Tuesday, October 31, 2017

EADitor now supports EAD and MODS to IIIF manifest generation

After migrating the Newell TEI notebooks to support serialization of facsimiles into IIIF manifests and the render of these manifests in an embedded Mirador viewer, I implemented a transformation of EAD finding aid image collections and MODS records for photographs into manifests.

EAD updates

The EAD finding aids were updated to replace the daogrp's linking to flickr images to link to thumbnail, reference, and IIIF service URLs (dao[@xlink:role='IIIFService']). An XSLT transformation of the EAD into manifest JSON occurs, with an intermediate process of iterating through the IIIFService info.json files with the Orbeon XForms processor in XPL to extract the height and width to generate canvases for each image.

The Brett finding aid now includes clickable thumbnails that will launch the zoomable Leaflet viewer in a fancybox popup window. At the top of the page, the user can download the manifest, and there's also a link to view the manifest in our internal Mirador viewer. You can view the EAD XML (link at top) for more details.

MODS updates

The updates to the MODS were twofold. First, in the previous version of Archer, all photographs were suppressed from the public regardless of copyright concerns. We have re-evaluated these concerns by applying one of several Rights Statements. Two of these rights statements are most permissible, and therefore, we will display the high resolution image when we have every right to do so. In any case, thumbnails are Fair Use, and therefore, they are always visible in the record page and the search results pages.

Where copyright allows us to do so, the MODS file includes a URL for the reference image and a URL[@access='raw object' and @note='IIIFService']. When a IIIFService URL is present in the MODS record, the XSLT transformation will include a Leaflet div and initiate the display of the image. See A Portrait Photograph of Margaret Thompson, for example. Like the finding aid, a manifest is dynamically generated from MODS, but only one XForms processor is called to extract the height and width from the info.json for the single image linked in the MODS file.

Pelagios Updates

Since the Brett collection links many photographs to ancient places defined in the Pleiades Gazetteer of Ancient Places, I have updated the EADitor RDF output for Pelagios. The output now includes IIIF service metadata conforming to the Europeana Data Model specification. Rainer Simon has imported these photographs into Peripleo.

Friday, October 6, 2017

Newell notebooks migrated to IIIF

As part of our transition to IIIF for high resolution photographs for the numismatic collection in MANTIS (see http://numismatics.org/collection/1944.100.45250 for example), I have begun to migrate our archival images into IIIF as well. These new features will be available in our new dedicated server as soon as the migration of Wordpress from one server to another is complete, which I expect in the next few weeks. The implementation of IIIF for our archival resources entails three overhauls of the current metadata model and HTML/IIIF Manifest serialization: TEI (for Newell notebooks of facsimile images), Encoded Archival Description (EAD) finding aids, and MODS. The transformation of the TEI notebooks into IIIF compliance is completed, and the functionality for EAD and MODS has been built, but the XML data have not been fully updated to link to IIIF services (mainly because the high resolution images haven't been uploaded to the server yet).

Annotated Newell notebook IIIF manifest displayed in Mirador

TEI to IIIF Manifest

The first Newell notebook was published to Archer (built on EADitor) more than three years ago. There are now about 50 notebooks published, but only a handful have been annotated to link to people, IGCH hoards, and coins in our collection (we will complete the annotation as part of the Hellenistic Royal Coinages project). To summarize the technical underpinnings, each notebook is a TEI file with facsimile elements for each page. The facsimile contains a link to the image and 0-n surface elements representing annotations. These surface elements were created by roundtripping the Annotorious/OpenLayers annotation JSON <-> TEI. The @ulx, @uly, @lrx, and @lry attributes represent the coordinates of the upper left and lower right hand corners of the annotations, and the coordinates were relative ratios based on OpenLayers bounds.

 For IIIF compliance, I ran the TEI through an XSLT 3 transformation to load the info.json metadata from our IIIF image server to extract the height and width of each image, and then recalculate the coordinates to be more in line with Web Annotation segments. The lower right coordinates are still stored in the TEI, but upon generation of annotation lists for the manifest, the left coordinates are subtracted to the right to correctly establish the annotation height and width.

      <surface lrx="1540" lry="155" ulx="1182" uly="54" xml:id="aho40v9vbhq7">
            <ref target="http://coinhoards.org/id/igch1516">IGCH 1516</ref>

The tei:facsimile to annotation list transformation outputs:


The tei:graphic was replaced with tei:media[@type='IIIFService'], with the @url pointing to the IIIF service URI instead of an image location. XSLT transformations for the manifest, HTML, RDF, and Solr outputs do the rest.

The Javascript has been updated so that clicking on a page under the index of annotations will force Mirador to change the the correct canvas.

You can see an example here: http://numismatics.org/archives/id/nnan187715

I will post another update on EAD and MODS -> IIIF next week. 

