The Digital Diplomatics 2022 conference took place from the 28th to 30th of September 2023. It was the fourth instance of an international conference series (Munich 2007, Naples 2011, Paris 2013)[1], which originally started as a workshop in Munich in 2004. In the long gap between the last conference and this one, research has advanced significantly, although some problems remain unsolved and ongoing concerns.
Several of the 2022 speakers had already been present at previous instances of the conference series, allowing us to follow the development of their respective projects and datasets over the years. In a joint talk of BALÁSZ CSIBA and JURAJ ŠEDIVÝ, the former gave an introduction to the history of institutionalised printed editing of diplomatic material in 19th and 20th century Slovakia and his own editing work. Among others, he mentioned the Documenta Posoniensia collection of documents regarding the city of Bratislava, parts of which are already accessible digitally on Monasterium[2] and Crossborder Archives SK-AT (CRAC)[3], with minimal metadata. More data, including charter abstracts as well as some transcriptions and translations, to 3,411 units of said collection can be found on the general cultural heritage portal pamMap[4], which had been discussed at the 2007 conference and combines geographical, thematic, and prosopographical cross-links between different archival material.
The scope of the project Diplomatarium Serbicum Digitale (DSD), proposed at the 2011 conference, launched in 2015 and presented by ŽARKO VUJOŠEVIĆ (Belgrade), is to edit the corpus of deeds of Serbian issuers distributed across various European archives. During the first phase of the project until 2018 at the Institute for Balkan Studies, a database of about 330 diplomas and more than 600 legal documents was put online, the oldest one dating back to 1186. After the shut-down of the portal in 2020, it got picked up by the Audiovisual Archives and Center for Digitization of the Serbian Academy of Sciences and Arts in Belgrade, which now provides new funding for the project to be re-published in collaboration with the Centre for Information Modelling in Graz. The data for this digital edition will be annotated following the TEI guidelines, enhanced by necessary additions from the Vocabulaire Internationale de la Diplomatique (VID).
ZSOLT HUNYADI’s (Szeged) talk covered the history and current status of a database for charter calendars of the 86-year period in which Hungary was ruled by kings of Angevin lineage[5]. In the pilot project of 2011, the implementation of data from around 19,000 charters had been achieved, as reported at the 2013 conference in Paris. A continuation was recently decided in collaboration with the DigiDeeds project in Toronto, associated with the goal of growing the database also beyond the Angevin period, resolving abbreviations, geo-tagging, adding persistent identifiers, and linking to the extensive editions and facsimiles of the charters.
Other repeat guests at the conference series are the two well-known German research institutions of the Monumenta Germaniae Historia (MGH) and the Regesta Imperii (RI), continuously reporting on the progress of their digitization processes and the progress of their respective databases. As in Munich in 2007, the status of the MGH, curating critical editions of mediaeval sources relating to German and European history since 1819, was reviewed once again by CLEMENS RADL (Munich), who also did so in Munich in 2007. After a general introduction to MGH history (Freiherr von Stein and Theodor Sickel) and the structure of the MGH Diplomata printed editions, he gave a brief overview of the MGH digital “ecosystem”, including new research results in the “Additions” database, seal collections, the Stumpf data on GitHub etc., demonstrated by the example of the Diplomata Karolinorum. The most recent openMGH project aims to successively provide TEI-XML files of all the printed editions in collaboration with partner projects like PaLaFra, Formierung des Lehnswesens, Corpus Corporum, and the Latin Text Archive.
The RI, which had been protagonists of the 2007 as well as the 2013 conferences, were represented by STEFFEN KRIEB (Mainz). He argued that the preface to their first volume by the founder Johann Friedrich Böhmer could be read as encouraging the use of the regesta for digital diplomatics purposes, when he pointed out that the value of the regesta not only lies in being a preliminary work for a full-text edited cartulary but also as a research source on their own. For those purposes, he concluded that further improving the machine-readability of the data is a focus of the RI branch in Mainz.
Exploring new data and the creation of new databases has been a topic from the very beginning of the Digital Diplomatics conference series, but so has been the subject of gathering sources from multiple existing collections and linking the knowledge of different projects. It is no surprise that also in 2022 this collaborative approach was very prominent: JULIEN THÉRY (Lyon) presented the 2017 launched APOSCRIPTA project, whose scope is to gather a unified corpus of as many mediaeval papal documents as possible (mostly litterae) on the TELMA platform of the Institut de recherche et d’histoire des textes (IRHT). It currently holds the texts and metadata for more than 25.000 letters, decretals, and privileges, with a strong concentration in the 13th century. The project is intended to be a collective and collaborative addition to ongoing research efforts like the third edition of the Regesta Pontificum Romanorum or the Index actorum Romanorum pontificum.
