Biblissima is an online digital library, which provides easy and coordinated access to a huge and complex mass of documentation on manuscripts and early printed books, the texts contained therein, their circulation and their readers, from the 8th to 18th centuries. This workshop presentation will give an overview of the steps followed and decisions made along the way to releasing a rst prototype of the Biblissima portal: from mapping data to a common ontology, via the establishment of a thesaurus, to the technical development of a single interface and a common triple store for data deriving from di erent iconographic databases on medieval manuscripts.
The two main components of the observatory are a cluster of the project's data on manuscripts and early printed books currently found in as many as 40 databases in di erent formats and with di erent research interests (including illuminated manuscripts, history of the transmission of texts and history of collections) and a digital image repository. The databases will be interconnected using semantic web technologies and linked to a platform for digital editions and to the project's digital image repository. 1.1
In order to handle the heterogeneity of the database formats (MySQL, EAD,
TEI P5, UNIMARC, etc.) and the variety of Biblissima's data (manuscript
cataloguing databases, textual editions, iconographic databases) we have chosen
to use the CIDOC Conceptual Reference Model (Comite International pour la
Documentation Conceptual Reference Model [
CIDOC CRM is an accepted ISO standard (ISO 21127). As an event-centric ontology it covers di erent phenomena in space and time like provenance, copying of texts, creation of works and expressions, as well as the production of information carriers and attribute assignments. As CIDOC CRM and FRBRoo (which combines the CIDOC CRM approach with the common vocabulary for the transmission of works (WEMI) that is provided by the FRBR model) are generic models for the museum and library domains, it was decided to de ne a few more speci c classes and properties related to manuscripts, early printed books and illuminations. For example, within the scope of the Biblissima project a medieval manuscript is an instance of the class bibma:Manuscript, which is a subclass of frbroo:F4 Manifestation Singleton (\This class comprises physical objects that each carry an instance of F2 Expression, and that were produced as unique objects, with no siblings intended in the course of their production"). An instance of a bibma:Manuscript might be composed of several parts (bibma:Component) and might carry both text and illustrations.
As regards the illustrations, there are several possible modelling solutions
in CIDOC CRM, such as E38 Image (\This class comprises distributions of
form, tone and colour that may be found on surfaces such as photos, paintings,
prints and sculptures or directly on electronic media") or its subclass E36 Visual
Item (\This class comprises the intellectual or conceptual aspects of recognisable
marks and images"). These solutions have been adopted both for book
illustrations by the \Illustrations of Goethe's Faust " project [
:i a bibma:Illumination .
:c crm:P56 bears feature :i .
This is a shortcut for the fully developed path: :folio a crm:E53 Place .
:c a bibma:Component ; crm:P59 has section :folio . :i a bibma:Illumination ; crm:P53 has former or current location :folio .
Instances of E53 Place are a folio or a particular zone on a folio, for example.
The ontology interacts with a thesaurus of technical terms used in medieval studies (codicology, palaeography, iconography etc.) and descriptors used for indexing medieval illuminations in the project's databases. The data is structured in a thesaurus compliant with the international standard for thesauri and interoperability with other vocabularies (ISO 25964). The di erent lexical and semantic relationships that can be de ned between the descriptors will have an intrinsic (semantic) role, in that they will help to show the relationships of hyponymy or synonymy, as well as an extrinsic (technical) function for the search engine. In addition, the project's data on people, corporate bodies, places and titles are aligned with existing authority les and linked data repositories, such as Rameau, VIAF, and GeoNames. 1.2
The majority of Biblissima's databases contain descriptive and structural metadata for medieval manuscripts and early printed books, issued from the cataloguing of these documents or scienti c research, but the project also includes digital editions in TEI P5 of library inventories and texts and records on illuminations. Metadata like the date of creation and place of origin of a manuscript and its illuminations, the identi cation of the scribe, translator or commentator of the copied text, former owners (people and corporate bodies) of a manuscript throughout the centuries and lists of books kept in libraries at a given moment in time can be used to study the history of the texts and manuscripts, as well as reading and collecting practices.
In order to develop the portal step by step we have chosen to begin by creating a uni ed access point to two iconographic databases.
Using Semantic Web technologies, the Biblissima prototype aims to demonstrate the potential of the available metadata produced by the Biblissima project. It was developed using open source solutions and all the data is publicly available under an open licence in order to facilitate reuse.
The prototype is built on a subset of two iconographic databases: Mandragore
[
It provides federated access to a subset of data present in the two databases, such as illumination related data: caption, descriptor, folio carrying the illumination, illumination record, digital surrogate of the illumination, artist, context of the illumination (author and title of the textual work per artistic unit), date of origin and place of origin. The data set also contains manuscript related data such as shelfmark, common name, grouping, repository, digital surrogate, manuscript record.
The subset is limited, in the case of both databases, to records on illuminations indexed with at least one geographical descriptor, which equates to almost 5 000 descriptors for approximately 20 000 illuminations.
A SPARQL endpoint and a federated search engine make it possible to search
all the data in the cluster. Users can search by descriptor, artist, date of origin,
place of origin, author or work title, and can re ne their search with a series of
facet lters. The results are displayed in a user-friendly manner by grouping them
in lists. Pages about manuscripts, their units and illuminations include frames
that display the corresponding digital surrogate using IIIF manifests [
Both databases have been used to index manuscript illuminations for the last 25 years and di erent systems were chosen for structuring the descriptors. A polyhierarchical classi cation of Biblissima's thesaurus may make it possible to retain the original descriptor classi cations while reordering them in a new systematisation. However, these classi cations do not re ect the medieval practices of organising knowledge. The identi cation of the iconographic elements is sometimes based on internal information such as heading titles, chapter titles, inscriptions or notes present in the manuscript; in the absence of this kind of information, identi cation is dependent on the scholar's culture, especially in the case of living things and artefacts. This means that when using these descriptors to study medieval illuminations, we must keep in mind that their identi cation has a disparate chronological and cultural origin. Another feature of the indexing practices speci c to these databases is that the data does not provide co-textual information, and the contextual information is not always available. This makes quite di cult to trace the diachronic evolution of the meaning of a word and of its iconographic representation.
The technical solutions adopted by Biblissima open new avenues and yield new ways of searching through data that could contribute to the analysis of iconographic representations. On the basis of the geographic descriptors, one could attempt to answer several questions regarding the status of cities in artistic imagery and de ne the notion of the city through iconographic choices: what are the criteria which confer an urban identity to a community and what makes the di erence between an urban and a rural environment? From what point in time do cities begin to be represented and what cities are the most represented over the centuries? How could one explain the cases of single occurrence? Could one analyse the anachronistic representations of places, be they cities or battle elds? What are the most common descriptors associated with toponyms? 3
The semantic web solutions that Biblissima has chosen could be adapted in order
to provide answers to other kinds of research topics. As such, Biblissima's
polyhierarchical thesaurus makes it possible to establish new classi cations of the
descriptors that already exist in the databases and to recreate a medieval
taxonomy of living species as it was conceived by an encyclopedist or a physician, for
example. One might also connect the thesaurus to the digital edition of
exegetical texts such as the biblical Glossa [
By adopting common standards for the ontology and for the thesaurus, Biblissima's data might also be aggregated with and used by other semantic web projects in the future. http://