Thursday, August 10, 2017

First DOIs minted for ANS Digital Library items

Several weeks ago, we migrated an older, circa 2002 TEI ebook on the Taranto 1911 hoard, authored by John Kroll and Sebastian Heath, into our Digital Library. The original TEI file and subsequent updates have been loaded into our TEI Github repository. The updates follow transcription precedents that we have set in older ANS-published printed monographs as part of the Mellon-funded Open Humanities Book Program: relevant places, objects, people, etc. have been linked to entities in LOD systems, such as Nomisma.org. All of the objects within this hoard (itself linked to IGCH 1864) are in the British Museum and linked to their URIs. Upon publication into the ANS Digital Library, the document parts are now accessible from the IGCH 1864 record and in (eventually) in Pelagios, connected to relevant ancient places.

Since Sebastian is an active scholar, with an ORCID, this document served as a proof of concept for the next iteration of ANS digital publication: that our current and future monographs and journal articles, once issued openly online, should be connected to ORCIDs for their authors, and publication metadata should be submitted to Crossref to mint a DOI and enhance accessibility. Furthermore, since there's a direct connection between ORCID and Crossref submissions, this new digital publication workflow would automatically populate an author's scholarly profile with ANS publications. This is a vast improvement over the likes of Academia.edu, which requires manual submission. The broad vision is this:

Regardless of whether an author submits works through the American Numismatic Society Digital Library, Zenodo.org, Humanities Commons, their own institutional repository, or an Open Access journal system, their ORCID profile is the central, canonical aggregation of the entirety of their intellectual output (which includes datasets, software, etc.).

This aggregation system between DOIs and ORCIDs, following Linked Open Data principles, is the future of academic publication. Ideally, it should be expanded beyond citations to modern works with DOIs and ORCIDs to include more historic works defined by Worldcat and linked to historic scholars with ISNI identifiers. It would take a tremendous amount of work, but in theory, it would be possible to create a network graph of citations across all disciplines, going back in history to the advent of the printed book, charting the evolution of how knowledge is generated and disseminated. Therefore, Crossref, ISNI, and ORCID would perhaps play a greater role than providing simple (and superficial) citation metrics in enabling us to develop a broader historiography and analysis of scholarship itself. We plan to mint DOIs for our historical publications eventually, if Crossref extends its XML schema to support ISNI identifiers.

Under the Hood

Some extensions were implemented in ETDPub, the TEI/MODS publication framework that underlies the ANS Digital library. First, I authored XSLT stylesheets that would crosswalk TEI or MODS into the appropriate Crossref XML model according to their schema version 4.4.0. You can see an example of my MA thesis here: http://numismatics.org/digitallibrary/ark:/53695/gruber_roman_numismatics.xref.

If the author/editor URI matches an ORCID URI in the TEI, then the Admin panel in ETDPub will enable the publication of the metadata to Crossref. Similarly, within the MODS ETD editing interface (in XForms), a user can insert a mods:nameIdentifier[@type='orcid'] under the mods:name for an author/editor in order to capture the ORCID. So far, only TEI or MODS records with ORCIDs attached to people are available for submission into Crossref to mint a DOI.

Submission Workflow

In the admin panel, if a document is eligible for submission to Crossref, a checkbox is available. Clicking on this will fire off a series of actions in the XForms engine:
  1. The TEI/MODS-to-Crossref XML transformation is executed and loaded into an XForms instance
  2. The Crossref XML is serialized to /tmp because it must be attached via multipart/form-data
  3. Still having difficulty getting multipart/form-data to execute correctly in the XForms engine, the XForms engine instead interacts with a PHP script in CGI
  4. After the PHP script responds with a successful HTTP code, the MODS/TEI document is loaded in the XForms engine in order to insert the DOI in the proper location within the document
  5. The TEI/MODS file is saved back to eXist, and the standard publication workflow is executed (a chain of XForms submissions), updating the Solr search index and the triplestore/SPARQL endpoint
So far two documents in the Digital Library have DOIs connected to ORCIDs:

Taranto 1911: http://dx.doi.org/10.26608/taranto1911
My thesis (Recent Advancements in Roman Numismatics): http://dx.doi.org/10.26608/gruber_roman_numismatics

Friday, July 14, 2017

Improved mapping in EADitor - Brett archaeology photos as a test

At long last, I have migrated from OpenLayers to Leaflet in EADitor. This required modifications in two areas: the HTML pages for rendering EAD finding aids and the map interface. As a result, I introduced two new serializations:

  • The map interface renders Solr search results rendering into GeoJSON (instead of OpenLayers displaying Solr->KML as before)
  • A transformation of an EAD finding aid into GeoJSON. A GeoJSON point is created for all unique mappable places from Geonames or Pleiades, and coordinates are extracted in real time by reading Geonames APIs or Pleiades RDF. The GeoJSON features include references to all uniquely addressable components that include that place in the controlaccess element. You can append the extension '.geojson' to get JSON response. Content negotiation will be implemented eventually. See http://numismatics.org/archives/ark:/53695/nnan0037.geojson for example.


