As 2016 stumbles to an end, I’ve put in a few days’ work on my new project Oceania, which is to be a Linked Data service for cultural heritage in this part of the world. Part of this project involves harvesting data from cultural institutions which make their collections available via so-called “Web APIs”. There are some very standard ways to publish data, such as OAI-PMH, OpenSearch, SRU, RSS, etc, but many cultural heritage institutions instead offer custom-built APIs that work in their own peculiar way, which means that you need to put in a certain amount of effort in learning each API and dealing with its specific requirements. So I’ve turned to the problem of how to deal with these APIs in the most generic way possible, and written a program that can handle a lot of what is common in most Web APIs, and can be easily configured to understand the specifics of particular APIs.

This program, which I’ve called API Harvester, can be configured by giving it a few simple instructions: where to download the data from, how to split up the downloaded data into individual records, where to save the record files, how to name those files, and where to get the next batch of data from (i.e. how to resume the harvest). The API Harvester does have one hard requirement: it is only able to harvest data in XML format, but most of the APIs I’ve seen offered by cultural heritage institutions do provide XML, so I think it’s not a big limitation.

The API Harvester software is open source, and free to use; I hope that other people find it useful, and I’m happy to accept feedback or improvements, or examples of how to use it with specific APIs. I’ve created a wiki page to record example commands for harvesting from a variety of APIs, including OAI-PMH, the Trove API, and an RSS feed from this blog. This wiki page is currently open for editing, so if you use the API Harvester, I encourage you to record the command you use, so other people can benefit from your work. If you have trouble with it, or need a hand, feel free to raise an issue on the GitHub repository, leave a comment here, or contact me on Twitter.

Finally, a brief word on how to use the software: to tell the harvester how to pull a response apart into individual records, and where to download the next page of records from (and the next, and the next…), you give it instructions in the form of “XPath expressions”. XPath is a micro-language for querying XML documents; it allows you to refer to elements and attributes and pieces of text within an XML document, to perform basic arithmetic and manipulate strings of text. XPath is simple yet enormously powerful; if you are planning on doing anything with XML it’s an essential thing to learn, even if only to a very basic level. I’m not going to give a tutorial on XPath here (there are plenty on the web), but I’ll give an example of querying the Trove API, and briefly explain the XPath expressions used in that examples:

Here’s the command I would use to harvest metadata about maps, relevant to the word “oceania”, from the Trove API, and save the results in a new folder called “oceania-maps” in my Downloads folder:

java -jar apiharvester.jar
directory="/home/ctuohy/Downloads/oceania-maps"
retries=5
url="http://api.trove.nla.gov.au/result?q=oceania&zone=map&reclevel=full"
url-suffix="&key=XXXXXXX"
records-xpath="/response/zone/records/*"
id-xpath="@url"
resumption-xpath="/response/zone/records/@next"

For legibility, I’ve split the command onto multiple lines, but this is a single command and should be entered on a single line.

Going through the parts of the command in order:

• The command java launches a Java Virtual Machine to run the harvester application (which is written in the Java language).
• The next item, -jar, tells Java to run a program that’s been packaged as a “Java Archive” (jar) file.
• The next item, apiharvester.jar, is the harvester program itself, packaged as a jar file.

The remainder of the command consists of parameters that are passed to the API harvester and control its behaviour.

• The first parameter, directory="/home/ctuohy/Downloads/oceania-maps", tells the harvester where to save the XML files; it will create this folder if it doesn’t already exist.
• With the second parameter, retries=5, I’m telling the harvester to retry a download up to 5 times if it fails; Trove’s server can sometimes be a bit flaky at busy times; retrying a few times can save the day.
• The third parameter, url="http://api.trove.nla.gov.au/result?q=oceania&zone=map&reclevel=full", tells the harvester where to download the first batch of data from. To generate a URL like this, I recommend using Tim Sherratt’s excellent online tool, the Trove API Console.
• The next parameter url-suffix="&key=XXXXXXX" specifies a suffix that the harvester will append to the end of all the URLs which it requests. Here, I’ve used url-suffix to specify Trove’s “API Key”; a password which each registered Trove API user is given. To get one of these, see the Trove Help Centre. NB XXXXXXX is not my actual API Key.