JAN BURGERS and RIK HOEKSTRA (Amsterdam) presented the Digitale Charterbank Nederland (DCN)[6], a database to gather the vast corpus of mediaeval and early modern charters preserved in the archives of the Netherlands. It already features (as of September 7th 2022) 205,625 charters from 4,235 archival funds in 95 archival institutions. In order to facilitate fast setup, most of the data comes from a collaboration with the company De Ree, whose software is being used by the majority of Dutch archives. Data from the archives not working with said software was in most cases exported automatically and added manually. The preference of mass import over a thorough diplomatic exploitation led to a less-detailed searchability. Yet,the speakers demonstrated for the case of the term “erfpacht” and its corresponding legal phenomenon, how it is still possible to track major historical developments through well-thought-out search terms. In light of the described generation of the digital data, the two emphasised the importance of thorough source criticism for dealing with databases.
Several speakers were new to the series but covered subjects continuing the discussions started at the earlier instances about how to create and make use of charter databases. JOANNA TUCKER and JOHN DAVIES (Glasgow) discussed the research tools of the Models of Authority: Scottish Charters and the Emergence of Government 1100–1250 project[7], funded by the Arts and Humanities Research Council (AHRC) from 2014 to 2017, which is continued since 2021 by Stewart J. Brookes and Joanna Tucker. It contains the images and transcripts of the surviving Scottish royal charters from 1100 to 1250, with annotations differentiating charter text structures, clauses, graphs, persons and places which are also supported by the search function. They showed its interoperability with the People of Medieval Scotland 1093–1371[8] database which links to charter images stored in Models of Authority. Finally, they presented a case study on boundary and holding, as well as warrandice clauses, summarising that Scottish royal charters became standardised from the beginning of the 13th century, following English models.
STEFAN BRÖHL (Karlsruhe) spoke about the joint efforts of six Southern German archives to collect, describe, and publish about 7,000 diplomatic units and their metadata of the counts palatine of Rhine from 1449–1508. The collaborative project aims to gather the results in Archivportal-D, the prototype of an online-repository to provide unified and standardised metadata for the fonds of the collaborating archives.
Archivist KRISZTINA ARANY (Budapest) introduced the audience to general facts about the history of the Hungarian National Archives and their research portals as e.g. their Digital Archives Portal[9] and Hungaricana[10], focussing especially on the “Reformation” database which is continuing an earlier database of medieval charters with material dating from the Battle of Mohács in 1526 up to the year 1571, and the “Vestigia” database collecting Hungarica (archival documents with provenience or relation to Hungary) stored in Northern Italian archives. Her colleague in research, BENCE PETERFI (Budapest), then demonstrated the use of those databases by a case study on the Teutonic order in Hungary.
While these resources focus on charter research in historical sciences, MARGUERITE DALLAS (Zurich) discussed it from a linguistic point of view. She is doing her research together with Martin Glessgen. The database Documents et analyses linguistiques de la Galloromania médiévale (GallRom) has been programmed at the Linguistic Research Infrastructure Department at Zurich University since 2020, aiming to combine bibliographical data from the Bibliographie des textes galloromans médiévaux (GallRomBibl) and the editions of the Documents linguistiques galloromans (DocLing), with lexical data from reference dictionaries in the network Lexique étymologique de la Galloromania médiévale (LEGaMe) and philological data as, for instance, the scriptological analyses of the Analyse scriptologique de la Galloromania médiévale (Scripta) project. The completed GallRom database will be a rich philological research tool with data for all four Gallo-Romance languages French, Francoprovençal, Occitan, and Gascon. It will allow for differentiated lexicological and grapho-phonetic analyses, for example on the linguistic evolution from Latin to Romance, the geochronological changes of vocabulary in mediaeval documentary texts, and the development and diffusion of writing in the Gallo-Romance territory.
