LoC gets semantic

This morning, the Library of Congress launched http://id.loc.gov/authorities/, their first serious entry into the semantic web.

The site makes the Library of Congress Subject Headings available as defererenable URLs. For example http://id.loc.gov/authorities/sh90005545.

NDHA demo and the National Library

This morning I went to the NDHA demonstration where a National Library techie talked us through the NDHA ingest tools. The tools are the most visible piece of the NDHA infrastructure, and are designed to unify the ingest of digital documents, whether they are born-digital documents physically submitted (i.e. producers mail in CDs/DVDs etc); born-digital documents electronically submitted (i.e. producers upload content via wizzy web tools); or digital scans of current holdings produced as part of the on-going digitisation efforts. The tools have a unified system with different workflows for unpublished material (=archive) and published material (=libarary). The unification of library and archival functionality seemed like futile ground for miscommunication.

The infrastructure is (correctly) embedded across the library, and uses all the current tools for collection maintenance, searching and access.

As a whole it looks like the system is going to save time and money for large content producers and capture better metadata for small donors of content, which is great. By moving the capture of metadata closer to the source (while still allowing professional curatorial processes for selection and cataloguing), it looks like more context is going to be captured, which is fabulous.

A couple of things struck me as odd:

1. The first feedback to the producer/uploader is via a human. Despite having an elaborate validation suite, the user wasn't told immediately "that .doc file looks like a PDF, would you like to try again?" or "Yep, that *.xml file is valid and conforming XML" Time and again studies have shown that immediate feedback to allow people to correct their mistakes immediately is important and effective.
2. The range of metadata fields available for tagging content was very limited. For example there was no Iwi/Hapu list, no Maori Subject Headings, no Gazetteer of Official Geographic Names. When I asked about these I was told "that's the CMS's role" (=Collection Management Software, i.e. those would be added by professional cataloguers later), but if you're going to move the metadata collection to as close to content generation, it makes sense to at least have the option of proper authority control over it.

Or that's my take, anyway. Maybe I'm missing something.

Report from the NDHA's International Perspectives on Digital Preservation

NOTE: I'm a computer scientist by training and this was largely librarian/archivist gig, so it's entirely possibly I've got the wrong end of the stick on one or more point in the summary below. It's also my own summary, and not the position of my employer, even though I was on work time during the event.

The NDHA is about to announce that the NDHA project has been completed on time and under budget. This is particularly pleasing in light of the poor history of government IT failures over the course of the last 30 years and a tribute to all concerned. Indeed, when I was taking undergraduate courses in software engineering a contemporary national library project was used as a text-book example of how not to run a software development undertaking. It's good to see how far they've come.

The event itself was a one-day event in the national library auditorium, with a handful of overseas speakers. I'm not entirely certain that a handful of foreigners counts as "international," but maybe that's just me being a snob. Certainly there was a fine turn-out of locals, including many from the National Library, the Ministry of Culture and Heritage and from VUW, including a number of students, who couldn't possibly have been there for the free food.

There seemed to be an underlying tension between librarianship and archivistship running through the event. I see this as being a really crazy turfwar, personally, since I see the chances of libraries and archives existing as separate entities and disciplines in fifty years seems pretty slim. The separation between the two, the "uniqueness" of objects in an archive seems to be to be obliterated by the free-duplication of digital objects. I've heard people say that archives also work access controls and embargoes for their depositors, but then so can libraries, particularly those in the military and those working with classified documents.

It seemed to me that the word "reliability" was used in a confusing number different of ways by different people. Without naming the guilty parties:

1. reliability as the truthfulness of the documents in the library/archive. This is the old problem of ingestors having to determine the absolute veracity of documents
2. reliability as getting the same metadata every time. This seems odd to me, since systems with audit control give _different_ results every time, because information on the previous accesses is included in the metadata of subsequent accesses
   
3. reliability as the degree to which the system conformed to a standard/specification

On reflection this may have been a symptom of the different vocabulary used by librarians and archivists. Whatever the cause, if we're wanting to spend public money, we have to be able to explain to the public what we're doing, and this isn't helping.

The organisers told us the presentations would be up by tonight (the evening of the presentation), but you won't find them on google if you go looking, because they tell google to please f**k off. I guess this is what someone was referring to when they said we had to work to make content accessible to google. The link is http://ndha-wiki.natlib.govt.nz/ndha/pages/IPoDP%2009%20Presentations and most were up at the time of writing.

I was hugely encouraged by the number of pieces of software that seemed to be being open sourced, as I see this as being a much better economic model than paying vendors for custom software, particularly since it's potentially scalable out from the national and top-tier libraries/archives/museums out to the second and third tier libraries/archives/museums, which by dint of their much larger numbers actually serve the most users and have the most content. It was unfortunate that the national library hasn't looked beyond propriety software for non-specialist software but continues to use AbodePhotoshop / Microsoft Windows, which are available only for limited periods of time on certain platforms (which will inevitably become obsolete), rather than openoffice, GIMP, etc, which are cross platform and licensed under perpetual licences which include the right to port the software from one platform to another. I guessPhotoshop / Windows is what their clients and funders know and use.

