This paper describes an ongoing work on the construction of a SKOS thesaurus to support multi-disciplinary studies on the transmission of zoological knowledge throughout historical periods, combining the analysis of ancient literature, iconographic and archaeozoological resources. We rst describe the I2AF, a national archaeozoological and archaeobotanical inventory database integrating data from archaeological excavation reports. Then we describe the TAXREF taxonomical reference designed to support studies in Conservation Biology, that was enriched with bioarchaeological taxa from I2AF. Finally we describe the TAXREF-based SKOS thesaurus under construction and its intended use within the Zoomathia research network.
Animal bones and plant remains from archaeological excavations are a rich and
original source of information on the history of biodiversity and its
interaction with human societies. When compared with the knowledge about diversity
and current locations of human populations, these remains help to gure out
the scenarios of past extinction, biological invasions and anthropic impact. This
is particularly true during the Holocene, when the in uence of human
activities overrode that of climatic factors. Therefore, gathering archaeozoological
and archaeobotanical data in a sustainable bioarchaeological database, publicly
available, represents a major challenge for Natural Sciences and Conservation
Biology. Archaeozoological and Archaeobotanical Inventories of France database
[
Historians address a related challenge. Identifying the reported species in
ancient literary and iconographic resources, and assessing the documentation is a
momentous issue of the History of Zoology. An increasing amount of primary
material (such as textual or iconographic resources) is encoded in domain-speci c
digital formats. For instance, the SourcEncyMe3 and Ichtya4 projects aim to
encode mediaeval encyclopedias in the XML-TEI standard5 while adding
manual annotations with regards to mediaeval compilers, author sources and taxa.
These works succeed in making material about mediaeval scienti c knowledge
more easily exploited by a broad scienti c community, and support researchers
studying e.g. the transmission of zoological knowledge throughout historical
periods. Yet, the sharing with related scienti c communities remains hampered by
the lack of formal semantic reference and terminological standards. For instance,
the dolphin is a research topic for modern studies on biodiversity, for
archaeozoologists, as well as for studies on Greek mythology wherein the dolphin played an
important symbolic role [
Those challenging questions can be addressed through the use of controlled
and widely accepted semantic references. A reference thesaurus shared by sibling
scienti c disciplines would help to clear the many misinterpretations or con
ations made by ancient authors and debated at length in modern critic literature
referring to Ancient sources (from P. Belon, 1551, to I. Geo roy Saint-Hilaire,
1841). The Zoomathia research network7 addresses this challenge, speci cally on
the study of rich mediaeval compilation literature on Ancient zoological
knowledge, supported by archaeological and iconographic knowledge. The Semantic
Web provides powerful models and technologies for connecting and sharing pieces
of data while making their semantics explicit. RDF facilitates the combination
and sharing of di erent data sets thanks to the underlying Web technologies
and the subsequent Linked Data paradigm. Zoomathia intends to leverage those
technologies to annotate and link together various medieval compilations such as
the Hortus Sanitatis 8, archaeozoological data (I2AF database) and iconographic
material. In this context, we chose the TAXREF [
medieval/programme-sourcencyme-corpus-et-sources-des-encyclopedies-medievales 4 http://www.unicaen.fr/recherche/mrsh/document numerique/projets/ichtya 5 http://www.tei-c.org/index.xml 6 https://www.unicaen.fr/puc/sources/depiscibus/accueil 7 http://www.cepam.cnrs.fr/zoomathia/ 8 This very popular text that enjoyed numerous editions and translations between 1491 et 1547 is not only a landmark in the history of encyclopedias, but also, concerning the naturalistic knowledge, representative of the whole medieval tradition. It provides taxonomy to build a SKOS thesaurus supporting the integration of these heterogeneous data sets.
