Sword icona di Icons8

LODlybaeum

A Punic Wars Linked Open Data Project
Aut viam inveniam aut faciam - Hannibal

Idea

Illustration of Roman Senate

Circa 264-146 BCE, the ancient Mediterranean region would witness three major conflicts fought between Rome and Carthage. Significant for their lifelong impact on the region, the victory of Rome would shape its rise as the dominant power in the ancient world.

The first Punic War (264-241 BCE) was caused by a complex competition over trade routes and territories in the Western Mediterranean, especially in Sicily and North Africa. It was a naval conflict centred around Sicily with Rome winning and gaining control of the island.

The second Punic War is largely the most famous out of the three wars, where the Carthaginian general Hannibal led his elephant fortified army on an overland reach from Spain across the Alps and eventually into Italy. On his march, he inflicted many taxing defeats on the Roman Empire, despite this the war was lost and Carthage had to cede territories in both Spain and North Africa and pay heavy indemnity to Rome.




The third and final Punic War (149-146 BCE) was sparked by renewed tensions between Rome and Carthage. Viewing Carthage’s presence as a continuous threat to the empire, Rome laid siege to the city and eventually captured it only to destroy it in 146 BCE. Carthage’s downfall allowed Rome a monopoly of control in the Mediterranean region and solidified its dominance.

These wars challenged the notions of warfare in ancient times and pushed two superpowers to concede, adapt and re-adjust. This Linked Open Data project aims to explore the significance of these events in the region by exploring the gains, losses and achievements of the two respective forces.

Illustration of a Carthaginian Senate

Items

The selected items are objects of relevance to the Punic Wars, inclusive of notable people, places and events; maintaining a historical perspective but also a modern retelling of what transpired.

Conceptual map

The items were selected during the creation of the conceptual map, which guided us through several iterations to narrow down to the ten items that best described the event at hand. We chose to group them based on their respective creation periods, with the upper section of the model representing modern items and the lower section representing ancient ones. The scenario of the project, the Punic Wars, naturally occupies the central position within the model.

conceptual map

E/R model

Following the selection of the ten items we constructed an Entity-Relationship model to depict the relevant relationships between them. Once again they have been arranged based on the respective creation periods; the relevant relationships outside the primary event have been provided.

E/R model

Metadata analysis

After having appropriately mapped the items, we identified the metadata standards used by the institutions to describe them.

When the institution did not mention a specific standard they used, as in the case of the podcast and the article, and the cuirass we selected one that could describe the item accurately.

Metadata standards marked with the asterisks (*) are generic. This occurs in case of incomplete or non-existing metadata.

# Item name Object type Provider Metadata standard
1 The Histories Book University of Chicago Schema.org*
2 Punic Nightmares Podcast iTunes Schema.org*
3 Barbarians Rising: Resistance Episode History Channel IMDb
4 Ksour Essef cuirass Cuirass Bardo National Museum Wikidata
5 The Death of Archimedes Painting National Trust Collections Wikidata
6 Capuan Bust of Hannibal Marble Sculpture National Archeological Museum of Naples DCAT-AP
7 The Battle of Zama Painting Art Institute Chicago Wikidata
8 La muerte de Ásdrubal Song MusicBrainz MMD
9 Hannibal Coin Coin British Museum DCAT-AP
10 To Be Taken with a Pinch of Salt Article JSTOR Schema.org*

Metadata alignment

To account for interoperability between standards we have made explicit the correspondences between their properties, which have been grouped based on the question they answer : who, when, where or what.

Starting from the predicates already chosen in the description of the items, we collected all the analogue predicates from different standards, to finally match corresponding predicates together.

WHO

PROPERTY DCTerms MMD DCAT-AP Schema.org Wikidata IMDb
Creator creator <artist><name> creator ScholarlyArticle.author
ScholarlyArticle.creator
P50 writers
Publisher publisher <label> publisher ScholarlyArticle.publisher P123 N/A
Contributor contributor <name><namePart> N/A ScholarlyArticle.contributor P767 N/A

Theoretical model

Through an abstraction of our Entity-Relationship model, we derived a theoretical framework that describes the range of entities and relationships across the ten items.

A note on the choices made in the realisation of the model and all subsequent materials: although the podcast has 3 episodes, only one of them has been explicitly represented in the model in order to avoid overcrowding. The two remaining episodes are linked in exactly the same manner. Each instance has been replaced with its own class.

Theoretical model

Enhanced E/R model

New entities and relationships have been added to the original E/R model on the basis of the theoretical model, providing a more intricate and comprehensive view of the items and the primary event.

Enhanced E/R model

Conceptual model

Tracing from our work on the enhanced E/R model, the conceptual model utilises the triple statements as written by their respective standards. Elements that enhance the relationships between items also have a syntax consolidated from the Github URIs that we have created.

Conceptual model Conceptual model

Example XML/TEI Document

For the example XML/TEI document that we had to prepare as part of the project, we selected the last item, the article. The preparation of the XML/TEI document has started from preparing the teiHeader.

The first principal component of teiHeader is fileDesc. The elements of fileDesc prepared for our document were titleStmt, to provide a title and the author of the article; publicationStmt, to provide information about the publisher of the article; seriesStmt, to provide the information about the journal, in which the article was published in and where exactly it can be found; and sourceDesc, to provide bibliographic reference to the original occurence of the article. There are three elements omitted: editionStmt, as there is no information available regarding the edition of the journal; extent, as the physical dimensions of the journal are not available, and the page count is already included elsewhere; and notesStmt, as no additional information has been found.

Our version of encodingDesc contains only the information describing the project (projectDesc), as well as the editorial declaration (editorialDecl) in regards to correction and normalization. The model provides a lot more possibilities than was accomplished here, of course, as more information regarding e.g. geographical locations (geoDecl) could be included in this part.

