designing a human literate library for the digital age
TRANSCRIPT
designing a human literate library for the digital ageTom ScottHead of Digital Engagement | Wellcome Collection
wellcome collection is a free museum & library exploring health, life and our place in the world
we seek to create opportunities for people to think deeply about the connections between science, medicine, life and art by making thought-provoking content and improving access to the diverse perspectives represented by our collections and through research
wellcome libraryinspired by the collections assembled by Henry Wellcome, we encourage great ideas about health by connecting science, medicine, life and art.
a history of medicine library? yes, but not only… medical professionals and scientists are people too. In acquiring their archives we also acquire their life’s work both professional and personal
Thomas Hodgkin (1798-1866)
pathologist (Hodgkin’s Lymphoma) and…
• an anti-slavery agitator
• a general social reformer
• a traveller
• a Quaker…
John Dixon (1832-1930)
Medical Officer of Health for Bermondsey but his papers also include:
• writings on languages
• board games
• photography
• writings on cacti
library of human life everything we do happens within the human body; wellcome library is therefore a box of human stories covering everything from conception and birth to illness and death and everything in between.
but people don’t know that, they don’t know what’s in the library… 1
what’s in the library? | alpha.wellcomelibrary.org
people don’t necessarily know what they are searching2
what do we have?images and metadata
lack of context 3
John Moore (1620-1702)Loving brother
London 19 June 1665
I hope these lines will finde you and yo[ur]s and all our freinds in the country well as blessed {be god} I am and all my family so long as god pleaseth: for we have {a} very crasie sickly time att London since June came in and are very fearefull it will grow worse every weeke while summer weather continues. for the plague increaseth much and spreads it selfe very strangely in the Citty and suburbs. 17 dyed one week 43 next and last weeke 112 of the plague and of all diseases 558 {last weeke} and much feared this weeks bill will farr exceed the last. it comes not out till Thursday morning. Now knowing young persons are most apt to take infection, [I] thought good to give you an accompt of it, to have your advice about Cusen John, to know your mind - whether you do not desire him home againe, or judge it the best way to have him into the Country againe till these sickly times be gone againe, and lett me know yo[ur] mind p[er] next, if you think fitt to lett him continue at London, I shalbe as carefull of him as my selfe, but as I said before youth is in more danger to take infection then older p[er]sons and if the sicknesse increases we shall have nothing to doe for it will put a stopp to all businesse: If god in mercy to us all put not a stopp to it. pray remember me & wife to yo[ur] brother George & sister & our Cusens Mr Mould & other friends as you see them with kinde love to yo[ur] selfe rest
your lov[ing] brother
John Moore
london’s dreadful visitationor, a collection of all the Bills of Mortality for this present year: beginning the 27th of December 1664 and ending the 19th of December following.
paper catalogues digital catalogues
searchable 🙁 😃
understandable 😃 😩We have atomised the collections to the point where you can find everything but have no idea what you’re looking at!
provide access for all 4
digital access digitisation and open licensing
reading experiencelibraries are designed to provide a great reading experience
why not online?
this isn’t about designing a ‘digital library’; it’s about looking at the contextual experience of our users
digital is a platform unto itself not (just) a catalogue for the physical library
we need to design a digital platform that helps users
…by encapsulating a librarian!
how are we going do that?
design a digital platform that’s as smart as a puppy…• helpful not passive• pays attention• try to do what you want not what
you ask for• learns
single domain model
traditional hierarchical model but the world isn’t hierarchical and knowledge is hidden
series model no way in! users need a top level entry point to collate and give context.
hybrid model collection level descriptions as authority files
combining datastored in such a way that we can choose on a case-by-case basis whether to use it/ how to use each dataset
Platform
data sources
combining data
Adapt
Transform
Ingest
API
data source
to domain
to search
for clientswellcomecollection.org
anyone…
understanding intent
paying attention to the context of queries
find the right collectionfind the right boxfind what’s in the boxsearch an item (book) in the box
datesthe data is complicated but not complex
• multiple date systems
• numerous modifiers
• ambiguous dates e.g. Spring Time
• fuzzy dates (19th century)
the complexity comes when dealing with the front end. How to present this information, know what people intend, facet data, use it for recommendations.
extracting meaning
optical character recognitionprinted text can be OCR’ed easily enough to identify:• text• tables• figures and images
what about handwriting?
right handLord Nelson
left handLord Nelson
image recognition and entity extraction
rekognitionAWS thinks this is:
• people (98.9%)
• person (98.9%)
• human (98.9%)
• brochure (70.3%)
• flyer (70.3%)
• poster (70.3%)
rekognitionAWS thinks this is:
• people (99%)
• person (99%)
• human (98.9%)
• playground (55.2%)
• lighting (52.4%)
OK can be good enoughAccuracy matters more if you link/display the relationship but e.g. knowing an entity is a person can be enough to improve search results or find related material.
The data might not be good enough to display but can be used as a hint to an algorithm to modify the sort order etc.
Can also use other data to improve the guesses…
synonyms in context
changing use of language
triangulate multiple data sources
providing context
provenance
the adamson collection
telling storiesthat we know because we research our collections
bidirectional links between stories, exhibitions & items in the collection what about books, articles etc. not in the collections?
[email protected] | @derivadowTHANK YOU | TOM SCOTT