This paper is organized as follows: Section 2 presents the I2AF project. Section 3 describes the TAXREF taxonomical reference. Then, section 4 presents our ongoing work on the construction of a SKOS thesaurus based on TAXREF. Finally, section 5 concludes and suggests leads for future works. 2
During the eighties decade, it was acknowledged that the access to archaeological data by researchers was increasingly challenged by the growing amount of data produced, and hampered by its scattering. The risk of permanent loss was even more worrying. Thus, it appeared obvious that data in archaeological reports had to be systematically and sustainably collected and inventoried, in a heritage perspective, while making them available to all potential users. From 2003 on, several programs supported by multiple French institutes designed, deployed and maintained such a national inventory database. Today, the I2AF is a collection of the French National Museum of Natural History (MNHN). It is continuously and increasingly populated with data on ora and fauna from reports of all excavations performed in French territories, whether the bioarchaeological material was already studied or not. Since January 2014, the inventory and knowledge dissemination e ort has been actively sustained by a national multi-institute network of bioarchaeologists9. When data from excavation reports is imported into the I2AF, it is aligned on two thesauri: a chronocultural thesaurus provides temporal terms with regards to cultural periods (the oldest records date back to the Middle Palaeolithic), and a taxonomic thesaurus of zoological and botanical names, namely the TAXREF taxonomical reference (see section 3).
As the national reference for nature and biodiversity, the MNHN is responsible for scienti c and technical coordination of the natural heritage inventory. To this end, it develops and distributes the TAXREF taxonomical reference, and maintains the National Inventory of Natural Heritage 10 (INPN), an information system that gathers current (contemporary) occurrence data on fauna and ora of metropolitan France and overseas departments and collectivities. To date, INPN gathers data from approximately 800 data sources aligned on TAXREF. In this context, the I2AF was naturally identi ed as a potential data contributor to the INPN. This was however challenging due to the discrepancies between both databases in terms of temporal periods and inventoried species. Indeed, while the INPN gathers actual environmental data on wild life, the I2AF also most of the data available between 1260 and 1320 in western Europe, derived from the late antiquity compilations. 9 GDR 3644 BioArcheoDat, "Societies, biodiversity and environment: archaeozoological and archaeobotanical data and results on the French territory". 10 Inventaire National du Patrimoine Naturel: http://inpn.mnhn.fr. Museum National d'Histoire Naturelle [Ed]. 2003-2015. provides archaeological data on domestic species, exotic species (not inventoried on any French territory, notably imported by menageries as soon as Roman Antiquity) and possibly extinct species. This issue was solved progressively by enriching TAXREF with new taxa along with the integration of I2AF data into the INPN. As examples we can cite extinct species such as the mammoth and the cave bear, domestic species such as the dog and the ox, and exotic species such as the Barbary macaque. 3
TAXREF: a Taxonomic Reference in Conservation
Biology
TAXREF[
TAXREF can be browsed on the INPN web site, and downloaded in TSV format (tab-separated values). An on-going work aims to set up a Web service allowing to query the taxonomy and retrieve results in XML or JSON formats. TAXREF is organized as a unique, controlled, hierarchical list of scienti c names. Conceptually, it consists of a table wherein one row uniquely describes one scienti c name. All taxonomical names are presented in the same way, whatever their taxonomical rank. Most salient elds are listed below: { CD NOM : unique identi er of the scienti c name. { CD SUP : identi er of the upper taxon in the classi cation. { CD REF : identi er of the reference taxon. This may be either the same as CD NOM or a di erent one. In the latter, CD NOM identi es a synonym of the reference name identi ed by CD REF. { Nom: taxon scienti c name. { Nom Vern and Nom Vern Eng : French and English vernacular names. { Auteur : taxon authority (author name and publication year). { Rang : taxonomical rank (phylum, class, order, family, gender, species...), represented by a code of two to four letters. { HABITAT : type of habitat in which the taxon usually lives marine, fresh water, terrestrial...) coded as values from 1 to 8. { A set of biogeographical statuses, one for each geographical region (metropolitan France and overseas departments and collectivities). E.g.: P stands for present, E for endemic, X for extinct, etc. 11 http://www.naturefrance.fr/sinp/presentation-du-sinp
As an example, Listing 1.1 shows a JSON excerpt describing the common
dolphin using its reference scienti c name Delphinus delphis, and its synonym
Delphinus tropicalis. Annotation "HABITAT":1 states that it lives in a marine
habitat. Annotation "Rang":"ES" states that the taxon belongs to the species
taxonomical rank (ESpece in French). Annotation "GUA":"P" states that its
biogeographical status is P (present) in Guadeloupe, a French overseas department.