Having been touched only marginally at the previous conference instances[11], databases using Semantic Web annotation can be seen as a rather modern approach of database construction in our conference series. CHRISTIAN DOMENIG (Klagenfurt) gave an update to his presentation of 2007 and reported about his personal experiences working on a digital edition of the charters of the counts of Cilli (Quellen zur Geschichte des Alpen-Adria-Raumes). Initially, it was a home-made Microsoft Access based database. After losing the first database to stopped technical support and a major security break, it was then rebuilt by moving the project to SemanticMediaWiki. Domenig demonstrated how such tools could be used for zero-budget-projects and especially for teaching purposes.
ADAM ZAPALA (Warsaw) discussed a workflow for the processing of papal letters, which is part of the Monumenta Vaticana res gestas Polonicas Illustrantia project compiling papal letters for late mediaeval Poland (mainly from the 15th century). The project digitally processes existing printed editions, automatically extracting and normalising their metadata, to an XML database with persistent identifiers by Wikibase. An image indexing software based on GIS technology (INDXR) is used to annotate content information from the documents, treating their digital representations as maps. The information is stored in a PostgreSQL database. The final scope of the project is to make the annotated full text editions available by the use of TEI Publisher and a plugin connecting it to Wikibase. In addition to the presentation of the project, Zapala also shared his general thoughts on the great potential of digital tools like HTR, NLP and ontologies like Grotefend digital[12], as well as the linking of aligned data, for historical research.
A rather unique and modern collection of seals was presented during the talk by ÁDÁM NOVÁK and SÁNDOR ÓNADI (Debrecen). They introduced two digital projects of the 2014 born research group “Hungary in Medieval Europe” at the University of Debrecen. The research group continuously works on the digital source database Memoria Hungariae[13], where different kinds of relics of Hungarian relation (understanding the Hungarica term very broadly) from all over Europe are ought to be collected. The second project was an experimental one, using the reflectance transformation imaging (RTI) technique which John McEwan had already proposed in 2018 to use for mapping seal impressions. In cooperation with Mensor 3D Ltd., 12 medieval seals from the State Archive in Prešov had been photographed with different lighting and angles, before they were displayed in the exhibition “Gothic in the Monastery of Pilis – French Connections of Andrew II”. The speakers showed how “reconstructions” of seals could be put together from RTI data of multiple incomplete impression remnants from the same seal matrix, how the original seals’ preservation could be enhanced by exhibiting digital 3D images or 3D printed copies of the RTI captured originals, and that this also means another possibility of “bringing home” Hungarica seals whose original impressions remain in collections abroad. After the talk, the audience also had the opportunity to take a closer look at copies of some of the mentioned seals that the presenters had brought with them.
Another promising aspect of modern database development, the use of crowdsourcing, was addressed by AMABLE SABLON DU CORAIL (Paris). He discussed some general trends in the French public archives since the digital revolution: The archivists’ work has changed from producing new descriptive data to making widely available any data that until now was hardly accessible and hence ignored by most scholars. Also, the audience of the archives’ work has changed to smaller interest communities who are volunteering to contribute to the indexing and descriptive work, leading to a revival of the inventaires analytiques type of inventories. To allow a community-based annotation, the French National Archives have developed a web tool for HTR and the conversion of archival descriptions into XML/EAD encoded data (Saisie d’Inventaires Manuscrits Assistée par Reconnaissance Automatique (SIMARA)). The images of the archival files and finding aids are uploaded in IIIF standard to the servers of TEKLIA, where also the document processing platform Arkindex with the SIMARA user interface and the calculation servers are hosted. The SIMARA tool then allows the archives to benefit from crowdsourcing to obtain the annotation data for the analytical inventory of the digitised documents. Traditionally, there is a big difference between the approaches of archivists and diplomatists in accessing documents: Where archivists strictly organise in fonds based on provenance, diplomatists build corpora based on contextual features. The “Records in Contexts” conceptual model proposes a “record set” solution which could substitute the former fonds, showing the document’s network of records, agents, and institutions. To achieve this change in frames of references, massive amounts of descriptive data are needed, in which crowdsourcing tools like the presented SIMARA will play a big role.