 Restructuring the Agnes Baldwin Brett finding aid

Agnes Baldwin Brett was a curator at the ANS from 1909-1912 and a prominent scholar of Greek numismatics. Our archives hold a variety of interesting materials, including photographs from her travels around Greece, Italy, and Turkey in the early 1900s. Numerous photos have been digitized, were uploaded to flickr Commons, and linked to the Brett EAD finding aid. Some photographs were identified and described (with brief text snippets) by ANS archivist, David Hill, but all photographs were placed in a single series-level component. All identifiable places were linked in EADitor's Geonames lookup mechanism in a top-level controlaccess element. There was no direct correlation between individual photographs and the people, places, and things depicted.

In order to demonstrate the full functionality of the new mapping interface, I finally took the time to restructure the finding aid so that each photograph would appear in its own item-level component with a controlaccess element enabling individual identification of the place depicted in the photo. Furthermore, while many finding aids have been linked to modern places defined in Geonames, the Brett collection of archaeological photographs provided an opportunity to link photos to ancient places in Pleiades, which would, in turn, open the door to the integration of these valuable materials into the wider Linked Ancient World Data cloud via Pelagios. The photos feature Mycenaean tombs, Greek temples, and even the Grave Stele of Hegeso.

Identifying individual monuments within Athens

Not only that, some photographs feature other students from the American School of Classical Studies at Athens that went on to be prominent scholars later in life. Since many of these scholars have produced published works and archival materials held at other institutions, they have URIs in the Social Network and Archival Context project. EADitor has had SNAC lookups for quite some time, and so I was able to link photos to these URIs when applicable. I hope that we can make these photos available to researchers even beyond the ancient world.

Linking people to SNAC
In addition to the tagging of places and people, many photographs feature known archaeological monuments that are notable enough to warrant their own Wikipedia articles, and therefore Wikidata entity URIs. I extended the subject lookup mechanism in EADitor beyond the standard Library of Congress Subject Headings to query the Wikidata API, embedding entity IDs directly into the EAD finding aid, which are then transformed into dcterms:subject URIs upon RDF serialization.



Since each individual component has an ID in EADitor, each component is uniquely addressable by fragment identifiers, e.g., http://numismatics.org/archives/ark:/53695/nnan0037#d1e131. After making some minor modifications to the RDF output to conform with the emerging schema.org archival extension, These Wikidata, SNAC, Pleiades, and Geonames URIs are exposed in the RDF for each component, which are hierarchically linked together.

@prefix arch: <http://purl.org/archival/vocab/arch#> .
@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix schema: <http://schema.org/> .
@prefix xml: <http://www.w3.org/XML/1998/namespace> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .

<http://numismatics.org/archives/ark:/53695/nnan0037#d1e131> a schema:ArchiveItem ;
    dcterms:coverage <http://www.geonames.org/264371> ;
    dcterms:date "1900-12-07"^^xsd:date ;
    dcterms:identifier "06-00242" ;
    dcterms:isPartOf <http://numismatics.org/archives/ark:/53695/nnan0037#c_92f631e3f903281a8cdedbfebfca0654> ;
    dcterms:subject <http://socialarchive.iath.virginia.edu/ark:/99166/w61c5qjp> ;
    dcterms:title "American School students wearing bug bags" ;
    dcterms:type <http://vocab.getty.edu/aat/300046300> ;
    foaf:depiction <http://farm9.staticflickr.com/8320/8003385533_c83827b679_o.jpg> ;
    foaf:thumbnail <http://farm9.staticflickr.com/8320/8003385533_55f1f093b1_t.jpg> .

This RDF is posted into Archer's SPARQL endpoint.

Archer RDF → SPARQL → Pelagios RDF

Now that we have numerous uniquely addressable photographs linked to Pleiades URIs published in our SPARQL endpoint, it was a breeze to create an RDF export for Pelagios. It is essentially a DESCRIBE query, and our model of RDF is run through XSLT into the Pelagios data model.

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX dcterms: <http://purl.org/dc/terms/>
 ?s dcterms:coverage ?place FILTER (strStarts(str(?place), 'https://pleiades.stoa.org'))  

The link to the Pelagios VoID is available on the front page of Archer. It is generated by an ASK query similar to above to see whether there are any objects in the SPARQL endpoint with Pleiades places expressed by the dcterms:coverage property.