The remaining parameters are all XPath expressions. To understand them, it will be helpful to look at the XML content which the Trove API returns in response to that query, and which these XPath expressions apply to.

• The first XPath parameter, records-xpath="/response/zone/records/*", identifies the elements in the XML which constitute the individual records. The XPath /response/zone/records/* describes a path down the hierarchical structure of the XML: the initial / refers to the start of the document, the response refers to an element with that name at the “root” of the document, then /zone refers to any element called zone within that response element, then /records refers to any records within any of those response elements, and the final /* refers to any elements (with any name) within any of of those response elements. In practice, this XPath expression identifies all the work elements in the API’s response, and means that each of these work elements (and its contents) ends up saved in its own file.
• The next parameter, id-xpath="@url" tells the harvester where to find a unique identifier for the record, to generate a unique file name. This XPath is evaluated relative to the elements identified by the records-xpath; i.e. it gets evaluated once for each record, starting from the record’s work element. The expression @url means “the value of the attribute named url”; the result is that the harvested records are saved in files whose names are derived from these URLs. If you look at the XML, you’ll see I could equally have used the expression @id instead of @url.
• The final parameter, resumption-xpath="/response/zone/records/@next", tells the harvester where to find a URL (or URLs) from which it can resume the harvest, after saving the records from the first response. You’ll see in the Trove API response that the records element has an attribute called next which contains a URL for this purpose. When the harvester evaluates this XPath expression, it gathers up the next URLs and repeats the whole download process again for each one. Eventually, the API will respond with a records element which doesn’t have a next attribute (meaning that there are no more records). At that point, the XPath expression will evaluate to nothing, and the harvester will run out of URLs to harvest, and grind to a halt.

Happy New Year to all my readers! I hope this tool is of use to some of you, and I wish you a productive year of metadata harvesting in 2017!

The first post gave an introduction and overview of the architecture of the publication software, and the second dealt quite specifically with how names and identifiers work in the LOD publication software.

In this post I’ll cover how the publication software takes the data published by Museum Victoria’s API and reshapes it to fit a common conceptual model for museum data, the “Conceptual Reference Model” published by the documentation committee of the Internal Council of Museums. I’m not going to exhaustively describe the translation process (you can read the source code if you want the full story), but I’ll include examples to illustrate the typical issues that arise in such a translation.

The CIDOC Conceptual Reference Model

The CIDOC CRM, as it’s usually called, is a system of concepts for analysing and describing the content of museum collections. It is not intended to be a replacement for the Collection Management Systems which museums use to store their data; it is rather intended to function as a kind of lingua franca, through which content from a variety of systems can be expressed in a generally intelligible way.

The Conceptual Reference Model covers a wide range of museological concerns: items can be described in terms of their materials and mode of construction, as well as by who made them, where and when, and for what purpose.

The CRM also provides a framework to describe the events in which objects are broken into pieces, or joined to other objects, damaged or repaired, created or utterly destroyed. Objects can be described in terms of the symbolic and intellectual content which they embody, which are themselves treated as “intellectual objects”. The lineage of intellectual influence can be described, either speculatively, in a high-level way, or by explicitly tracing and documenting the influences that were known have taken place at particular times and locations. The legal history of objects can also be traced through transfer of ownership and custody, commission, sale and purchase, theft and looting, loss and discovery. Where the people involved in these histories are known, they too can be named and described and their histories interwoven with those of other people, objects, and ideas.

Core concepts and additional classification schemes

The CRM framework is quite high level. Only a fairly small number of very general types of thing are defined in the CRM; only concepts general enough to be useful for any kind of museum; whether a museum of computer games or of classical antiquity. Each of these concepts is identified by an alphanumeric code and an English-language name. In addition, the CRM framework allows for arbitrary typologies to be added on, to be used for further classifying pretty much anything. This is to allow all the terms from any classification system used in a museum to be exported directly into a CRM-based dataset, simply by describing each term as an “E55 Type". In short, the CRM consists of a fairly fixed common core, supplemented by a potentially infinite number of custom vocabularies which can be used to make fine distinctions of whatever kind are needed.

