Seman&c	
  Linking	
  of	
  
Informa&on,	
  Content	
  
and	
  Metadata	
  for	
  
Early	
  Music	
  	
  
	
  (SLICKMEM)	
...
People	
  
Goldsmiths	
  College	
  
Tim	
  Crawford	
  
David	
  Lewis	
  
	
  
BBC	
  
Ian	
  Knopke	
  
City	
  Univers...
Early	
  Music	
  Online	
  
(EMO,	
  JISC	
  Rapid	
  Digi&sa&on,	
  2011)	
  
Digi&sed	
  from	
  archival	
  microfilms	...
Vocal	
  part-­‐books	
  
Electronic	
  Corpus	
  of	
  Lute	
  Music	
  
(ECOLM	
  III,	
  AHRC	
  2012)	
  
Music	
  in	
  lute	
  tablature	
  en...
Lute	
  tablature	
  
EMO	
  metadata	
  
BL	
  MARC	
  records	
  exported	
  as	
  XML	
  
New	
  cataloguing	
  includes	
  a	
  range	
  of	...
EMO	
  metadata	
  
EMO	
  metadata	
  
EMO	
  metadata	
  
SLICKMEM	
  Objec&ves	
  	
  
Establish	
  concept-­‐lists	
  for	
  EMO	
  person	
  names,	
  
places,	
  dates,	
  and	...
Sources	
  of	
  error	
  
OCR	
  errors	
  (necessitate	
  human	
  interven&on)	
  
Cataloguers	
  have	
  to	
  deal	
 ...
Adrian	
  Willaert	
  
(c.	
  1490	
  –	
  1562)	
  
Adr.	
  Vuillart	
  
Adri.	
  Vuillart	
  
Adrian	
  Villart.	
  
Adr...
‘Susanne	
  un	
  jour’	
  
(Orlande	
  de	
  Lassus,	
  1532-­‐1594)	
  
Susane	
  un	
  jour	
  :	
  canzone	
  francese...
Authority	
  lists	
  
BL	
  uses	
  the	
  Library	
  of	
  Congress	
  Name	
  Authority	
  
files	
  for	
  personal	
  ...
Music	
  Informa&on	
  Retrieval	
  (MIR)	
  
We	
  can	
  use	
  symbolic	
  MIR	
  methods	
  to	
  ascertain	
  
iden&t...
Future	
  work	
  
Hopefully,	
  EMO	
  will	
  soon	
  expand	
  to	
  include	
  over	
  
2,000	
  early	
  printed	
  m...
www.ecolm.org	
  
www.earlymusiconline.org	
  
Nächste SlideShare
Wird geladen in …5
×

Semantic Linking of Information, Content and Metadata for Early Music (SLICKMEM) - Tim  Crawford (Semantic Media @ The British Library, 23 September 2013)

979 Aufrufe

Veröffentlicht am

This talk was given by Tim  Crawford (Goldsmiths, University of London) at the "Semantic Media @ The British Library" event on 23 September 2013.

Veröffentlicht in: Technologie, Unterhaltung & Humor, Bildung
  • Als Erste(r) kommentieren

  • Gehören Sie zu den Ersten, denen das gefällt!

Semantic Linking of Information, Content and Metadata for Early Music (SLICKMEM) - Tim  Crawford (Semantic Media @ The British Library, 23 September 2013)