The Brett collection is incredibly interesting, and I hope that we will be able to digitize more photographs and the corresponding travel diary at some point in the future. There are still many photographs that haven't been identified, and so perhaps we might be able to accomplish this through crowdsourcing. We will implement a IIIF server by the end of summer and begin the transition of our archival materials into IIIF--not only photographs, but also the Newell diaries. Perhaps one day we will be able to annotate the people, places, and things from the Brett diary and photographs with Mirador or a similar IIIF viewer. While Pelagios integration is somewhat imminent, the aggregation of disparate archival holdings through shared SNAC identifiers is still further along the horizon.

Tuesday, February 28, 2017

Final four Mellon-funded TEI ebooks published

The final four of a group of 86 American Numismatic Society-published books have been checked and uploaded to our Digital Library. Here are some stats I was able to produce from various SPARQL queries of the TEI->Open Annotation RDF:

  • 349 mentions of 164 different Greek coin hoards published in IGCH in 193 sections in 14 books.
  • 266 unique references to nomisma URIs. 146 are mints or regions, and 87 of these identifiers are matches with Pleiades places. These mint references appear in 600 sections 51 books. Including direct Pleiades references (and not only those which are implicit by means of Nomisma concordances), there are 621 sections in these 51 books which will be accessible through the Pelagios Project.
  • 97 of the 266 references are to people, most of whom are linked to Wikidata and VIAF entities that are, in turn, linked to other systems, such as Social Networks and Archival Context
  • More than 1,400 coins in the ANS collection are referenced
  • 139 Roman Imperial coin types in OCRE
  • 4 Roman Republican coin types in CRRO 
These four are the final of 86 total books digitized as part of the NEH-Mellon Open Humanities Book program.  Many thanks to both the National Endowment for the Humanities and the Mellon Foundation for making this possible. The framework and methodologies implemented in this project will be applied to further digitization here at the ANS as we move toward making our entire collection of monographs freely and openly accessible, and I hope that other academic publishers and learned societies will follow in our footsteps in this endeavor.

These books go beyond simple transcription and publication as EPUB files. With links to our own research databases internally and externally to Linked Open Data information systems, we hope that these works will be transformed into research portals to further context about the people, places, events, etc. mentioned in the text. On the other side of the coin, so to speak, researchers interested about the entities, objects, coin hoards, etc. will have access to a wealth of historical information about these things and will gain access to our monographs not only from our own Library, Archive, and Museum systems, but through projects like Pelagios, Digital Public Library of America, and other large scale aggregators of cultural heritage materials.

Friday, January 13, 2017

More than 80 LOD-enhanced ebooks published to the ANS Digital Library

The American Numismatic Society has nearly completed its Mellon Foundation-funded Humanities Open Book program. Eighty-two of 86 books have been enhanced by a Whitney Christopher, a TEI specialist from the King's College London DH program to link to people and places defined on Nomisma.org, Pleiades (either directly linked or by means of Nomisma's internal concordance system), VIAF, Wikidata, and the ANS's own archival authority control system. The final four books will go online soon. They are all available in the ANS Digital Library.

The number of people and places mentioned in these texts is a staggering figure, and it should be noted that we have focused on linking those entities that are most relevant to the texts, but we will continue to refine the linking over time, especially when it comes to Nomisma concepts and bibliographic references to Worldcat Works (links to which have not yet been incorporated). As Nomisma expands further into the Greek world and other domains of numismatics (after the ancient period), we will return to these ebooks to insert or replace links to Nomisma mints, people, and political entities.

Beyond relevant people and places, we have inserted hundreds of links to IGCH records (about 170 different coin hoards are cited in 400 locations in a handful of books), to the ANS collection, and to coin types defined in OCRE or CRRO. So far, more than 100 coins in the ANS and 6 in the Smithsonian American Art Museum have been identified by their accession numbers, although one of the four remaining books to be published will soon include nearly 70 more links to ANS coins. There are many more coins referenced in these books that may now belong to the ANS, but were not accessioned at the date of publication. A curator with more specific knowledge will need to identify these in the future.

One of the most often cited hoard is the Demanhur Hoard (IGCH 1664), which is mentioned in four books and on various pages of two of Edward Newell's notebooks. By linking archival authorities mentioned in these texts, we have greatly enhanced access to the works by and about Edward Newell and other prominent numismatic figures associated with the Society. A user of the ANS's authority portal (built on EAC-CPF) will have access to books written by Newell in our digital library, as well as his archival materials. Furthermore, mentions of Newell from the books written by other scholars will appear under annotations. In his case, he is mentioned in 18 other books, sometimes in multiple sections.

Like Mantis, the OCRE and CRRO config files have been updated to link to our archival SPARQL endpoint, and therefore annotations about specific types are accessible directly through types defined in these system. Nearly 50 types in OCRE are linked from Roman Medallions, and a researcher can drill down into a specific section of the book from RIC 5 Gallienus and Salonina 1.

Finally, through the links to Pleiades, each section in each book that mentions an ancient place will be accessible in Pelagios.