Therefore, a dataset based on the CRM will generally be directly comparable with another dataset only in terms of the core CRM-defined entities. The different classification schemes used by different datasets remain “local” vocabularies. To achieve full interoperability between datasets, these distinct typologies would additionally need to be aligned, by defining a “mapping” table which lists the equivalences or inequivalences between the terms in the two vocabularies. For instance, such a table might say that the term “moulded” used in Museum Victoria’s collection is more or less the same classification as “molding (forming)” in the Getty Art and Architecture thesaurus.

Change happens through “events”

To model how things change through time, the CRM uses the notion of an “event”. The production of a physical object, for instance, is modelled as an E12 Production event (NB concepts in the CRM are all identified by an alphanumeric code). This production event is linked to the object which it produced, as well as to the person or persons who played particular creative roles in that event. The event may also have a date and place associated with it, and may be linked to the materials and to the method used in the event.

On a somewhat philosophical note, this focus on discrete events is justified by the fact that not all of history is continuously documented, and we necessarily have a fragmentary knowledge of the history of any given object. Often a museum object will have passed through many hands, or will have been modified many times, and not all of this history is known in any detail. If we know that person A created an object for person B, and that X years later the object turned up in the hands of person C, we can’t assume that the object remained in person B’s hands all those X years. A data model which treated “ownership” as a property of an object would be liable to making such inflated claims to knowledge which is simply not there. Person C may have acquired it at any point during that period, and indeed there may have been many owners in between person B and person C. This is why it makes sense to document an object’s history in terms of the particular events which are known and attested to.

Museum Victoria’s API

How does Museum Victoria’s data fit in terms of the CIDOC model?

In general the model works pretty well for Museum Victoria, though there are also things in MV’s data which are not so easy to express in the CRM.

Items

Museum Victoria describes items as “Things made and used by people”. These correspond exactly to the notion of E22 Man-Made Object in the CIDOC CRM (if you can excuse the sexist language), described as comprising “physical objects purposely created by human activity.”

Every MV item is therefore expressed as an E22 Man-Made Object.

Titles

Museum Victoria’s objects have an objectName property which is a simple piece of text; a name or title. In the CIDOC CRM, the name of an object is something more complex; it’s an entity in its own right, called an E41 Appellation. The reason why a name is treated as more than just a simple property of an object is that in the CRM, it must be possible to treat an object’s name as an historical phenomenon; after all, it will have been applied to an object by a particular person (the person who created the object, perhaps, or an archaeologist who dug it out of the ground, or a curator or historian), at some historical point in time. An object may have a number of different names, each given it by different people, and used by different people at different times.

However, because the Museum Victoria names are simple (a single label) we can ignore most of that complexity. We only need to define an E41 Appellation whose value is the name, and link the E41 Appellation to the E22 Man-Made Object using a P1 is identified by association.

Articles, Items and their relationships

The MV API provides access to a number of “articles” which are documents related to the Museum’s collection. For each article, the API shows a list of the related collection items; and for each item, you can get the corresponding list of related articles. Although the exact nature of the relationship isn’t made explicit, it’s reasonable to assume that an item is in some way documented by the articles that are related to it. In the CIDOC CRM, such an article is considered an E31 Document, and it bears a P70 documents relationship to the item which it’s about.

If the relationship between an item and article is easy to guess, there are a couple of other relationships which are a little less obvious: an article also has a list of related articles, and each item also has a list of related items. What is that nature of those relationships? In what way exactly does article X relate to article Y, or item A to item B? The MV API’s documentation doesn’t say, and it wouldn’t surprise me if the Museum’s collection management system leaves this question up to the curators’ judgement.

A bit of empirical research seemed called for. I checked a few of the related items and the examples I found seemed to fall into two categories:

• One item is a photograph depicting another item (the specific relationship here is really “depicts”)
• Two items are both photographs of the same subject (the relationship is “has the same subject as”).

Obviously there are two different kinds of relationship here in the Museum’s collection, both of them presented (through the API) in the same way. As a human, I can tell them apart, but my proxy software is not going to be able to. So I need to find a more general relationship which subsumes both the relationships above, and fortunately, the CIDOC CRM includes such a relationship, namely P130 shows features of.