While many talks of the conference dealt with the provision of source material in the form of sustainable online databases, a significant part of the talks showcased the use of such data with machine learning approaches for specific research questions. These topics have only rarely been discussed at the previous conference instances until 2013. For instance, automated text recognition and digital palaeography continue to be an important topic in the digital humanities, as shown by three talks in Graz. PETER A. STOKES (Paris) recalled his personal research history in digital diplomatics and palaeography (including the projects LangScape, ASChart, eSawyer, DigiPal, Models of Authority, and eScriptorium). He reminded us of the importance of modelling activity, e.g. in defining “what a charter is”. As the several projects did not share a common definition, it was impossible to realise the vision of the aggregate AngloSaxon Cluster. He spoke on the importance of sharing data and some limitations of machine learning approaches. He emphasised just how much the results of a machine learning model depend on its training data, rather than on the software itself, and therefore how valuable data is. We were encouraged to collaborate and share as much training data as possible, even when this might be challenging due to differences in transcription and segmentation standards, images, and script-style.
SERENA AMMIRATI and PAOLO MERIALDO (Rome) presented two projects of their research group at the Roma Tre University, starting with the In Codice Ratio project on two different HTR approaches for papal registers of Honorius III. from the Vatican Apostolic Archive. One experiment was a neural network for the automatic recognition of single characters in an over-segmented image and an automatic decision based on sequence probabilities for the candidate transcriptions. Here, they tested a crowd-based approach for the production of its training data where school students recognized symbols based on pattern matching. The other, seemingly more promising HTR approach recently tested was based on experts’ transcriptions of entire lines for the training data instead. Their current project Matrices is a sequel of the In Codice Ratio project, which, in collaboration with the State Archive in Rome, adapted versions of the first project’s machine learning approaches to 12th and 13th century documents from the monastery of SS. Cosmas and Damian in Mica Aurea, which then are analysed with further machine learning processes. In this project, they follow a two-stage self-supervised approach, training a neural network to represent images by the same scribe (scriniarius), and then another neural network capable of identifying similarities and differences in the data from different scriniarii. By applying the local interpretable model agnostic explanation (LIME) explainable AI (XAI) technique, they seek to understand which parts of the data the ML system considers as important for the classification process – knowledge that can be useful for palaeographers working on the documents as well, as it could help the experts to focus on details and possibly discover new features relevant for their palaeographic analyses.
TOBIAS HODEL (Bern) asked if the underlying mechanics, specifically the confidence matrices, of automated text recognition (ATR) could be used as tools for palaeographic questions as the identification of scribes. He pointed out that distant diplomatics can benefit from large models for HTR becoming available. He used the confidence matrix of a model trained on materials from the Abbey of Einsiedeln to make language-independent palaeographical statements. He came to the conclusion that, while HTR does have a different view on the charters compared to a palaeographer, it may not be suitable yet for scribe identification. Nevertheless, the study of its confidence matrices can help us to better understand how ATR algorithms are working.
The broad field of Computer Vision covers more than palaeography. It offers also other methods for diplomatics, as the following two talks demonstrated. The NOT A writtEn word but graphic symbols (NOTAE) project, presented by FRANCESCO LEOTTA (Rome), studies graphical signs drawn into documentary sources of all possible kinds (e.g. charters, letters, …) from the 5th to the 8th centuries. After pre-processing the collected document images, the project team segments them into boxes and classifies these as showing graphical signs or not. To augment the ground truth for object classification, the project team uses a generative adversarial network to produce AI-generated additional examples. The algorithm is continuously trained by implementing the feedback of experts in the field who add missing boxes and correct false positives, as the results obtained by machine learning are only meant to be suggestions for a manual evaluation. In the future, NOTAE aims for an automated suggestion and description of symbol categories, as well as the historical analysis of the studied material. Additionally, the team strives for publication of the knowledge graph created in the project as a public repository of graphic symbols in original documentary records from the Late Antiquity to early Middle Ages.
MARK BELL (London) and JOHN STELL (Leeds) promoted the use of qualitative spatial representation (QSR) as a method in tackling digital diplomatics challenges. The method tries to use spatial relationships in document layout to infer meaning like causal or temporal relationships, e.g., via common patterns of spatial reasoning. The continuous and in-progress work at the National Archives (London), which uses this technique in the analysis of historical documents, is a proof of its vast potential – especially in the analysis of registers, as, for example, the close rolls of Henry III.
The well-established field of natural language processing (NLP) was represented in multiple talks during the conference. TIMO KORKIAKANGAS‘ (Helsinki) talk covered digital linguistic studies on Late Latin charters. He is the curator of the Late Latin Charter Treebank (LLCT), a set of three dependency treebanks of about 1200 Early Medieval Latin documentary texts, mainly from Tuscany. He demonstrated in three case studies how distant approaches to the aforementioned source material could look like and how linguistics and philology themselves can profit from their results as well. Similar grammatical errors or usage of formulaic “templates”, for instance, could lead to conclusions on Latin teaching, scribal and chancery schools, and educational ideals which then would reflect back into linguistic research.