A comprehensive description of allowed values for the habitat, taxonomical rank
and biogeographical status is provided in [
Listing 1.1. Example of a JSON representation of TAXREF entries
Currently, more than 450.000 taxa are registered, covering the continental and marine environments. From the temporal perspective, all current living beings are considered as well as those of the close natural history, that is, from the Palaeolithic until now. Usage statistics12 attest the large variety of people using TAXREF, far beyond the research community: botanic conservatories, associations, public institutions and collectivities, private companies, individuals. Given its wide adoption in various communities, we chose it to build a SKOS reference thesaurus that should be published and linked on the Linked Data. 4
A TAXREF-based Thesaurus for the Linked Data In this section we present our ongoing work on the creation of a SKOS vocabulary faithfully representing the TAXREF taxonomical reference. SKOS13 is the acronym of Simple Knowledge Organization System; it is a W3C standard designed to represent controlled vocabularies, taxonomies and thesauri. It is used extensively to bridge the gap between existing knowledge organisation systems and the Semantic Web and Linked Data. 12 TAXREF usage statistics are not published publicly but can be provided on demand. 13 http://www.w3.org/2009/08/skos-reference/skos.html 1 @prefix skc: <http://www.w3.org/2004/02/skos/core#>. 2 @prefix skx: <http://www.w3.org/2008/05/skos-xl#>. 3 @prefix tc: <http://lod.taxonconcept.org/ontology/txn.owl#>. 4 @prefix gn: <http://www.geonames.org/ontology#> . 5 @prefix nt: <http://purl.obolibrary.org/obo/ncbitaxon#> . 6 @prefix taxr: <http://inpn.mnhn.fr/taxref/>. 7 8 <http://inpn.mnhn.fr/taxref/taxon/60878> a skc:Concept; 9 skx:altLabel <http://inpn.mnhn.fr/espece/cd_nom/60881>; 10 skx:prefLabel <http://inpn.mnhn.fr/espece/cd_nom/60878>. 11 skc:broader <http://inpn.mnhn.fr/taxref/taxon/191591>; 12 taxr:hasHabitat <http://inpn.mnhn.fr/taxref/habitat#Marine>; 13 taxr:bioGeoStatusIn [ 14 taxr:bioGeoStatus <http://inpn.mnhn.fr/taxref/bioGeoStat#P>; 15 gn:locatedIn <http://sws.geonames.org/3017382/> ]; 16 taxr:bioGeoStatusIn [ 17 taxr:bioGeoStatus <http://inpn.mnhn.fr/taxref/bioGeoStat#P>; 18 gn:locatedIn <http://sws.geonames.org/3579143/> ]; 19 taxr:bioGeoStatusIn [ 20 taxr:bioGeoStatus <http://inpn.mnhn.fr/taxref/bioGeoStat#B>; 21 gn:locatedIn <http://sws.geonames.org/935317/> ]. 22 23 <http://inpn.mnhn.fr/espece/cd_nom/60878> a skx:Label; 24 taxr:isPrefLabelOf <http://inpn.mnhn.fr/taxref/taxon/60878>: 25 skx:literalForm "Delphinus delphis"; 26 tc:authority "Linnaeus, 1758"; 27 nt:has_rank <http://inpn.mnhn.fr/taxref/taxrank#Species>; 28 taxr:vernacularName "Dauphin commun a bec court"@fr; 29 taxr:vernacularName "Short-beaked common dolphin"@en. 30 31 <http://inpn.mnhn.fr/espece/cd_nom/60881> a skx:Label; 32 taxr:isAltLabelOf <http://inpn.mnhn.fr/taxref/taxon/60878>; 33 skx:literalForm "Delphinus tropicalis". 34 tc:authority "Van Bree, 1971"; 35 nt:has_rank <http://inpn.mnhn.fr/taxref/taxrank#Species>; 36 taxr:vernacularName "Dauphin commun d’Arabie"@fr; 37 taxr:vernacularName "Arabian common dolphin"@en. 38 39 <http://inpn.mnhn.fr/taxref/taxrank#Species> a skc:Concept; 40 skc:prefLabel "Species"@en; 41 skc:exactMatch 42 <http://purl.obolibrary.org/obo/NCBITaxon_species>; 43 skc:exactMatch 44 <http://rdf.geospecies.org/ont/geospecies#TaxonRank_species>. 45 46 <http://inpn.mnhn.fr/taxref/habitat#Marine> a skc:Concept; 47 skc:prefLabel "Marine habitat"@en; 48 skc:relatedMatch 49 <http://lod.taxonconcept.org/ontology/txn.owl#MarineHabitat>; 50 skc:exactMatch 51 <http://purl.obolibrary.org/obo/ENVO_00002227>.