This property generalises the notions of “copy of” and “similar to” into a dynamic, asymmetric relationship, where the domain expresses the derivative, if such a direction can be established.
Otherwise, the relationship is symmetric. It is a shortcut of P15 was influenced by (influenced) in a creation or production, if such a reason for the similarity can be verified. Moreover it expresses similarity in cases that can be stated between two objects only, without historical knowledge about its reasons.

For example, I have a photograph of a piece of computer hardware (which is the relatedItem), and the photo is therefore a kind of derivative of the hardware (though the Museum Victoria API doesn’t tell me which of the objects was the original and which the derivative). In another example I have two photos of the same house; here there’s a similarity which is not due to one of the photos being derived from the other.

Ideally, it would be preferable to be able to represent these kinds of relationships more precisely; for instance, in the case of the two photos of the house, one could generate a resource that denotes the actual physical house itself, and link that to the photographs, but because the underlying data doesn’t include this information in a machine-readable form, the best we can do is to say that the two photos are similar.

Production techniques

Some of the items in the Museum’s collection are recorded as having been produced using a certain “technique”. For instance, archaeological artefacts in the MV collection have a property called archeologyTechnique, which contains the name of a general technique, such as moulded, in the case of certain ceramic items.

This corresponds to the CRM concept P32 used general technique, which is described like so:

This property identifies the technique or method that was employed in an activity.
These techniques should be drawn from an external E55 Type hierarchy of consistent terminology of
general techniques or methods such as embroidery, oil-painting, carbon dating, etc.

Note that in CIDOC this “general technique” used to manufacture an object is not a property of the object iself; it’s a property of the activity which produced the object (i.e. the whole process in which the potter pressed clay into a mould, glazed the cup, and fired it in a kiln).

Note also that, for the CIDOC CRM, the production technique used in making these tea-cups is not the text string “moulded”; it is actually an abstract concept identified by a URI. The string “moulded” is just a human-readable name attached as a property of that concept. That same concept might very well have a number of other names in other languages, or even in English there’s the American variant spelling “molded”, and synonyms such as “cast” that could all be alternative names for the same concept.

Translating a Museum Victoria item with a technique into the CRM therefore involves identifying three entities:

• the object itself (an E22 Man-Made Object);
• the production of the object (an E12 Production activity);
• the technique used in the course of that activity to produce the object (an E55 Type of technique)

These three entities are then linked together:

• The production event “P32 used general technique" of the technique; and
• The production event [edit: "P94 has created"] "P108 has produced" the object itself.

Notes

The items, articles, specimens and species in the Museum’s API are all already first-class objects and can be easily represented as concepts in Linked Data. The archeologyTechnique field also has a fairly restricted range of values, and each of those values (such as “moulded”) can be represented as a Linked Data concept as well. But there are a number of other fields in the Museum’s API which are in the form of relatively long pieces of descriptive text. For example, an object’s objectSummary field contains a long piece of text which describes the object in context. For example, here’s the objectSummary of one our moulded tea cups:

This reconstructed cup was excavated at the Commonwealth Block site between 1988 and 2003. There is a matching saucer that was found with it. The pattern is known as 'Moss Rose' and was made between 1850 and 1851 by Charles Meigh, Son & Pankhurst in Hanley, Staffordshire, England.


Homewares.
Numerous crockery pieces were found all over the Little Lon site. Crockery gives us a glimpse of everyday life in Melbourne in the 1880s. In the houses around Little Lon, residents used decorated crockery. Most pieces were cheap earthenware or stoneware, yet provided colour and cheer. Only a few could afford to buy matching sets, and most china was probably acquired second-hand. Some were once expensive pieces. Householders mixed and matched their crockery from the great range of mass-produced designs available. 'Blue and white' and the 'willow' pattern, was the most popular choice and was produced by English potteries from 1790.

It’s not quite as long as an “article” but it’s not far off it. Another textual property is called physicalDescription, and has a narrower focus on the physical nature of the item:

This is a glazed earthenware teacup which has been reconstructed. It is decorated with a blue or black vine and leaf design around outside and inside of the cup which is known as 'Moss Rose' pattern.


