Comprehension of past events and its reconstruction is one of the tasks performed by historians. With the introduction of computeraided methods the way in which historians perform their work has been transformed. One of these inclusions is the Semantic Web which can act as an alternative for publication, conciliation, standardisation and integration. Asturian notaries public contracts are a valuable material to understand the society of this epoch, specially in the Middle Ages where a renovation process in the institution was taking place. Therefore, in this work we explore the transformation of TEI-based XML transcriptions of notarial contracts to RDF by means of an heterogeneous data mapping tool which can improve the mechanism to publish Linked Data from existing transcriptions.3
Elucidating past events and bringing them to our days is one of the tasks
performed by historians which search evidences to reconstruct historical discourses.
As in many elds, the introduction of computer aided methodologies has opened
a new dimension in historical research in which has been coined as Digital
Humanities. One eld that has had a good reception is the Semantic Web whose
technologies have been envisaged as a mean of publication, conciliation,
standardisation and integration in the Humanities eld [
A particular main problem, in Digital Humanities, is the generation of Linked
Data from historical material. Although many procedures were proposed, most
of them involve creating ad-hoc solutions that could only handle with a certain
model, and modi cations would need much e ort or a software expert. Another
problem is how to deal with heterogeneous models without creating one solution
per schema. In the Semantic Web community new tools that try to deal with data
transformation and, also with heterogeneity|of formats and data models|have
appeared [
Among other historical resources, notarial contracts are particularly valuable
for the study of western mediterranean societies in the Middle Ages [
In this paper we describe our work on transforming transcriptions of medieval Asturian notarial contracts encoded with Text Encoding Inititive (TEI), how this can be achieved with heterogeneous data mapping tools and what are the challenges that this methodology poses. 2
In the last years several works have explored the idea of transforming
XMLbased historical artefacts to Linked Data. This is the case of [
Regarding the notarial deeds, the most similar work to the one presented in
this paper is presented in [
Although these works explore di erent ways of transforming data to RDF, none of them tackle the use of heterogeneous data mapping tools, which are capable to integrate|in the same script|data in various forms and formats. This proposal could lead to a faster transformation process due to the centralisation in one tool, the higher exibility against model changes and addition of other sources of information|with heterogeneous formats like: CSV, HTML, JSON, etc.|and the improvement on learning time from the domain experts. 3
In the XIII century, during the reign of king Alfonso X, a renewed doctrine in uenced the elaboration of a legal frame tted to the times and the particularities of the Castilian Crown.
This frame included a new legal corpus and the transformation of judicial and documentary practices. In this context, the traditional scribes|most of them coming from clerical institutions|were replaced by public notaries.
For Asturias, an ancient kingdom located in the north of Castile, the new policy established by king Alfonso X meant several changes, as it was the king's intention to modernise not only the administration of his realm, but to transform rural areas into a more dynamic urban ones.
Public notaries assumed their predecessors' role with a whole new meaning: rst, written culture no longer belonged exclusively to the Church; second, they o ered their services to everyone, no matter their economical solvency or social background; third, their profession was de ned by the law; thus, they recorded every single legal action and contract in the daily life of the Asturian society.
As no notarial registers from this early period remain today, we are working with documents issued by these notaries. Most of them were preserved by Asturian monasteries and cathedral, as they frequently used notaries' services to record their economical activities. Nowadays, a great amount of these documents are still guarded by an ecclesiastical institution|the monastery of San Pelayo of Oviedo5 is one of the richest private archives in the region|while many are also preserved at the Archivo Historico Nacional6|the biggest public-state archive in Spain. 4
Asturian notarial contracts from XIII and XIV centuries are held by several
private and public-state archives which, in some cases, can hinder their access.
Furthermore, in many cases digitised versions are not available. But even with
digitised versions, it is still to be proven that accuracy of promising
state-of-theart Optical Character Recognition (OCR) techniques [
As a rst step manuscripts are transcribed to TEI-XML using vocabulary
features plus some additions which cover diplomatic elements [
This version of manuscripts transcription can be queried and published but it has the problem of entities reconciliation and integration with other datasets. As a way to solve these lacks, the translation of these TEI-XML transcriptions
6 http://www.culturaydeporte.gob.es/cultura/areas/archivos/mc/archivos/ahn/portada.html 7 An example manuscript transcribed to XML can be seen on:
https://github.com/albarranelena/AsturianNotaries/blob/master/AAA 7.xml to Linked Data is explored. In this work we have decided to use ShExML as it o ers a simple syntax and, as being developed by two of the authors, it can be tuned if it is necessary. 5
The transformation process begins with the creation of the transformation script in ShExML syntax8.
