Top Banner
Linking the American Art Museum to the Cloud What is Linked Open Data? Data published by exis0ng internet protocols that use a URI (Unique Resource Indicator) as the primary discoverable en0ty for a resource (e.g. person, object, web page, etc.) THE FIVE STARS OF LOD: make your stuff available on the web (whatever format) under an open license ★★ make it available as structured data (e.g., Excel instead of image scan of a table) ★★★ use nonproprietary formats (e.g., CSV instead of Excel) ★★★★ use URIs to iden0fy things, so that people can point at your stuff ★★★★★ link your data to other data to provide context
12

Linking American Art to the Cloud

May 15, 2015

Download

Technology

Overview of the project at the Smithsonian American Art Museum to publish its collections data as 5-star Linked Open Data.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  What  is  Linked  Open  Data?    Data  published  by  exis0ng  internet  protocols  that  use  a  URI  (Unique  Resource  Indicator)  as  the  primary  discoverable  en0ty  for  a  resource  (e.g.  person,  object,  web  page,  etc.)    THE  FIVE  STARS  OF  LOD:  ★  make  your  stuff  available  on  the  web  (whatever  format)  under  an  open  license    ★★  make  it  available  as  structured  data  (e.g.,  Excel  instead  of  image  scan  of  a  table)    ★★★  use  non-­‐proprietary  formats  (e.g.,  CSV  instead  of  Excel)    ★★★★  use  URIs  to  iden0fy  things,  so  that  people  can  point  at  your  stuff    ★★★★★  link  your  data  to  other  data  to  provide  context  

Page 2: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  What  is  good  for?    •  Making  your  data  more  discoverable  and  useful  by  everybody  •  Making  the  web  machine-­‐readable  at  a  more  granular  level  •  Allowing  for  more  sophis0cated  queries  using  inference  •  Connec0ng  your  data  to  other  people’s  data  

For  American  Art,  Linked  Open  Data  will:  •  Make  our  collec0ons  data  more  findable  on  the  web  •  Create  connec0ons  with  other  museums  that  have  related  artworks  •  Create  connec0ons  with  other  non-­‐museum  resources,  such  as  the  New  York  Times  •  Create  connec0ons  with  our  dispersed  content  on  social  media  (e.g.  Flickr)  •  Help  us  beSer  adapt  to  the  changing  web    

Page 3: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  Examples    Europeana    •  Digi0zed  collec0ons  of  museums,  libraries,  archives  and  galleries  across  Europe.  •  Open  metadata  on  20  million  texts,  images,  videos  and  sounds  •  A  subset  of  2.4  millions  objects  from  8  direct  Europeana  providers  encompassing  over  200  

cultural  ins0tu0ons  from  15  countries  is  served  according  to  the  Linked  Data  recipes.  •  Virtual  exhibi0ons  showcase  some  of  the  content  available.    

Page 4: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  Examples    Pelagios    •  Stands  for  'Pelagios:  Enable  Linked  Ancient  Geodata  In  Open  Systems’  •  Aim  is  to  help  introduce  Linked  Open  Data  into  online  resources  that  refer  to  places  in  the  

Ancient  World.  •  Allows  you  to  find  content  related  to  a  specific  place  

Page 5: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  GeJng  Started    IniKal  QuesKons  •  Will  it  take  a  lot  of  0me  and  resources  to  prepare  our  data?  •  How  does  LOD  differ  from  what  a  Google  search  can  do?  •  Is  it  foolish  to  be  doing  this  before  standards  are  in  place?  •  What  if  people  create  inappropriate  links  to  our  data?  •  Will  it  be  worth  the  0me  and  effort  in  the  end?  •  How  do  we  handle  all  of  the  non-­‐public  data  that  we  have?  •  Is  it  possible  to  make  sense  of  all  the  acronyms?  

The  Project  •  Working  with  the  Informa0on  Sciences  Ins0tute  (ISI)  and  Department  of  Computer  Science  at  

the  University  of  Southern  California.  •  Goal:  Publish  5-­‐star  Linked  Open  Data  of  our  complete  collec0ons  data  (41,000  objects,  

8,000  ar0sts).  •  Project  Phases:  Prepare  the  data,  Create  an  ontology,  map  the  data  to  RDF,  link  the  data  to  

hub  datasets,  publish  the  data.  

Page 6: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  The  Process    Preparing  the  data  •  Collec0ons  data  is  stored  in  TMS.  We  have  over  100  tables  •  We  decided  to  publish  only  the  data  that  is  already  visible  on  our  website  •  We  used  an  exis0ng  output  report  from  our  database  •  Several  fields  needed  to  be  interpreted  first  before  they  could  be  mapped  to  RDF    Designing  the  Ontology  •  We  built  our  ontology  around  exis0ng  ontologies    

•  An  augmented  version  of  Europeana  Data  Model  v.2  for  overall  framework;  SKOS  for  classifica0on  of  artworks,  ar0st  and  place  names;  Dublin  Core  for  tombstone  data;  RDA  Group  2  Elements  for  biographical  informa0on;  schema.org  for  geographical  data.  

