means of the object-property speci esProductionActivity. Finally, nds are related with instances of the class ProductionActivity by means of the object-property produced (subproperty of the CIDOC property P108 has produced ). 3.1

Conclusions

We presented OntoCeramic 2.0, an OWL 2 ontology storing new survey and legacy data on ancient pottery stored in the archives of Heritage Superintendence of Syracuse and Catania, in the Regional Technical O ce of Sicily, and in the State Archives of Palermo and Catania, and collected within the Ru.N.S. project.

We integrated OntoCeramic 2.0 in the standard CIDOC CRM and de ned important features of ceramics such as class, shape, type, dough, and chronological periods of archaeological nds. We plan to extend OntoCeramic 2.0 in such a way as to support stratigraphic excavations, production factories, topographical information, and bibliographic references management. In addition, we consider to integrate OntoCeramic 2.0 with data from the Eastern side of the Sicily and with ontologies for other types archaeological nds.

Finally, we intend to de ne a set-theoretic representation of OntoCeramic 2.0 in the avour of [ 6 ]. However, since OntoCeramic 2.0 contains existential restrictions, we also need to modify the underlying set-theoretic fragment in such a way as to allow a restricted form of the composition operator. The related reasoning procedure will then be adapted to the new set-theoretic fragments exploiting the techniques introduced in [ 7, 8 ] in the area of relational dual tableaux.

References

1. D. Allemang and Hendler J. Semantic Web for the Working Ontologist: E ective Modeling in RDFS and OWL . Elsevier, 2011 .

2. F. Baader , I. Horrocks, C. Lutz , and U. Sattler . An Introduction to Description Logic. Cambridge University Press, 2017 .

3. T. Berners-Lee , J. Hendler , and O. Lassila . The Semantic Web. Scienti c American , 284 ( 5 ): 34 { 43 , 2001 .

4. E. Bonacini . Il Territorio Calatino nella Sicilia Imperiale e Tardoromana. British Archaeological Reports British Series , Oxford, 2007 .

5. R. Brancato . Pro lo Topogra co dei Paesaggi Rurali della Piana di Catania . In forthcoming, Ph.D. Thesis , University of Catania, 2019 .

6. D. Cantone , C. Longo , M. Nicolosi-Asmundo , and D. F. Santamaria . Web Ontology Representation and Reasoning via Fragments of Set Theory . In Cate, B. and Mileo , A . (eds) Web Reasoning and Rule Systems . LNCS, vol. 9209 . Springer, 2015 .

7. D. Cantone , M. Nicolosi-Asmundo , and E. Orlowska . Dual tableau-based decision procedures for some relational logics . In Proceedings of the 25th Italian Conference on Computational Logic , CEUR-WS Vol. 598 , Rende , Italy, July 7-9 , 2010 , 2010 .

8. D. Cantone , M. Nicolosi-Asmundo , and E. Orlowska . Dual tableau-based decision procedures for relational logics with restricted composition operator . Journal of Applied Non-Classical Logics , 21 ( 2 ): 177 { 200 , 2011 .

9. D. Cantone , M. Nicolosi-Asmundo , D. F. Santamaria , and F. Trapani . Ontoceramic: an OWL Ontology for Ceramics Classi cation . In Proceedings of CILC 2015 , CEUR-WS , vol. 1459 , pp. 122 { 127 , Genova , July 1-3 , 2015 .

10. L. Corti. I Beni Culturali E La Loro Catalogazione . Mondadori, Milano, 2003 .

11. M. Doerr . The CIDOC CRM An Ontological Approach to Semantic Interoperability of Metadata . AI Magazine , Vol. 24 , N. 3 , 75 - 92 , 2003 .

12. A. Felicetti and F. Murano . Scripta Manent: a CIDOC CRM Semiotic Reading of Ancient Texts . International Journal on Digital Libraries , 18 ( 4 ): 263 { 270 , 2017 .

13. A. Felicetti , F. Murano , P. Ronzino , and F. Niccolucci. CIDOC CRM and Epigraphy: a Hermeneutic Challenge . In Proc. of the Workshop on Extending, Mapping and Focusing the CRM, Poznan, Poland, September 17 , pages 55 { 68 , 2015 .

14. A. Grigoris , P. Groth , F. van Harmelen , and R. Hoekstra . A Semantic Web Primer , Third Edition . The MIT Press, 2012 .

15. T. Hofweber . Logic and Ontology. Edward N. Zalta (ed.), The Stanford Encyclopaedia of Philosophy (Summer 2018 Edition) , 2018 .

16. A. La Fragola. L' Atlante delle Forme Ceramiche dell'Enciclopedia dell'Arte Antica: Ipotesi di Progetto per una Gestione e Fruizione Elettronica dei Dati in Ambiente XML. Bollettino d'Informazioni, Centro di Ricerche per i Beni Culturali , XII, n. 1, Pisa , pages 113 { 119 , 2002 .

17. R. Letricot and A.V. Szabados . L'ontologie CIDOC CRM appliquee aux objets du patrimoine antique . In Archeologia e Calcolatori, sup. 5 , pages 257 { 272 , 2014 .

18. M. Mazza. L'Economia Siciliana tra Impero e Tardo-Impero. in Contributi per una Storia economica della Sicilia, Regione Siciliana , pages 15 { 62 , 1987 .

19. D. Oberle , N. Guarino , and S. Staab . What is an ontology? Handbook on Ontologies . Springer, 2009 .

20. C. Orton , P. Tyers , and A. Vince . Pottery in Archaeology. Cambridge, 1993 .

21. F. Privitera . Dall'Alcantara agli Iblei: la Ricerca Archeologica in Provincia di Catania . Spigo, U . (eds), Regione Siciliana , 2005 .

22. D. F. Santamaria . A Set-Theoretical Representation for OWL 2 Pro les . LAP Lambert Academic Publishing, ISBN 978-3-659-68797-6 , 2015 .