OLIVER SCHALLERT (Munich), CARSTEN BECKER (Berlin), and HELMUT SCHMID (Munich) covered areal variation in Middle High German, especially its methodological and quantitative aspects. The researchers showed that a vast, electronically accessible corpus of Middle High German charters provides fine-scaled data on area-specific language variation over the study period, unlike other historical documents. With the help of regular expressions drawn from the Wörterbuch der mittelhochdeutschen Urkundensprache (WMU) and semi-automated annotations created by Schmid’s RNNTagger[15], one can study fine-scaled grammatical variation or the transmission of linguistic innovations. Schallert demonstrated that contraction forms, i.e., occurrences of shortening that occur with multiple verbs in this language period, such as lazen/lan (“to let”), haben/han (“to have”), have a distinct areal profile, which was corroborated by their date profile and leads to further questions in historical linguistics.
JEROEN DEPLOIGE and MARIJKE BEERSMANS (Ghent) presented the Diplomatica Belgica database, publicly available since 2015 and previously discussed at the Digital Diplomatics conference by Els De Paermentier in 2013. The ongoing work on it aims at expanding its corpus content, adding information on toponyms and anthroponyms, and rendering the database’s access more effective and user-friendly. The talk continued with a case study of named entity recognition (NER) in a small corpus of 155 Middle Dutch charters and 5,000 tokens. For this task, two neural networks have been used: A custom bidirectional long-short term memory (Bi-LSTM) model and the state-of-the-art spaCy library., The latter outperformed the speakers’ own model. They concluded that, while both of these models performed worse in the Middle Dutch case study compared to the recent work[16] on French medieval charters, the results still showed that the application of NER methods to bigger Middle Dutch corpora is feasible and promising.
MARLENE ERNST and ALINA OSTROWSKI (Passau) covered their recently started research in a DFG funded project on papal-delegated jurisdiction in the Iberian Peninsula of the 12th century based on the data collected in the Iberia Pontificia project and other sources like the Regesta Imperii and printed editions as the “Papsturkunden in Spanien” series and “Papsturkunden in Portugal”. The speakers’ current work focuses on the application of text re-use detection to identify formulaic language. They hope that building of these formulaic expressions allows automated detection of diplomatic charter sections. While the project currently still is in the source-work phase, another big task for the future will be different historical network analyses to gain insight on the personal and institutional spread of delegated jurisdiction in the studied region, and on spatio-temporal clusters in the context of this phenomenon.
EVELINE LECLERCQ (Ghent) applied stylometric methods to aid her philological interpretation of the charters of the bishops of Cambrai. Eveline Leclercq analysed over one thousand charters using the Diplomata Belgica and Chartae Galliae. She created a bag-of-words model with 1000 most frequent words to create a ‘document vector matrix’. TF-IDF normalisation and square distance matrices combined with close reading allowed her to identify members of the chancery. It has to be noted, that even the close reading could apply digital methods for text-reuse detection like Text-Matcher (J. Reeve)[17] and Medieval Latin (Manjavacas)[18]. Leclerq concluded by pointing out the need for a larger corpus and database functionalities applied to it to realise a combination of distant and close reading methods. The procedure, the report’s author emphasised, not only allows for a thorough examination of the papers, but also allows for the identification of editorial or ecclesiastical influence.
Statistical approaches are part of traditional diplomatics questions and now supported by modern querying possibilities, as shown by MIGUEL CALLEJA-PUERTA (Oviedo) who represented the Notariado y construcción social de la realidad. Hacia una codificación del documento notarial, siglos XII–XVII project (NotFor). The project pursues a better understanding of the public notariat in the Portuguese and Spanish kingdoms, based on TEI-annotated documents for which, because of the limited possibilities of TEI in describing diplomatic structure, a new tag set was developed. Calleja-Puerta presented the diplomatics aspects of the project, emphasising the possibilities of domain-specific XML markup and XQuery-based selections. He demonstrated this with his recent research results on the public notariat of the Galician city of Ribadavia and the comparison of the usage of tax clauses (cláusulas de plusvalía) and penal clauses in the urban environment of Oviedo and the rural environment of Grado, both Asturian cities.