Listing 1.2. Example SKOS representation of TAXREF entries
Listing 1.2 shows the proposed SKOS representation of the taxon presented in Listing 1.1, using the Turtle RDF syntax. The keystone of our modelling of TAXREF in SKOS is as follows. Each taxon is represented by a SKOS concept (line 8); its URI is in namespace http://inpn.mnhn.fr/taxref/taxon/, which local name is CD NOM, the TAXREF taxon identi er (see section 3). The skc:broader property is used to model the relationships between a taxon and the upper taxon in the classi cation (CD SUP). The reference scienti c name of a taxon and its synonyms are de ned as values of properties skx:prefLabel and skx:altLabel respectively (lines 9 and 10). They are URIs in namespace http://inpn.mnhn.fr/espece/cd nom/. These URIs have been de ned by INPN; today they are dereferenced to a Web page providing a HTML description of the taxon. The label literal values themselves are de ned with property skx:literalForm (lines 25 and 33). The habitat and biogeographical status are represented by a property value which subject is the URI representing the taxon (lines 12 to 21), while the authorities, taxonomical rank, and vernacular names are attached to labels (lines 26 to 29 and 34 to 37).
We identi ed existing vocabularies that can be reused in our context, keeping in mind that we wish to link the TAXREF thesaurus with existing, well-adopted data sets, in particular within the Linking Open Data cloud. We rst focussed on classes and properties that represent taxon characteristics (habitat, taxonomical rank, name authority, etc.). For example, taxonomical ranks are de ned in various ontologies such as the NCBI taxonomic classi cation14 and the GeoSpecies ontology15. Similarly, the type of habitat is commonly de ned in several ontologies such as the ENVO16 environment ontology. To keep full control over the TAXREF vocabulary, we chose to de ne terms (SKOS concepts) for the taxonomical ranks (lines 39 to 44), types of habitat (lines 46 to 51) and biogeographical statuses in a speci c TAXREF namespace (http://inpn.mnhn.fr/taxref/), and align them with concepts of existing vocabularies using the skc:exactMatch or skc:closeMatch properties. In future works, we intend to align the TAXREF taxa themselves with taxa in other well-adopted taxonomies.
To perform the translation of TAXREF into a SKOS vocabulary, we use
xR2RML [
In this paper, through a few simple example questions, we have highlighted
today's needs of some scienti c disciplines, as diverse as Conservation Biology,
Bioarchaeology, and Ancient literature, to gather and make sense of
heterogeneous data and material. Then, we have described I2AF, a national
archaeozoological and archaeobotanical inventory database integrating data from
archaeological excavation reports. We have presented the TAXREF taxonomical
reference designed to support studies in Conservation Biology. To meet the needs of
Archaeozoology and Archaeobotany, TAXREF was progressively extended with
taxa from I2AF. It is the rst taxonomical reference used to integrate data from
Bioarchaeology and Conservation Biology[
Then we have presented our ongoing work on the construction of a SKOS thesaurus based on TAXREF. In the context of the Zoomathia research network, we aim to use this thesaurus to support multi-disciplinary studies on the history and transmission of zoological knowledge throughout historical periods, combining the analysis of ancient and mediaeval literature, iconographic and archaeozoological resources. This will require the enrichment of the TAXREFbased thesaurus with philological and cultural information and its geographical extension to other Mediterranean areas (Greece, Italy, etc.). Besides, in order for a large community to bene t from this work, and to spur its adoption by linkeddata based applications, future works target the automatic creation of links with other well-adopted open data sets and thesaurus, may they be non-specialized like DBpedia, or domain-speci c like the NCBI taxonomical reference.