With a number of participants I had conversations about preservation. Andrew Wilson in his presentation used the quote:

“traditionally, preserving things meant keeping them unchanged; however our digital environment has fundamentally changed our concept of preservation requirements. If we hold on to digital information without modifications, accessing the information will become increasingly difficult, if not impossible” Su-Sing Chen, “The Paradox of Digital Preservation”, Computer, March 2001, 2-6

If you think about what intellectual objects we have from the Greeks (which is were us Westerners traditionally trace our intellectual history from), the majority fall into two main classes: (a) art works, which have survived primarily through roman copies and (b) texts, which have survived by copying, including a body of mathematics which were kept alive in the Arabic translation during a period when we Westerners were burning the works in Latin and Greek and claiming that the bible was the only book we needed. I'll grant you that a high-quality book will last maybe 500 years in a controlled environment, maybe even 1000, but for real permanence, you just can't get past physical ubiquity. If we have things truly worthy of long-term preservation, we should be striking deals with the Warehouse to get them into every home in the country, and setting them as translation exercises in our language learning courses.

I had some excellent conversations with other participants at the event, including Phillipa Tocker from Museums Aotearoa / Te Tari o Nga Whare Taonga o te Motu who told me about the http://www.nzmuseums.co.nz/ site they put together for their members.

Looking at the site I'm struck by how similar the search functionality is to http://www.nram.org.nz/. I'm not sure whether their relative similarity is a good thing (because it enables non-experts to search the holdings) or a bad thing (because by lowering themselves to the lowest common denominator they've devalued their uniqueness). While I'm certain that these websites have vital roles in the museums and archives community respectively, I can't help but feel that from an end-users perspective have two sites rather than one seems redundant, and the fact that they don't seem to reference/suggest any other information sources doesn't help. I can't imagine a librarian/archivist not being forth-coming with a suggestion of where to look next if they've run out of local relevant content---why should our websites be any different?

I recently changed the NZETC to point to likely-relevant memory institutions when a search returns no results (or when a user pages through to the end of any list of results).

I also talked to some chaps from Te Papa about the metadata they're using to to represent places names (Getty Thesaurus of Geographic Names) and species names (ad-hoc). At the NZETC we have many place names marked up (in NZ, Europe and the Pacific), but are not currently syncing with an external authority. Doing so would hugely enable interoperability. Ideally we'd be using the shiny new New Zealand Gazetteer of Official Geographic Names, but it doesn't yet have enough of the places we need (it basically only covers places mentioned in legislation or treaty settlements). It does have macrons in all the right places though, which is an excellent start. We currently don't mark up species names, but would like to, and again an external authority would be great.

It might have been useful if the day had included an overview of what the NDHA actually was and what had been achieved (maybe I missed this?).

flickr promoting the commons / creative commons

flickr is promoting photos in what it calls "The Commons", but only to logged in users. Normal users (who can't comment / tag the photos in the commons anyway) don't get an obvious link to them (except via about a billion third party sites such as blogs and google search). The page also shows how the national library's choice not to include text in their logo has come up trumps.

The confusingly similarly named "The Commons" and "Creative Commons" parts of the website apparently don't reference each other. Odd.

Excellent stuff from New Zealand Geographic Board Ngā Pou Taunaha o Aotearoa

A while ago, motivated by the need for an authoritative list of New Zealand place names for our with at the NZETC, I criticised the NZGB fairly roundly.
While they haven't produced what I/we want/need, in the last couple of months they've made huge progress in an unambiguously right direction.
Their primary work is the New Zealand Gazetteer of Official Geographic Names, a list of all official place names in New Zealand. It uses have a peculiar definition of "official" (= mentioned in legislation or a Treaty of Waitangi settlement), they have very few names of inhabited places (and no linking with the much larger ones maintained by official bodies such as the police and fire service), They have no elevation data for mountains and pass (which are defined by their height) and they define some things as points when they appear to be areas (such as Arthur Pass National Park), but it's much better than the New Zealand Place Names Database since:

1. It has a statutory reference for every place, given the source of the officialness of the name
2. It fully support Macrons
3. It has a machine readable-list of DoC administered lands --- I can imagine this being used for all sorts of interesting things, getting people out in other scenic and marine reserves.