The exploitation of machine learning methods for digital-born statistical questions played a more prominent role now at earlier instances of the conference series. NICOLE BERGK-PINTO (Paris), JACQUELINE SCHINDLER (Graz) and NIKLAS TSCHERNE (Graz) presented their ongoing project BeCoRe – Between Composition and Reception. The project studies visual signs of authority (seals, monograms, initials, other graphical signs like the chrismon, etc.), but also takes textual features and layout into account, investigated with machine learning/Computer Vision approaches. The theoretical foundation of the project’s hypothesis is Fichtenau’s concept of “diplomatic landscapes”, i.e. geographically localised traditions in both the internal and external composition of charters. One approach is comparing clusters created by computational methods: Do they relate to regional or systematic categories (monastic order, monastery, or others)? In the current phase of corpus creation from various French and Austrian charter collections in the Monasterium.net platform, the authors of the paper could already observe linguistic differences in the Cluny corpus which are related to the location of the recipients of the charters. The next stage will be to work with machine learning techniques. It was emphasised how much a machine learning-driven clustering approach must be aware of the pitfalls of cherry-picking results and pushing the training process to return results that one expects and has decided on beforehand.
PIA GEISSEL (Wuppertal) talked about the opportunities and possibilities of stylometric approaches to mediaeval charters, a method she applied to Brescian and Milanese charters from 10th to 12th century in the Codice diplomatico della Lombardia medievale (CDML)[19] corpus. She reported on the necessary pre-processing steps, including the quality of source data. Her experiments with principal component analysis (PCA) on temporal and geographic classes in the CDML showed the homogeneity of the material, which was supported in the discussion with the audience; the digital corpus is based on the selection by Maria Franca Baroni with the plan to be homogeneous, and this is not necessarily representative for the charter production in the Milan area in this period. This showed the importance of corpus creation in the application of digital methods.
JOHN MCEWAN (St. Louis) talked about the problem of dating seal matrices from archaeological finds. In contrast to seal impressions attached to charters, whose chronological classification can usually be based on the dating clauses of the respective documents, matrices lack this important information. Therefore, only a few of them have been dated at all, in particularly elaborate research efforts. McEwan trained a machine learning algorithm to first extract information on the shape and class (meaning the depiction seen on the seal) from the seals found on the Digital Sigillography Resource (Digisig)[20] platform of the Saint Louis University Centre for Digital Humanities, designed by John McEwan himself and currently containing data for more than 42,000 seals. His algorithm was used to automatically date seals, working with the previously obtained information. As a last step, he compared the outcomes of his machine learning approach to datings from the Portable Antiquities Scheme Dataset.
The aforementioned shift from traditional concepts to higher statistics was shown in the conference’s keynote as well. MICHAEL GERVERS and GELILA TILAHUN (Toronto), from the Documents of Early England Data Set (DEEDS) project[21], have been frequent guests of the Digital Diplomatics series. The primary goal of DEEDS was to provide digitised English documents from mainly the 12th to 13th century. It became evident that only a small subset of the primary texts from the Battle of Hastings (1066) until the early years of King Edward II’s (1284–1327) reign carried fixed dating clauses, turning assigning dates to these charters into a significant problem for English diplomatics. By the use of two-word shingles and statistical analyses, the DEEDS team was able to date any charter from this corpus with an accuracy of ±7 years. Currently, they apply topic modelling, using a latent dirichlet allocation (LDA) model, to examine how certain economic and social changes in England led to the change of patterns in the vocabulary and style of the investigated documents. The researchers were able to identify eight topics (“Choosing how many topics to identify is more of an art rather than science.”) in this way that could be attributed to certain time frames in English history. This led to an understanding of the circumstances under which a charter would be internally dated in the time before 1189. A network analysis measuring the correlation distances between ‘communities’ of word terms as nodes already showed some visible pattern breaks. Gervers concluded that standalone databases like the DEEDS database are too limited to sufficiently answer this kind of diplomatic questions. Therefore, he advocated for gathering, standardising and linking more data and repositories, summarising his wishes for the future of digital diplomatics in the bi-gram “collaborative cooperation”.