The CIDOC CRM does include concepts related to the historical context and the physical nature of items, but it’s not at all easy to extract that detailed information from the descriptive prose of these, and similar fields. Because the information is stored in a long narrative form, it can’t be easily mapped to the denser data structure of a Linked Data graph. The best we can hope to do with these fields is to treat them as notes attached to the item.

The CIDOC CRM includes a concept for attaching a note: P3 has note. But to represent these two different types of note, it’s necessary to extend the CRM by creating two new, specialized versions (“sub-properties”) of the property called P3 has note, which I’ve called P3.1 objectSummary and P3.1 physicalDescription.

Summary

It’s possible to recognise three distinct patterns in mapping an API such as Museum Victoria’s to a Linked Data model like the CIDOC CRM.

1. Where the API provides access to a set of complex data objects of a particular type, these can
   be mapped straight-forwardly to a corresponding class of Linked Data resources (e.g. the items, species, specimens, and articles in MV’s API).
2. Where the API exposes a simple data property, it can be straightforwardly converted to a Linked Data property (e.g. the two types of notes, in the example above).
3. Where the API exposes a simple data property whose values come from a fairly limited range (a “vocabulary”), then those individual property values can be assigned identifiers of their own, and effectively promoted from simple data properties to full-blown object properties (e.g. the production techniques in Museum Victoria’s API).

Conclusion

It’s been an interesting experiment, to generate Linked Open Data from an open API using a simple proxy: I think it shows that the technique is a very viable mechanism for institutions to break into the LOD cloud and contribute their collection in a standardised manner, without necessarily having to make any changes to their existing systems or invest in substantial software development work. To my mind, making that first step is a significant barrier that holds institutions and individuals back from realising the potential in their data. Once you have a system for publishing LOD, you are opening up a world of possibilities for external developers, data aggregators, and humanities researchers, and if your data is of interest to those external groups, you have the possibility of generating some significant returns on your investment, and the possibility of “harvesting” some of that work back into your institution’s own web presence in the form of better visualizations, discovery interfaces, and better understanding of your own collection.

Before the end of the year I hope to explore some further possibilities in the area of user interfaces based on Linked Data, to show some of the value that these Linked Data publishing systems can support.

I’m writing this up as an example of one way — a relatively easy way — to publish Linked Data off the back of some existing API. I hope that some other libraries, archives, and museums with their own API will adopt this approach and start publishing their data in a standard Linked Data style, so it can be linked up with the wider web of data.

Two ways to skin a cat

There are two basic ways you can take an existing API and turn it into a Linked Data service.

One way is the “metadata aggregator” approach. In this approach, an “aggregator” periodically downloads (or “harvests”) the data from the API, in bulk, converts it all into RDF, and stores it in a triple-store. Then another application — a Linked Data publishing service — can read that aggregated data from the triple store using a SPARQL query and expose it in Linked Data form. The tricky part here is that you have to create and maintain your own copy (a cache) of all the data which the API provides. You run the risk that if the source data changes, then your cache is out of date. You need to schedule regular harvests to be sure that the copy you have is as up to date as you need it to be. You have to hope that the API can tell you which particular records have changed or been deleted, otherwise, you may have to download every piece of data just to be sure.

But this blog post is about another way which is much simpler: the “proxy” approach. A Linked Data proxy is a web application that receives a request for Linked Data, and in order to satisfy that request, makes one or more requests of its own to the API, processing the results it receives, and formatting them as Linked Data, which it then returns. The advantage of this approach is that every response to a request for Linked Data is freshly prepared. There is no need to maintain a cache of the data. There is no need for harvesting or scheduling. It’s simply a translator that sits in front of the API and translates what it says into Linked Data.

This is an approach I’ve been meaning to try out for a fair while, and in fact I gave a very brief presentation on the proxy idea at the recent LODLAM Summit in Sydney. All I needed was a good API to try it out with.