To create the data model we have taken the schema.org vocabulary to de ne the general attributes. This vocabulary, in its pending branch9, o ers new types that are suitable for generics attributes of works like the one presented in this paper. Therefore, the archive component type is used to model the content and meta-data of each TEI-XML transcription.
Some of these attributes require to have another type in the object part. For these cases a shape link is made which is a mechanism to de ne a new shape with a new form that will be linked to the upper one. This is the case of the schema:locationCreated which is a schema:Place and has a name and a link to a Linked Data Cloud10 entity. Here, we have linked the schema:sameAs attribute with their Wikidata counterpart entity. This process is made using the ShExML matchers feature11 which allows to replace a string for another string of our choice. For instance, the town of Aviles can be linked with its Wikidata12 entry. Therefore, linking shapes we are able to create links between generated triples and model schema.org types inside ShExML. With the iterator nesting we are able to cover the tree structure and, also, multiple children from one parent which must be considered as a one triple generation per child.
Once this script is generated we can use the ShExML engine13 to convert an arbitrary number of les following this encoding model to their RDF counterpart. To check the conversion presented in this paper we also o er an online demo14 where we can upload the generated script and select the "Convert to RDF" option to generate the RDF output15. 6
Although this conversion can cover a lot of what is described in the TEI-XML transcription there are some limitations|which are also in line with some limitations encountered in TEI vocabulary and related formal ontologies derivatives 8 Script available on:
http://herminiogg.github.io/whiseIII-paper-2020/notariesShort.shexml
9 http://pending.schema.org
10 https://lod-cloud.net/
11 http://shexml.herminiogarcia.com/spec/#matcher
12 https://www.wikidata.org/wiki/Q14649
13 https://github.com/herminiogg/shexml
14 http://shexml.herminiogarcia.com/editor
15 Full output available on:
http://herminiogg.github.io/whiseIII-paper-2020/notariesShort.ttl
[
This problem increases when a diplomatic study is raised. In this case, modelling aspects such as the legal action described in the contract, the tradition of the act or the number and role of the participants cannot be achieved with the current vocabulary. Although this limitation do not restrict the conversion to Linked Data16 and, it can also be queried through SPARQL queries, it is true that it can limit future inferences and, moreover, it could limit integration with other graphs which is the nal goal of Linked Data.
Other ontologies like FRBR [
Another problem is how to identify and disambiguate persons' names which will require a mechanism to identify and disambiguate them from other people with the same name and surname. It is also problematic that these people are not registered in any other repository as may be, for example, the Kings of Spain (e.g.: Wikidata, DBpedia, etc.). This would involve the addition of an entity disambiguation mechanism in ShExML plus the creation of speci c algorithms for this case. This kind of knowledge extraction from the text would imply in a simpler and faster process for the transcriber that can focus more on the transcription process and less in the identi cation and categorisation of entities. 7
In this work we have explored the possibility to apply heterogeneous data mapping tools to a TEI-based XML transcription of notarial contracts in order to convert them to RDF. This transformation was carried out using ShExML, which aims to o er a simple syntax to de ne these kinds of transformations, and using the schema.org vocabulary to assure the integration of these corpora with other existing or future resources. The process has shown that schema.org and other existing vocabularies are not able to synthesise what is needed for diplomatic 16 Full ShExML script with diplomatic features: http://herminiogg.github.io/whiseIIIpaper-2020/notariesFull.shexml
Full RDF result: http://herminiogg.github.io/whiseIII-paper-2020/notariesFull.ttl 17 https://github.com/nie-ine/Ontologies 18 https://leonvanwissen.nl/vocab/roar/docs/ studies. Therefore, we envisage the creation of a diplomatic ontology as an approach to cover this topic. Moreover, we highlight the need for an identi cation and disambiguation mechanism for person entities to favour further analyses. Funding This work has been partially funded by the Principality of Asturias through the Severo Ochoa call (grants BP17-29 and BP16-51) and by the Ministry of Economy, Industry and Competitiveness under the calls of "Programa Estatal de I+D+i Orientada a los Retos de la Sociedad" (project TIN2017-88877R) and "Proyectos de I+D de Generacion de Conocimiento" (project PGC2018093495-B-100).