METHODS

We created a SPARQL query using data.carnegiehall.org, which finds people from Carnegie Hall's performance history (e.g. performers, and/or creators like composers, arrangers, lyricists, etc.) born on today's date. Since birthdates have been stored as ISO-8601 dates assigned datatypes like xsd:date (YYYY-MM-DD), xsd:gYearMonth (YYYY-MM), or xsd:gYear (YYYY), we can use SPARQL's FILTER to find only those people born on today's month and day. Birthplaces are identified using GeoNames URIs (when the birth city is not known, birth country will be used; people with no birthplace recorded will not appear in the query). The query will also return the Wikidata item ID for anyone whose Carnegie Hall ID has been aligned with Wikidata using the skos:exactMatch property.

              PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
              PREFIX schema: <http://schema.org/##>
              PREFIX geo-pos: <http://www.w3.org/2003/01/geo/wgs84_pos##>
              PREFIX skos: <http://www.w3.org/2004/02/skos/core##>
              SELECT ?personName ?birthPlace ?birthPlaceLabel ?lat ?long ?opasID ?wikidataLink (YEAR(?date) as ?year)
              (IRI(CONCAT("https://www.carnegiehall.org/About/History/Performance-History-Search?q=&dex=prod_PHS&pf=",
                            (STR(ENCODE_FOR_URI(?personName))))) AS ?perfLink)
              (IRI(CONCAT("https://www.carnegiehall.org/About/History/Performance-History-Search?q=&dex=prod_PHS&cmp=",
                            (STR(ENCODE_FOR_URI(?personName))))) AS ?compLink)
              WHERE
              {
                  BIND(MONTH(NOW()) AS ?nowMonth)
                  BIND(DAY(NOW()) AS ?nowDay)

                  ?personID schema:birthDate ?date ;
                          schema:name ?personName ;
                          schema:birthPlace ?birthPlace .
                  ?birthPlace rdfs:label ?birthPlaceLabel ;
                              geo-pos:lat ?lat ;
                              geo-pos:long ?long .
                  OPTIONAL { ?personID skos:exactMatch ?wikidataLink .
                      filter contains(str(?wikidataLink), "wikidata")}
                  BIND(REPLACE(str(?personID), "http://data.carnegiehall.org/names/", "") as ?opasID)
                  FILTER (MONTH(?date) = ?nowMonth && DAY(?date) = ?nowDay)

              }
              ORDER BY ?year
              LIMIT 100
            

In order to provide an easily human-readable version of each person’s history at the hall, we also use SPARQL to create a link to Performance History Search, an HTML presentation of essentially the same dataset that we published first in 2013 (and predates our experiments with LOD). (In the query, this is found right after the SELECT statement, where you'll see (IRI(CONCAT( etc.)

CONCLUSIONS

what we learned

You might be asking why we need to formulate different versions of the PHS link. The HTML version launched in 2013, well prior to our release of the same data as RDF in 2017; although the source database is the same, the process that translates the data for display is a bit different and was developed separately. This creates a few challenges when attempting to create links to PHS search filters:

• Our source database for CH’s performance history data, a proprietary SQL-based product designed for concert planning, stores performers and composers in separate tables. When the data is surfaced in the HTML Performance History Search (PHS), that separation between composers and performers remains. Query filters are constructed from a search index based on the name string of the composer or performer.
• Our RDF version of the data solves this problem of (potential) dual IDs by creating a single ID for each named entity, with statements defining their role according to associations with creative works (as a composer, arranger, lyricist, etc.) and/or events (as a performer).
• In order to construct the PHS link, a URL-safe version of the Wikidata item label (i.e. the name of the composer or performer, with URL-encoded characters replacing spaces and other reserved characters) must be concatenated with a base URL, e.g. https://www.carnegiehall.org/About/History/Performance-History-Search?q=&dex=prod_PHS&pf=Juan%20Tizol.

further investigation

Eventually our goal is to bring all online historical content — our performance history and digital collections — into a single, unified user experience using our LOD as the metadata "backbone". The Carnegie Hall Data Lab is a first step in that direction, where we can begin experimenting with user-friendly ways to surface our performance history data.