Museum Victoria Collection API

The Museum Victoria Collection API was announced on Twitter by Ely Wallis on August 25th:

The cat's out of the bag – @museumvictoria has a new collections online. We hope you like it! http://t.co/c0oNTWDk5t

— Ely Wallis (@elyw) August 25, 2015

Well as it happened I did like it, so I got in touch. Since it’s so new, the API’s documentation is a bit sparse, but I did get some helpful advice from the author of the API, Museum Victoria’s own Michael Mason, including details of how to perform searches, and useful hints about the data structures which the API provides.

In a nutshell, the Museum Victoria API provides access to data about four different sorts of things:

• Items (artefacts in the Museum’s collections),
• Specimens (biological specimens),
• Species (which the specimens belong to), and
• Articles (which document the other things)

There’s also a search API with which you can search within all of those categories.

Armed with this knowledge, I used my trusty XProc-Z proxy software to build a Linked Data proxy to that API.

Linked Data

Linked Data is a technique for publishing information on the web in a common, machine-understandable way.

The central principle of Linked Data is that all items of interest are identified with an HTTP URI (“Uniform Resource Identifier”). And “Resource” here doesn’t just mean web pages or other electronic resources; anything at all can be a “Resource”: people, physical objects, species of animal, days of the week … anything. If you take one of these URIs and put it into your browser, it will deliver you up some information which relates to the thing identified by the URI.

Because of course you can’t download a person or a species of animal, there is a special trick to this: if you send a request for a URI which identifies one of these “non-information resources”, such as a person, the server can’t simply respond by sending you an information resource (after all, you asked for a person, not a document). Instead it responds by saying “see also” and referring you to a different URL. This is basically saying “since you can’t download the resource you asked for (because it’s not an information resource), here is the URL of an information resource which is relevant to your request”. Then when your browser makes a request from that second URL, it receives an information resource. This is why, when browsing Linked Data, you will sometimes see the URI in your browser’s address bar change: first it makes a request for one URI and then is automatically redirected to another.

That information also needs to be encoded as RDF (“Resource Description Framework”). The RDF document you receive from a Linked Data server consists of a set of statements (called “triples”) about various things,including the original “resource” which your original URI identified, but usually also other things as well. Those statements assign various properties to the resources, and also link them to other resources. Since those other resources are also identified by URIs, you can follow those links and retrieve information about those related resources, and resources that are related to them, and so on.

Linked Data URIs as proxies for Museum Victoria identifiers

So one of the main tasks of the Linked Data proxy is to take any identifiers which it retrieves from the Museum Victoria API, and convert them into full HTTP URIs. That’s pretty easy; it’s just a matter of adding a prefix like “http://example/something/something/”. When the proxy receives a request for one of those URIs, it has to be able to turn it back into the form that Museum Victoria’s API uses. That basically involves trimming the prefix back off. Because many of the things identified in the Museum’s API are not information resources (many are physical objects), the proxy makes up two different URIs, one to denote the thing itself, and one to refer to the information about the thing.

The conceptual model (“ontology”)

The job of the proxy is to publish the data in a standard Linked Data vocabulary. There was an obvious choice here; the well-known museum ontology (and ISO standard) with the endearing name “CIDOC-CRM”. This is the Conceptual Reference Model produced by the International Committee for Documentation (CIDOC) of the International Council of Museums. This abstract model is published as an OWL ontology (a form that can be directly used in a Linked Data system) by a joint working group of computer scientists and museologists in Germany.

This Conceptual Reference Model defines terms for things such as physical objects, names, types, documents, and images, and also for relationships such as “being documented in”, or “having a type”, or “being an image of”. The proxy’s job is to translate the terminology used in Museum Victoria’s API into the terms defined in the CIDOC-CRM. Unsurprisingly, much of that translation is pretty easy, because there are long-standing principles in the museum world about how to organise collection information, and both the Museum Victoria API and the CIDOC-CRM are aligned to those principles.

As it happened I already knew the CIDOC-CRM model pretty well, which was one reason why a museum API was an attractive subject for this exercise.

Progress and prospects

At this stage I haven’t yet translated all the information which the Museum’s API provides; most of the details are still simply ignored. But already the translation does include titles and types, as well as descriptions, and many of the important relationships between resources (it wouldn’t be Linked Data without links!). I still want to flesh out the translation some more, to include more of the detailed information which the Museum’s API makes available.