NZGB sent around an email in which they explicitly addressed some of the points I'd earlier raised (I'm sure I wasn't the only one):

It should be noted that some of the naming practices of the past will have to be lived with, despite inconsistencies. Moving forward, the rules of nomenclature followed by the NZGB are designed to promote standardisation, consistency, and non-ambiguity. The modern format for dual names is '<Maori name> / <non-Maori name', which the NZGB has applied for the past 10 years, though Treaty settlement dual names sometimes deviate from this convention, because the decision is ultimately made by the Minister for Treaty of Waitangi Negotiations. Older forms of dual names, with brackets, will remain depicted as such until changed through the statutory processes of the NZGB Act 2008. These are not generally regarded as alternative names.
Macrons in Maori names have posed problems for electronic databases. Nevertheless they are part of the orthography, recommended by the Maori Language Commission, and the Board endorses their use. The Gazetteer will include macrons where they are formalised as part of the official name. When Section 32 of the new Act comes into force, official documents will be required to show official names, and these will need to include macrons where they have been included as part of the official name (unless the proviso is used). A list of those official names which have macrons is at http://www.linz.govt.nz/placenames/researching-place-names/macrons/index.aspx . LINZ's Customer Services has some solutions for showing macrons in LINZ's own databases and on published maps and charts, and is currently investigating how bulk data extracts might include information about macrons, for the customer's benefit.

Despite the name, it isn't clear in my mind exactly what's official and what isn't. Is the content of the "coordinates" column official? For railway lines this is a reference to the description, which in the cases of railways is usually of the form "From X to Y", where X and Y are place names, frequently place names that aren't on the list, so are thus presumably not official. Unless I'm going blind there is also no indication of accuracy on the physical measurements.

fuzzziness

I've been using topic maps in my day job, so I decided to try out http://www.fuzzzy.com/, a social bookmark engine that uses an underlying topic map engine.
I tried to approach fuzzzy with an open mind, but the increasingly stumbling on really annoying (mis-)features.

1. This is the first bookmark engine I've ever used hat doesn't let users migrate their bookmarks with them. This is perhaps the biggest single feature fuzzzy could add to attract new users, since it seems that most people who're likely to use a bookmark engine have already played with another one long enough to have dozens or hundreds of bookmarks they'd like to bring with them. I know this is non-ideal from the point of view of the social bookmark engine they're migrating too, since it makes it hard to do things completely differently, but users have baggage.
2. While it'd possible to vote up or vote down just about everything (bookmarks, tags, bookmark-tags, users, etc), very little is actually done with these votes. If I've viewed a bookmark once and voted it down, why is it added to my "most used Bookmarks"? Surely if I've indicated I don't like it the bookmark should be hidden from me, not advertised to me.
3. For all the topic map goodness on the site, there is no obvious way to link from the fuzzzy topic map to other topic maps.
4. There doesn't seem to be much in the way of interfacing with other semantic web standards (i.e. RDF).
   
5. The help isn't. Admittedly this may be partly because many of the key participants have English as a second language.
6. There's a spam problem. But then everywhere has a spam problem.
7. It's not obvious that I can export my bookmarks out of fuzzzy in a form that any other bookmark engine understands.

These (mis-)features are a pity, because at NZETC we use topic maps for authority (in the librarianship sense), and it would be great to have a compatible third party that could be used for non-authoritative stuff and which would just work seamlessly.

Place name inconsistencies

I've been looking at the "Dataset of New Zealand Geographic Place Names" from LINZ. This appears to be as close as New Zealand comes to an Official list of place names. I've been looking because it would be great to use as an authority in the NZETC.

Coming to the data I was aware of a number of issues:

1. Unlike most geographical data users, I'm primarily interested in the names rather than the relative positions
2. New Zealand is currently going through an extended period of renaming of geographic features to their original Māori names
3. The names in the dataset are primarily map labels and are subject to cartographic licence
   

What I didn't expect was the insanity in the names. I know that there are some good historical reasons for this insanity, but that doesn't make it any less insane.

1. Names can differ only by punctuation. There is a "No. 1 Creek" and a "No 1 Creek".
2. Names can differ only by presentation. There is a "Crook Burn or 8 Mile Creek", an "Eight Mile Creek or Boundary Creek" and an "Eight Mile Creek" (each in a different province).
3. There is no consistent presentation of alternative names. There is "Saddle (Mangaawai) Bivouac", "Te Towaka Bay (Burnside Bay)", "Queen Charlotte Sound (Totaranui)", "Manawatawhi/Three Kings Islands", "Mount Hauruia/Bald Rock", "Crook Burn or 8 Mile Creek" and "Omere, Janus or Toby Rock"
4. There is no machine-readable source of the Māori place names with macrons, and the human readable version has contains subtle difference to the machine-readable database (which contains no non-ASCII characters). For example "Franz Josef Glacier/Kā Roimata o Hine Hukatere (Glacier)" and "Franz Josef Glacier/Ka Roimata o Hine Hukatere" differ by more than the macrons. There appears to be no information on which are authoritative.

Right now I'm find finding this rather frustrating.

(grammar edit)