The Digital Diplomatics Conference 2022 also offered the opportunity to briefly present research in the condensed form of a poster session. LAURA BITTERLI (Zurich) presented the e-learning platform Ad fontes where students can acquire basic analogue skills in e.g. palaeography which are still necessary when working with digital tools like Automatic Text Recognition (ATR). A new learning module on the latter topic is in preparation, helping students and researchers to overcome the initial hurdles that come with its practical application. MARTINA BOLOM-KOTARI (Hradec Králové) discussed the problems and chances of the digital edition of seals. She showed that their representation in the form of an online database is in the interest of both archivists and researchers, since – if done well and in full consideration of the sphragistics needs in terms of high-quality photos and detailed metadata as well as the connection to the document the seals were attached to – it could fully replace the original for research proposes and therefore save the originals from the risks of further abrasion coming with even their most careful use. ZSOLT SIMON (Târgu Mureș) gave an overview of six databases for Romanian charters, outlining their advantages and shortcomings in comparison with their predecessors. SERAFINA FILIPPELLI (Bergamo) presented her work on the credit and debt register (Registro di Silvestro de Benedictis) of the monastery Santo Sepolcro of Astino, highlighting how important the digital images of the register entries are for the understanding of its external characteristics and context (e.g. the registers structure and watermarks on the pages). Last one to shortly speak on his poster was GIUSEPPE CONSOLO (Naples) whose project aims to produce a digital edition of account books from the Dominican monastery of SS. Pietro and Sebastiano from the second half of the 15th century which will be published as a part of the Graz-based Digital Edition Publishing Cooperative for Historical Accounts (DEPCHA)[22] project. The ultimate goal of the project will be the presentation of data taken from the edited account books in an interactive form at the Museum of Agricultural Science at Portici.
Summarising, one common thread in many talks of the conference was the importance of well-processed and sustainable data which gains its value from being accessible and useful to not only singular projects but also further research. This is a core concern of the project which organised the conference, the ERC Advanced Grant From Digital to Distant Diplomatics (2022–2026)[23], which is working on a conference volume to be published as a Beiheft of the Archiv für Diplomatik. Collaboration was a keyword of the conference, as also stressed by Michael Gervers at the end of his keynote. Many of the conference’s speakers and projects had already been present at earlier editions of the series, where they presented their plans to produce data. Now this data is available and we can start to analyse it, which is an encouraging message to junior researchers: while some of the applied machine learning methods might not necessarily be new approaches and well-established in their respective computational research fields, their application to corpora of diplomatic sources is rather modern and results in new and exciting research questions for our traditional discipline. They allow us to climb on the shoulders of the data giants that have emerged in the last 15 years, with new ideas and unwavering enthusiasm.
Franziska Decker and Georg Vogeler
References:
[1] https://www.didip.hypotheses.org/conference-2022/digital-diplomatics-conference-series
[2] https://www.monasterium.net/
[3] https://www.crossborderarchives.eu/
[4] https://www.pammap.sk/
[5] http://www.eruditio.hu/lectio/mokka-ms
[6] https://charterbank.huygens.knaw.nl
[7] http://www.modelsofauthority.ac.uk/
[8] http://www.poms.ac.uk/
[9] https://www.eleveltar.hu/
[10] https://www.hungaricana.hu/en/
[11] See the contribution by Aleksandrs Ivanovs / Aleksey Varfolomeyev, Some Approaches to the Semantic Publication of Charter Corpora. The Case of the Diplomatic Edition of Old Russian Charters, in: Digital Diplomatics. The Computer as a Tool for the Diplomatist?, Antonella Ambrosio et. al. (ed.), Archiv für Diplomatik, Schriftgeschichte, Siegel- und Wappenkunde. Beiheft 14, Köln 2014, pp. 149–68.
[12] https://gams.uni-graz.at/grotefend
[13] http://lendulet.memhung.unideb.hu/
[14] http://www.notae-project.eu/further-info/project
[15] https://www.cis.uni-muenchen.de/~schmid/tools/RNNTagger/
[16] Sergio Torres Aguilar / Dominique Stutzmann, Named Entity Recognition for French Medieval Charters, in: Mika Hämäläinen et al. (ed.), The Association for Computational Linguistics Anthology. Proceedings of the Workshop on Natural Language Processing for Digital Humanities, Silchar 2021, pp. 37-46.
[17] https://github.com/JonathanReeve/text-matcher
[18] https://github.com/emanjavacas/text-reuse
[19] https://lombardiabeniculturali.it/cdlm/
[20] http://www.digisig.org/
[21] https://www.deeds.library.utoronto.ca/
[22] https://www.gams.uni-graz.at/depcha
[23] https://www.didip.eu
A shortened version of this conference report can be found at www.hsozkult.de.