This exercise was a test of my XProc-Z software, and of the general approach of using a proxy to publish Linked Data. Although the result is not yet a complete representation of the Museum’s API, I think it has at least proved the practicality of the approach.

At present my Linked Data service produces RDF in XML format only. There are many other ways that the RDF can be expressed, such as e.g. JSON-LD, and there are even ways to embed the RDF in HTML, which makes it easier for a human to read. But I’ve left that part of the project for now; it’s a very distinct part that will plug in quite easily, and in the meantime there are other pieces of software available that can do that part of the job.

See the demo

The proxy software itself is running here on my website conaltuohy.com, but for ease of viewing it’s more convenient to access it through another proxy which converts the Linked Data into an HTML view.

Here is an HTML view of a Linked Data resource which is a timber cabinet for storing computer software for an ancient computer: http://conaltuohy.com/xproc-z/museum-victoria/resource/items/1411018. Here is the same Linked Data resource description as raw RDF/XML: http://conaltuohy.com/xproc-z/museum-victoria/data/items/1411018 — note how, if you follow this link, the URL in your browser’s address bar changes as the Linked Data server redirects you from the identifier for the cabinet itself, to an identifier for a set of data about the cabinet.

The code

The source code for the proxy is available in the XProc-Z GitHub repository: I’ve packaged the Museum Victoria Linked Data service as one of XProc-Z’s example apps. The code is contained in two files:

• museum-victoria.xpl which is a pipeline written in the XProc language, which deals with receiving and sending HTTP messages, and converting JSON into XML, and
• museum-victoria-json-to-rdf.xsl, which is a stylesheet written in the XSLT language, which performs the translation between Museum Victoria’s vocabulary and the CIDOC-CRM vocabulary.

One especially nice feature of Zotero is that you can use it to collaborate with a group of people on a shared library of data which is stored in the cloud and synchronized to the devices of the group members.

Getting data out of Zotero’s web API

If you then want to get the data out of Zotero to do other things with it, you have a number of options. Zotero supports many standard export formats, but the problem I found was that none of those export formats exposed the full richness of your data. Some formats don’t include the “Tags” that you can apply to items in your library; some don’t reflect the hierarchical structure of ‘Collections’ in your library; and so on. It seems the only way to get the full story is to use Zotero’s web API.

Like any web API, this API is a great thing; it makes it possible to use Zotero as a platform for building all kinds of other web applications and systems. The nice thing about a web API is that it’s open to being accessed by any other kind of software. You don’t need to write your software in Javascript or in PHP (the Zotero data server is written in PHP). To access a web API you only need to be able to make HTTP requests, so you’re not tied to any particular platform.

Zotero’s web API is pretty good as web APIs go, though it does have a weakness which is common to many “web APIs”. The weakness is that it’s not obvious how to interpret the data which Zotero provides, and this is a practical barrier to the use of the API. It certainly was for me.

REST

Zotero’s API documentation makes mention of the buzzword “REST”, which is an acronym for “Representational State Transfer”. REST is the name for a style of network communications, defined by a set of design principles or guidelines. A network protocol or web API that conforms to those guidelines is said to be “RESTful”. However, in practice a great many “RESTful” web APIs fail to conform to one or more of the principles, commonly the principle of the “Uniform Interface”, one corollary of which is that the packets of information sent back and forth must be “self-descriptive”.

Self-descriptive messages

To get it right, a RESTful web API needs to provide self-descriptive information; the information it sends you must describe itself sufficiently that you could work out what to do with it. Often the publishers of APIs rely on providing documentation of the different data formats their API provides, and they expect you to have found and read that documentation before you use their API, and to already know what kind of response you will get from the various different parts of their API. But if an API relies on you already knowing what kind of information it provides, then it’s not RESTful. This unfortunately is the case with Zotero.

So how should an API publish “self-descriptive” data?