The profileDesc of the text provides a description of non-bibliographic aspects of the text. As the subject of the document is a scientific article, only the information regarding the language used (langUsage) and the situational parameters of the text (textDesc) were used here. The latter is also the only element of the header where all possible content has been included.

xenoData, the fourth principal component of teiHeader has been omitted completely. It's primary purpose is to provide non-TEI metadata. While it is one of the goals of the project, constructing the RDF triples from the XML/TEI document and later adding them to it seemed counter-intuitive.

Finally, the revisionDesc provides basic temporal information about the preparations of the XML/TEI document.

The actual body of the text has been transcribed from the document with no changes to its content, except for the dehyphenation of some words that were originally put at the end of a line. There are two kinds of quotes, ones denoted in the original text by a change in formatting, the other denoted with double quotes. This is reflected in the TEI by use of <quote> and <q>, respectively. Bibliographic references are of three different kinds: the first corresponds directly to the aforementioned quotes and come with the bibliographic information (bibl), all wrapped up into <cit>s. They all have a corresponding footnote in the original text. The second kind uses bibls in place of the references, that have a corresponding footnote. The third kind are references put directly in text, and have been left in the transcription as is. Italics have been marked using emph.

The full document can be accessed here.

Transformation to HTML

For the transformation of the article into HTML, we created an .xslt file, as mentioned in the project specification.

There are two main parts of the transformation:

It's first part is selecting the metadata of the article out of the teiHeader. The title of the article is then put into the title of the webpage seen in the browser tab, and in the body of HTML the information regarding the title (again), the author and the publisher of the article is put as a header of the website. Other metadata, namely; publication place and date, journal in which the article was published, the volume of the journal, pages where the article can be found and the ISSN of the publication are put into a table.

The second part of the transformation regards the body of the article. Paragraph elements, which in the TEI file are denoted as <p>, are transformed into the exact same elements in the HTML. "Full" citations in the form of a cit block are transformed into <blockquote>s, which coincidentally corresponds to their original formatting. The two kinds of quotations are maintained, however the <q>'s are transformed into double quotes, and <quote>s are transformed into <q>'s. The <emph>s, which are conveying the original rendition of text in italics, are transformed into <i>'s, however they could be left unchanged and produce the same result. Finally, the bibliographic references that were originally in the footers of the pages, are transformed into square brackets to provide a visual break from the proper text of the article.

We have also prepared a CSS file to transform the visual look of the article

The result of the transformation can be accessed here. The full .xslt file can be seen below:

Transformation to RDF

After transforming the article into XML, we proceeded to convert its contents into RDF format. To begin, we established namespaces for key standards: TEI, Dublin Core, Schema.org, and Dublin Core Terms. Subsequently, we associated the prefix "SCHEMA" with the Schema.org namespace within the RDF graph to facilitate clear identification of elements.

Next, we retrieved data from the XML file. Two primary categories of information can be highlighted: contextual details related to the article itself, such as publisher, author and other relevant information; and the citations and the related bibliography. The bibliographic entries present within the article yet not directly linked to any citation have been dealt with separately and only associated with the article.

Utilising the extracted data, we constructed RDF triples that encapsulated the semantic relationships between entities. These triples were integrated into the RDF graph which was finally serialised.

The result of the transformation can be accessed here. The graphical representation can be accessed here. The full .py file can be seen below:

Data description

Book: Polybius - Histories

Subject Predicate Object
Book rdf:type dcterms:BibliographicResource
Book dc:creator person/Polybius
Book dc:title "Histories"
Book dc:language "grc"
Book dc:PhysicalMedium "papyrus"
Book dc:identifier "https://lccn.loc.gov/05000123"
Book dc:publisher "Trustees of Tufts University"
Book dc:created "1893"
Book owl:sameAs http://data.perseus.org/texts/urn:cts:greekLit:tlg0543.tlg001

RDF production

URI definition

In order to link the items of the domain of the project together, we decided to create unique names for all entities in form of https URIs.

The following paths were created to distinguish the entities: /event for temporal events (wars and sieges), /person for people and item for the 10 items of the project.

  • https://purl.archive.org/domain/lodlybaeum/event/PunicWars
  • https://purl.archive.org/domain/lodlybaeum/event/PunicWars
  • https://purl.archive.org/domain/lodlybaeum/event/SiegeOfSyracuse
  • https://purl.archive.org/domain/lodlybaeum/event/BattleOfZama

  • https://purl.archive.org/domain/lodlybaeum/person/hannibal
  • https://purl.archive.org/domain/lodlybaeum/person/Hasdrubal
  • https://purl.archive.org/domain/lodlybaeum/person/Polybius
  • https://purl.archive.org/domain/lodlybaeum/person/RonaldRidley
  • https://purl.archive.org/domain/lodlybaeum/person/Archimedes

  • https://purl.archive.org/domain/lodlybaeum/item/book
  • https://purl.archive.org/domain/lodlybaeum/item/podcast
  • https://purl.archive.org/domain/lodlybaeum/item/episode
  • https://purl.archive.org/domain/lodlybaeum/item/cuirass
  • https://purl.archive.org/domain/lodlybaeum/item/bust
  • https://purl.archive.org/domain/lodlybaeum/item/painting
  • https://purl.archive.org/domain/lodlybaeum/item/painting_2
  • https://purl.archive.org/domain/lodlybaeum/item/song
  • https://purl.archive.org/domain/lodlybaeum/item/coin
  • https://purl.archive.org/domain/lodlybaeum/item/article

TTL serialisation



RDF Visualisation

Knowledge graph


Knowledge graph

Team

Hubert
Iheb
Giorgia