Page 7: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  The  American  Art  Ontology  

Page 8: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  The  Process    Mapping  the  Data  to  RDF  (Resource  DescripKon  Framework)  •  Used  KARMA  tool  to  model  the  data  •  The  system  learns  with  each  dataset  so  the  process  becomes  easier  and  faster  

       

   For  Example:    Subject              Predicate      Object              

   www.americanart.si.edu/linkeddata/person/3406      saam:Person      “Thomas  Moran”    www.americanart.si.edu/linkeddata/person/3406    rdaGr2:dateOfBirth    “1837”    www.americanart.si.edu/linkeddata/person/3406    owl:SameAs      hSp://live.dbpedia.org/page/Thomas_Moran    

Page 9: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  The  Process    Linking  the  Data  to  External  Data  •  Verify  matches  before  publishing  •  Have  already  linked  ar0sts  to:  

•  DBPedia  -­‐  2,194  •  New  York  Times  -­‐  70  

•  Addi0onally,  can  link  ar0sts  to:  •  GeSy  Union  List  of  Ar0st  Names  -­‐  2,110  (ULAN  is  not  yet  published  as  LOD,  but  will  be)  •  Rijksmuseum  dataset  –  551  (links  are  not  yet  verified)  

•  In  the  works:  •  Linking  places  to  GeoNames  •  Linking  concepts  to  AAT  •  Linking  to  datasets  from  other  museums  •  Linking  to  social  media  content  

Publishing  •  Plan  to  publish  complete  dataset  and  all  verified  links  under  a  CC0  license  

•  Data  will  be  CC0,  but  images  will  be  maintained  under  a  restricted  license  •  Include  example  records  and  SPARQL  endpoint  

Page 10: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  Some  answers    Answers  to  IniKal  QuesKons:  •  Will  it  take  a  lot  of  0me  and  resources  to  prepare  our  data?  

•  Using  KARMA  to  model  the  data  and  a  visual  interface  to  verify  the  links  reduced  the  staff  Eme  that  would  have  been  needed  to  do  this  manually.  Working  with  ISI  certainly    helped  kick-­‐start  the  process.  

•  How  does  LOD  differ  from  what  a  Google  search  can  do?  •  LOD  eliminates  the  “noise”  of  a  Google  search.  With  LOD  you  can  query  specific  facts.  

With  Google  you  query  documents  and  then  have  to  read  the  document  to  get  the  facts.      •  Is  it  foolish  to  be  doing  this  before  standards  are  in  place?  

•  There  are  already  some  standards  in  place.  Plus,  being  one  of  the  first  means  that  we  have  the  opportunity  to  help  shape  the  standards.  

•  What  if  people  create  inappropriate  links  to  our  data?  •  You  cannot  control  what  people  say  about  you  on  the  internet!  

•  Will  it  be  worth  the  0me  and  effort  in  the  end?  •  We  believe  so!  It  will  allow  us  to  beTer  adapt  to  the  future  of  the  web.  

•  How  do  we  handle  all  of  the  non-­‐public  data  that  we  have?  •  We  opted  to  publish  only  our  public  data.  

•  Is  it  possible  to  make  sense  of  all  the  acronyms?  •  Yes!  It  takes  Eme,  but  you  do  eventually  grasp  all  the  different  terms.  

Page 11: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  Some  conclusions    •  We  ini0ally  planned  to  use  only  a  sample  of  collec0ons  data.  In  the  end,  we  used  data  for  

our  en0re  collec0on  –  over  41,000  objects!    •  Linking  to  datasets  like  DBPedia  and  the  New  York  Times  will  greatly  expand  the  content  we  

offer  on  our  website.  •  Linking  to  datasets  from  other  art  museums  will  increase  the  accessibility  and  reach  of  art  

collec0ons  and  cultural  heritage  online.  •  We’re  excited  for  the  poten0al  to  link  to  our  content  on  social  media  sites  –  an  object  page  

as  a  “hub”  to  all  types  of  content  about  that  object.  •  We  see  great  poten0al  in  using  Linked  Open  Data  to  curate  stories  about  artworks  and  

ar0sts  that  connect  museums  and  datasets  around  the  world  in  new  and  surprising  ways.  

Page 12: Linking American Art to the Cloud

Linking  the  American  Art  Museum  to  the  Cloud  What’s  next?    •  Embedding  linked  content  on  object  pages  and  ar0st  pages  on  our  website  (Wikipedia,  the  

New  York  Times,  etc.)  •  Improve  representa0on  of  ar0sts  on  Wikipedia,  adding  ar0cles  and  infoboxes  where  possible  

to  increase  the  number  of  matches  in  DBPedia.  •  Create  an  ongoing  maintenance  plan  to  ensure  that  the  linked  open  data  reflect  new  and  

edited  museum  data.  •  Tag  object-­‐  and  person-­‐related  museum  content  on  social  sites  like  Flickr  and  YouTube  so  

that  we  can  create  links  to  that  content  on  our  website,  too.  •  Inves0gate  mapping  and  linking  an  artwork’s  subject.  •  Expand  the  LOD  in  ways  that  will  enhance  research.  •  Create  a  tool  that  allows  users  to  “curate  stories”  using  LOD:  

•  hSp://prezi.com/htrvh2jrcsio/cura0ng-­‐stories-­‐with-­‐linked-­‐open-­‐data/  •  Encourage  others  to  build  applica0ons  with  our  data.