The HTTP Content-Type header

The main mechanism a web server uses to publish self-descriptive data is to include along with the data a Content-Type header which explicitly declares the format of that information using a code called an “Internet Media Type”. There are a zillion of these Internet Media Types, including image/jpeg and image/png for images, text/html for web pages, application/xml for generic XML documents, or application/json for generic JSON data objects. Of these examples, the last two stand out as different because they are not very specific. What does an XML document mean? What does a JSON object mean? They could mean anything at all, because XML and JSON are generic data formats which can be used to transmit all different kinds of information. It’s possible to be more specific about what kind of XML or JSON you are producing, by saying for instance application/rdf+xml (RDF data encoded in XML) or application/ld+json (Linked Data encoded in JSON). But if you only give a more generic Content-Type, then a client will need to look inside the data package itself to determine what it means.

If Zotero were to publish its data as application/zotero+json, that would be an improvement. It would mean that Zotero data in this format could be exchanged around in other systems, and still be understandable. As it stands, Zotero’s application/json data can only reliably be understood if you have just read it from Zotero.

Here’s an example of the JSON data you can read from Zotero’s API: https://api.zotero.org/groups/300568/items?v=3&format=json

Namespaces

One of the nice features of XML is the concept of “namespaces”. These are distinct vocabularies with globally unique names, which allow you to unambiguously identify what kind of XML data you are looking at. If a piece of software can recognise the namespace or namespaces that a document uses, then it’s in a position to understand what it means, and to process it usefully. Otherwise a human is going to have to look at the XML and try to make some sense of it. JSON doesn’t have an equivalent to XML Namespaces (although JSON-LD does), so that means that information served up as application/json can’t be considered very self-descriptive.

Another interesting point about XML Namespaces is that each of these vocabularies is uniquely identified by a URI; that is, the URI is the name of the vocabulary. This has the nice feature that you can open an XML file, find the namespace URI, plug that namespace URI into your browser, and magically be presented with some useful information about that vocabulary. In other words, any data in this format will always contain a hyperlink to its own documentation (called a “Namespace Document“).

If Zotero were to publish its data in XML, and use a “Zotero” namespace to label all the terms in its vocabulary, then that would be another improvement. Any XML documents of that type could be downloaded from Zotero and stored in any other kind of system, and because they would contain that identifier, they would still be identifiably Zotero-flavoured, even after they had long been detached from Zotero itself.

Formalised data formats

Although it is a problem that Zotero’s JSON data format doesn’t have its own formal name by which it can identify itself, the more critical issue for me in attempting to understand the Zotero data was that the data format exposed by the API is barely documented at all.

If you read the JSON data, you will see names such as publicationTitle, itemType, dateAdded, etc, and you can have a guess at what they mean, but it shouldn’t be necessary to guess what they mean, or to understand the relationships between them. I had to spend hours analysing the dataset I had extracted from the web API, before I could seriously attempt to convert it to some other form. There is some documentation scattered about here and there, but no authoritative description of the data format. Compare this to the situation with the more formalised formats which Zotero can export: TEI, RIS, MODS, etc, which have formal specifications defining all the terms in their vocabulary.

Is this something that Zotero could do? It’s hard to say; it would require some technical changes to the Zotero data server code, but probably more signficantly it would involve a change in collective mindset by the developers involved; to see Zotero’s data model as an abstraction independent of Zotero’s data server application; a genuine public language for communicating between arbitrary bibliographic systems, not merely a kind of window into the internal workings of a particular software system.

This is a common situation in web applications which offer an API: the application developers are focused intellectually on the application itself; its own internal workings and the functionality it provides, and they naturally tend to see the API as merely an aspect of that system. The idea that the data format of the API might have a life independent of their software, or that it might even outlive their software altogether, is a stretch. But if the data which their system works with is important, then it is surely important enough to accord some formal status: to give it a name; an Internet Media Type, even a namespace URI, to constrain it with a schema, and to explain it with formal documentation.

Next steps

As usual, the code I’m writing is published on GitHub; in the first instance this XProc pipeline to convert a Zotero library to EAD. But this was really a first stab at the problem; where I’d like to go is to try to formalise and specify the Zotero data format itself; to give it an XML encoding with a formal definition, and then to build other systems, such as Linked Data systems, on top of that formalised format.