Current design issues for digital archives Robert Munro (presented by David Nathan) Endangered Languages Archive (ELAR), School of Oriental and African.

Current design issues for digital archives

Robert Munro(presented by David Nathan)

Endangered Languages Archive (ELAR), School of Oriental and African Studies, London

2

Outline

1. Introduction

2. Archive architectures

3. Current Issues1. value-adding interaction from ‘end’ users

2. flexibility in access to materials

3. granularity of description of materials

4. Conclusions

3

Introduction – ELAR

Part of the Hans Rausing Endangered Languages Project (HRELP).

Open for deposits since October 2005.In the process of designing and implementing

key systems.

4

Introduction – ELAR

ELAR will be the first language archive that allows users to:add metadata in the language of their choice add new metadata (comments, descriptions, links) to

existing materialstranslate metadata into a language of their choiceselect language preference(s) for viewing existing

metadataadd metadata to archived materials at different levels of

granularity

5

Introduction – current issues

‘End’ users adding value to archive materialswho will moderate such additions?

Flexible support of accesscan an archive explicitly support multilingual users?

Metadata – comments / description of materials:should the granularity of description be at the level of:

files,collections of files,and/or sub-subsections of a file?

6

Archive architectures

Producers Silo

The classic ‘silo’ view of an archive:little more than disaster-proof backup

7

Archive architectures

Silo

The producers are not the only users:different dissemination formats are required…

Dissemination Producers

8

Archive architectures

Silo

The producers are not the only users:different dissemination formats are required…

…for different user communities

Dissemination Producers

Designated communities

9

Archive architectures

Silo

Working formats are not preservation formats:materials may need to be transformed on ingest

Dissemination Designated communities

IngestionProducers

10

Archive architectures

You cannot rigidly preserve digital data:file need to refreshed and migrated to current formats

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

11

Archive architectures

…but the objects, metadata and structures are still backed up in disaster-proof silo’s.

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

12

Archive architectures

Archives need to define three types of ‘packages’ingestion, archive and dissemination:

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

13

Ingestion (Accession) packages

Formats & structures that can be converted to archive formats with minimal effort:open file formatswell-documented structures: XML with schema ideal

The content needs to take into account the many potential uses of the materials:high quality sound and videoa variety of genresdetailed metadata and structural information

14

Dissemination packages

Many potential users of archived materials:researchersspeakers educators publishers

With many different requirements:access to materials by various methodsarchive servicescontinuation of ownership of language materials

15

Current issues – value adding

The current model is fairly uni-directionalbut users can/should add value to archive materials

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

16

Current issues – value adding

Users should be able to add to existing materials:speakers’ comments on contentresults of recent research

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

17

Current issues – value adding

The archive needs to trust certain users to add metadata to existing materials:should the identity of users be recorded / open?should users be able to challenge existing metadata?

Who to trust?depositors cannot moderate all comments on objects,

especially if comments can be in any languagebut can an archive deny a speaker’s request to add

comments to a recording of them speaking?

18

Current issues – flexibility of access

The archive cannot create different dissemination packages for every language and/or user:

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

19

Current issues – flexibility of access

Users should be able to personalize access:language preference(s) for metadatapreference on type of materials

Archive Dissemination

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

afd_34

dfa dfadf

fds fdafds

Designated communities

IngestionProducers

20

Current issues – flexibility of access

Flexibility of search / browse:keyword ‘search engine’ type searchrich relationships between objects for browsinggeographic searchesresearch community specific search

21

Current issues – flexibility of access

Flexibility of language:most metadata in most archives is in Englishshould metadata be multilingual?

22

Current issues – flexibility of access

If a user prefers to speak Quechua, then Spanish, then English:rather than accessing via one interface per

language…

ORbread

Photograph by Juan Pérez Martínez

January 2006

pan

Fotografia tomada por Juan Pérez Martínez

Enero 2006

pan

Fotografia tomada por Juan Pérez Martínez

Enero 2006

OR …

23

If a user prefers to speak Quechua, then Spanish, then English:…users should get all languages at once, according

to availability of data and their preferenceslabel in Quechua:photographer in Spanish:date in English:

Current issues – flexibility of access

t’anta

Fotografia tomada por Juan Pérez Martínez

January 2006

t’anta

Fotografia tomada por Juan Pérez Martínez

January 2006

24

Current issues – granularity

Archives tend to treat archived files as ‘atomic’metadata only refers to files as a whole

What abouta specific comment about a 20 second subsection of

the file? a general comment applying to many files?

25

Current issues – granularity

For example, suppose we have an annotated sound recording of some event:

26

Current issues – granularity

Some metadata is about the file as a whole:date recorded, speakers, title

27

Current issues – granularity

Some metadata is about sub-segments:name of a significant person or place specific linguistic phenomena

28

Current issues – granularity

It is likely that users will want to:add comments to such subsections richly link subsections to other items make unambiguous reference to subsections

At the time of deposit, no one can predict which subsections of files will later be significant:users need to be able to explicitly define subsections

of archive objects

29

Conclusions

Archives are not static repositories:an archive supports materials for multiple different

user communities in parallel

Value-adding interaction:archived materials can be further enriched by users

Flexibility in access to materials:personalizable interaction with archive materials

Granularity of description of materials:user defined granularity of materials

30

Thank you