Brief Notes from Kew

1
Brief Notes from Kew

• Mark Jackson Software Applications Manager

2
Focussing on...

• Herbarium digitisation
• electronic Plant Information Centre

3
Kew Herbarium

• Guesstimated
• 7 million specimens
• 250,000 types
• Less than 5 specimens databased
• A variety of personal databases

4
Preparation for Digitisation

• Computerise transactions
• Agree and document policy and procedures
• Establish core fields (HISPID pending ABCD)
• Develop hardware and software infrastructure
  (e.g. catalogue database, mass storage)

5
Digitisation Strategy

• Curators to barcode, database and image types for
  loan
• Repatriation research projects
• to use infrastructure and core fields
• data to be imported into Catalogue (eventually)
• Pursue digitisation projects

www.kew.org/data/repatbr
6
Specimen imaging

• Decision to try to match Cibachrome prints in
  terms of quality (e.g. suitable for many
  diagnostic purposes)
• 600 dpi delivers 200MB images
• Stored as uncompressed (but bzipped) TIFFs
• Acquisition of mass storage

7
HerbScan

• A3 flatbed scanner, inverted
• Cradle for specimens
• Distributed throughout Herbarium

8
Pros and cons
200 MB master images (600 dpi scans), based on
capturing the level of detail of Cibachromes.
Camera HerbScan

• 30-40,000
• 200MB images barely achievable
• 1 image per minute
• Fixed
• Versatile

• 7,500
• 200MB images easily achievable
• 10 images per hour
• Some mobility
• Suited to flat items

9
HerbCat enquiries
image enquiries
Client
Image Server
HerbCat
Images
Metadata
10
Focussing on...

• Herbarium digitisation
• electronic Plant Information Centre

11

• UK government funding for delivery of services
  electronically
• Resource-discovery interface to multiple Kew data
  sources (not necessarily at Kew)
• Data sources are heterogenous
• Simple interface overlaying other systems

ePIC Interface
Data source
Data source
Data source
Data source
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
Architecture
Interface (java servlet)/JSPs
Requests
Results
Multi-threaded Java server
Request queue
Data sources
Data sources
Handlers one per data source one for logging one
for spell-checking
Configuration files (XML)
16
Texts

• Web documents indexed using Lucene
• Flora Zambesiaca digitised and marked-up with XML
• Experimentation with options for query and output
  via Java servlet
• using XSL to output selections
• using Lucene to index the XML
• importing the XML into a database
• Other texts - jury still out, but Lucene route
  looks promising

17
Feedback

• Email mechanisms
• Web usability testing/focus groups
• Logging
• Quantitative success
• levels of usage, patterns trends
• beware crawlers, testing development staff,
  harvesters
• referring URLs, Google link popularity of site
• country, domain
• Qualitative success
• success of queries esp. zero hits (spelling,
  common names, families)
• performance system monitoring
• number of queries per session, return visits
• results pages viewed

18
World distribution of queries
19
Future

• www.kew.org/epic

• More data sources, including texts and images
• Hierarchical browsing front-end based around
  revamped Brummitt Families Genera with
  phylogenetic classification
• Looking forward to
• using the GBIF Names Service
• links with DiGIR/BioCASE resources...