  1. 1. Seman&c  Linking  of   Informa&on,  Content   and  Metadata  for   Early  Music      (SLICKMEM)   Tim  Crawford   (Goldsmiths)  
  2. 2. People   Goldsmiths  College   Tim  Crawford   David  Lewis     BBC   Ian  Knopke   City  University   Tillman  Weyde   Reinier  de  Valk     Oxford  eResearch  Centre   Kevin  Page  
  3. 3. Early  Music  Online   (EMO,  JISC  Rapid  Digi&sa&on,  2011)   Digi&sed  from  archival  microfilms  and  newly   catalogued  by  the  BL   300  books  of  16th-­‐century  printed  music   Roughly  10,000  pieces  of  music   Most  is  vocal  music  in  partbooks,  not  scores   About  30  books  in  various  forms  of  tablature   (mostly  lute,  but  also  guitar  and  keyboard);   including  many  arrangements  of  vocal  music  
  4. 4. Vocal  part-­‐books  
  5. 5. Electronic  Corpus  of  Lute  Music   (ECOLM  III,  AHRC  2012)   Music  in  lute  tablature  encoded  in  TabCode   Metadata  managed  in  SQL  database   Web-­‐based  interface  for  viewing,  edi&ng  and   playback  of  lute  music   EMO  lute  books  captured  by  automa&c  encoding   using  Op&cal  Tablature  Recogni&on   Correc&on  of  OTR  raw  data  crowd-­‐sourced  to  lute-­‐ playing  experts  using  web  interface   EMO  vocal  music  captured  by  OMR  (Aruspix)  
  6. 6. Lute  tablature  
  7. 7. EMO  metadata   BL  MARC  records  exported  as  XML   New  cataloguing  includes  a  range  of  types  of   musicological,  bibliographical  and  historical  informa&on   Also  o[en  includes  details  of  physical  condi&on,  prin&ng   history  and  ownership   Lists  musical  contents  with  composers  where  known,  but   does  not  describe  detailed  loca&ons  of  musical  items  in   the  part-­‐books   Names  can  include  composers,  poets,  printers,   publishers,  sponsors,  dedicatees,  owners  (early  and   modern)  
  8. 8. EMO  metadata  
  9. 9. EMO  metadata  
  10. 10. EMO  metadata  
  11. 11. SLICKMEM  Objec&ves     Establish  concept-­‐lists  for  EMO  person  names,   places,  dates,  and  work-­‐&tles;  evaluate  exis&ng   ontologies;  basic  RDF  of  EMO  metadata     Express  similar  concepts  in  EMO  metadata  with   Similarity  Ontology     Link  EMO/ECOLM  metadata  to  external  resources     Extract  music  features  (pitch/  &me,  chroma)  from   encoded  lute  and  vocal  music;  test  geometric  &  n-­‐ gram  pa`ern-­‐matching  across  EMO/ECOLM   Develop  pilot  Music  Descrip&on  Ontology  to   express  musical  features  
  12. 12. Sources  of  error   OCR  errors  (necessitate  human  interven&on)   Cataloguers  have  to  deal  with:   16-­‐c  spelling  conven&ons  (not  really  errors  as  such)   Prin&ng  errors  and  conven&ons,  abbrevia&ons,  etc.   Some  names  exis&ng  in  a  wide  variety  of  forms   The  same  musical  work  exis&ng  under  various  &tles   Cataloguers  some&mes  make  mistakes,  too!  
  13. 13. Adrian  Willaert   (c.  1490  –  1562)   Adr.  Vuillart   Adri.  Vuillart   Adrian  Villart.   Adrian.  Vuillart   Adrian  vuillart   Adrian.  Vvillaert   Adrian.  VVillart.   [Adrian  Willaert]   Adrian  Willart   Adriano   Adriano  Vuigliart   Adriano  VVillaert   Adrianus   Adrianus  Vuillart   Adrianus  VVillart   Adrianus  Willart   Adrianus  Williart   Adrien  vvillart   ...  etc.  
  14. 14. ‘Susanne  un  jour’   (Orlande  de  Lassus,  1532-­‐1594)   Susane  un  jour  :  canzone  francese  a  cinque   Susanna   Susanna  faire   Susanna  frumb   Susanne  ung  iour   Susanne  ung  Jour   Sussanna   Susane  un  jour  d'amour  solicitée   Susane  vn  jour  d’amour  solicitée   Susanne  un  jour  d’amour  sollicitée   Suzanne  ung  jour  damour  solicitee  
  15. 15. Authority  lists   BL  uses  the  Library  of  Congress  Name  Authority   files  for  personal  names   There  are  very  few  such  lists  for  musical  works   Rela&vely  li`le  of  the  music  in  EMO  has  been   recorded   What  about  the  rest?  
  16. 16. Music  Informa&on  Retrieval  (MIR)   We  can  use  symbolic  MIR  methods  to  ascertain   iden&ty  of  musical  works  and  disambiguate   conflicts   Similar  techniques  can  be  used  for  music   analysis  and  recogni&on  of  musical  quota&ons   Also  provides  means  to  link  with  external   resources,  such  as  RISM  OPAC   Publishing  cumula&ve  results  of  these   inves&ga&ons  will  provide  a  useful  resource  
  17. 17. Future  work   Hopefully,  EMO  will  soon  expand  to  include  over   2,000  early  printed  music  books  at  BL   Match  score-­‐content  with  recordings   (combining  symbolic  MIR  with  audio  MIR)   Link  EMO  data  with  BBC  Early  Music  Show   archive  as  proof  of  concept  for  wider  linking  of   broadcast  music  with  scores  and  other   resources  
  18. 18. www.ecolm.org   www.earlymusiconline.org  

×