Zegar astronomiczny w PradzeBy Steve Collis from Melbourne, Australia (Astronomical Clock Uploaded by russavia) [CC BY 2.0], via Wikimedia Commo

W jednym z poprzednich wpisów na blogu “Czy umiemy pisać daty?” omawiałem podstawy uniwersalnej notacji  czasu i dat, zdefiniowanej w międzynarodowym standardzie ISO 8601 i jego uproszczonej wersji konsorcjum W3C. Od tego czasu Biblioteka Kongresu Amerykańskiego zakończyła prace nad rozszerzonym standardem, Extended Date/Time Format (EDTF) 1.0. Większa część EDTF dotyczy zapisu nieprecyzyjnych dat. Taka niedokładna lub nieprecyzyjna informacja dotycząca czasu występuje często w zapisach wydarzeń historycznych, np. w archiwach czy naukach bibliotecznych. Standard ISO 8601 nie pozwala na wyrażenie takich konceptów jak “w przybliżeniu rok 1962”, “któryś rok pomiędzy 1920 a 1935” czy “wydarzenie miało prawdopodobnie miejsce w roku 1938, ale nie jesteśmy tego pewni”. Standard EDTF pozwala na zapisanie w postaci zrozumiałej przez komputer takich konceptów, wypełniając potrzeby istniejące w wielu polach wiedzy mających do czynienia z metadanymi o charakterze historycznym.

Mimo tego, że standard EDTF jest stosunkowo nowy i nie ma zbyt wiele narzędzi programowych pomagających wprowadzać takie dane, sądzę, że warto jest zaznajomić się z tą nowa notacją i używać jej w miarę możliwości.


Chciałbym rozpocząć dyskusję kilkoma definicjami; symbole pojawiające się przy definicjach będą opisane dalej.


Precyzja jest miarą zakresu, wewnątrz którego mieści się ‘prawdziwa’ wartość [1]. Precyzja jest jednoznacznie zdefiniowana w wyrażeniach daty i daty/czasu. Jeśli wydarzenie miało miejsce w roku 1318, zapis taki posiada precyzję jednego roku (mogło mieć miejsce w dowolnym czasie w ciągu tego roku). Jeśli podamy 1318-05, zwiększamy precyzję do jednego miesiąca, a 1945-09-15 posiada precyzję jednego dnia, itp [2]. W EDTF możemy rozszerzyć tę definicję określając precyzję dziesięcio- lub stulecia używając symbolu x (patrz 'precyzja maskowana' poniżej).

Part II: Product

(Guest blog by Rob Hudson)

Arthur Rubinstein (Linked Data)In Part I of this blog, I began telling you about my experience transforming Carnegie Hall’s historical performance history data into Linked Open Data, and in addition to giving some background on my project and the data I’m working with, I talked about process: modeling the data; how I went about choosing (and ultimately deciding to mint my own) URIs; finding vocabularies, or predicates, to describe the relationships in the data; and I gave some examples of the links I created to external datasets.

In this installment, I’d like to talk about product: the solutions I examined for serving up my newly-created RDF data, and some useful new tools that help bring the exploration of the web of linked data down out of the realm of developers and into the hands of ordinary users. I think it’s noteworthy that none of the tools I’m going to tell you about existed when I embarked upon my project a little more than two years ago!

As I’ve mentioned, my project is still a prototype, intended to be a proof-of-concept that I could use to convince Carnegie Hall that it would be worth the time to develop and publish its performance history data as Linked Open Data (LOD) — at this point, it exists only on my laptop. I needed to find some way to manage and serve up my RDF files, enough to provide some demonstrations of the possibilities that having our data expressed this way could afford the institution. I began to realize that without access to my own server this would be difficult. Luckily for me, 2014 saw the first full release of a linked data platform called Apache Marmotta by the Apache Software Foundation. Marmotta is a fully-functioning read-write linked data server, which would allow me to import all of my RDF triples, with a SPARQL module for querying the data. Best of all, for me, was the fact that Marmotta could function as a local, stand-alone installation on my laptop — no web server needed; I could act as my own, non-public web server. Marmotta is out-of-the-box, ready-to-go, and easy to install — I had it up and running in a few hours.


Rob HudsonRob Hudson - Photo by Gino Francesconi

Part I: Process

(Guest blog by Rob Hudson)

My name is Rob Hudson, and I’m the Associate Archivist at Carnegie Hall, where I’ve had the privilege to work since 1997. I’d like to tell you about my experience transforming Carnegie Hall’s historical performance history data into Linked Open Data, and how within the space of about two years I went from someone with a budding interest in linked data, but no clue how to actually create it, to having an actual working prototype.

First, one thing you should know about me: I’m not a developer or computer scientist. (For any developers and/or computer scientists out there reading this right now: skip to the next paragraph, and try to humor me.) I’m a musician who stumbled into the world of archives by chance, armed with subject knowledge and a love of history. I later went back and got my degree in library science, which was an incredibly valuable experience, and which introduced me to the concept of Linked Open Data (LOD), but up until relatively recently, the only lines of programming code I’d ever written was a “Hello, World!” - type script in Basic — in 1983. I mention this in order to give some hope to others out there like me, who discovered LOD, thought “Wow, this is fantastic — how can I do this?”, and were told “learn Python.” Well, I did, and if I can do it, so can you — it’s not that hard. Much harder than learning Python — and, one might argue, more important — is the much more abstract process of understanding your data, and figuring out how to describe it. Once you’ve dealt with that, the transformation via Python is just process — perhaps not a cakewalk, but nonetheless a methodical, straightforward process that you can learn and tackle, step by step.

Now let me tell you a bit about the data that I worked with for my linked data prototype. The Carnegie Hall Archives maintains a database that attempts to track every event, both musical and nonmusical, that has occurred in the public performance spaces of Carnegie Hall since 1891. (Since the CH Archives was not established until 1986, there are some gaps in these records, which we continue to fill in using sources like digitized newspaper listings and reviews, or missing concert programs we buy on eBay.) This database now covers more than 50,000 events of nearly every conceivable musical genre: classical, folk, jazz, pop, rock, world music, and no doubt some I’m overlooking.  But Carnegie Hall has always been about much more than music; its stages have also featured dance and spoken word performances, as well as meetings, lectures, civic rallies, political conventions — there was even a children’s circus, complete with baby elephants, in 1934. Our database has corresponding records for more than 90,000 artists, 16,000 composers and over 85,000 musical works. Starting in 2013, we began publishing some of these records to our website, where you can now find the records for nearly 18,000 events between 1891 and 1955.  The limited release reflects our ongoing process of data cleanup, and we’re continuing to publish new records each month.  For my linked data prototype, I chose to use this published data set, since I knew it was good, clean data.