Designing the Garden: Getting Grounded in Linked Data

Designing the Garden: Getting Grounded in Linked Data

#rwlod

Jenn Riley (@jenlrile)

Associate Dean, Digital Initiatives

McGill University

"Full fathom five thy father lies,

Of his bones are coral made,

Those are pearls that were his eyes,

Nothing of him that doth fade,

But doth suffer a sea-change,

into something rich and strange,

Sea-nymphs hourly ring his knell,

Ding-dong.

Hark! now I hear them, ding-dong, bell.”

--Shakespeare, The Tempest

#rwlod Beyond the Looking Glass, ALCTS Preconference 2015


Photo by Liam Moloney, https://flic.kr/p/7Qux27, CC BY-SA

https://flic.kr/p/7Qux27

1. It’s not “our” data and “their” data – it’s one big graph.


It’s about the connections

https://linkedjazz.org/network/

Graph-based world views


http://ebiquity.umbc.edu/blogger/2015/06/06/querying-rdf-data-with-text-annotated-graphs/

Connecting things together


dbpedia:William_Shakespeare

1615-04-23

1564-04-26

dbpedia:Stratford-

upon-Avon

dbpedia-owl:birthDate

worldcat-work:10745266

81

fast:1069678 # Political Refugees

“Tragicomedy’

“Shakespeare’s Comedy of the

Tempest”

schema:genre

viaf:96994048

# William Shakespeare

foaf:person

rdf: type

schema: CreativeWork

rdf: type

Creating so much data!

Where the scope of “library data” ends

What it is and isn’t “our job” to do

Making “complete” descriptions ourselves

Mapping data from “other” vocabularies into “library” vocabularies

This means we can stop worrying about…


Learning other metadata cultures

Being Linked Data ecosystem good citizens

How the technology and the data can most effectively work together

Making connections between things

Understanding other vocabularies and communities

And start worrying about…


2. We can expect more intelligence in the system.


http://hangingtogether.org/?p=4167; September 2014


Guido Reni (1575-1642)

Hercules Killing the Hydra of Lerna

c. 1620-1621

oil on canvas

commissioned along with other scenes from the mythology of Hercules in 1617 by FerdinandoGonzaga, for a room in the Villa Favorita in Mantua, Italy inv. 535

Musée du Louvre, Paris, France

Mine usage data to enhance relevance and utility in discovery

Start from the most relevant information and provide easy means for quick expansion on demand

Coherently display conflicting information

Give indications of provenance of information

What systems must do for users


If we’re serious about information literacy, we have to

give our users tools and thentrust them.


Flag dead ends for review and action

Normalize most string-based data

Highlight potentially conflicting information

Hide complexity (URIs, etc)

Mine and show candidate connections for review

What systems must do for metadata maintainers


Choosing one authoritative “correct” assertion in the face of conflicting data

Whether or not a given source meets a certain standard for authority

Authorized headings, access points

And textual justifications for them

A large proportion of the data cleanup tasks we used to do

For example, the formatting of strings

But we can expect a new set of these to emerge!



How to enhance system algorithms

How users best interact with complicated information

Methods for automated metadata creation and cleanup

Getting large amounts of new data into the system



3. The information age has provided a new definition of “authority.”


Current narrative:

Libraries create good metadata!

Because we’re trained to do so

We’re consistent and follow rules

That’s what makes good data

Other people create bad data

Because it’s not consistent

Using a well built record structure is a key part of good metadata

But…

We don’t read the books we catalogue

We don’t typically have expertise in the subjects of the works we describe

Sometimes we don’t even read or speak the language those works use

Our perspective is pretty different from our users’

Number of things to describe is quickly expanding and our budgets are shrinking

Let’s take a good, hard look at our “authoritative” data


Remember, we’re looking at more intelligent systems in the LD world

That can deal with inconsistency, masking it or cleaning it up

The LD graph allows us to not worry about metadata structures

So consistency and rules are no longer the primary drivers of good metadata

Which means we can turn large swaths of the creation of metadata over to domain experts

Wait, it’s not about consistency and rules?



https://www.digitalcommonwealth.org/search/commonwealth:5h73q9544

If we’re serious about good metadata, we need to start from

expert information.


We need this guy


http://www.betterlivingthroughbeowulf.com/scholars-lose-themselves-in-their-research/

And we need these folks too!


Photo by veggiesosage, https://flic.kr/p/5WjAsK, CC-BY-NC-ND

https://flic.kr/p/5WjAsK

Really, it’s going to be OK

https://xkcd.com/386/


The Linked Data community cares about provenance


http://www.w3.org/standards/techs/provenance#w3c_all



Deep research on materials for which there’s already a knowledgeable community

Descriptions being “complete”

Making sure everything is “right”



Mining the data that’s already out there

Promoting voices of those who engage with content

How usable systems can be built to generate Linked Data from activities real people already partake in

Seeding basic information for the rare and unique materials we hold that have never been released

Connecting data from communities operating in different languages



4. Our job is to tend the garden.


Metadata is an ecosystem

Photo by Temari 09, https://flic.kr/p/6UskT1, CC-BY-NC

https://flic.kr/p/6UskT1

The garden needs tending

Photo by Center for International Forestry Research, https://flic.kr/p/dbx1Gt,CC-BY-NC-ND

https://flic.kr/p/dbx1Gt

A new model – making connections

https://thenounproject.com/term/connection/25392/

“Original cataloguing” and “copy cataloguing”

Getting data into “our systems”

Doing all the work that needs to be done the first time we think about a specific item



Understanding the Linked Data environment

Locating large and useful datasets

Understanding vocabularies developed elsewhere

Finding good people that can analyze relationships and make new connections



The data and systems are all in the cloud

Library-based discovery less important but likely still around for a while

Several ways systems can navigate the graph

Crawling

Dereferencing

Query federation

(See http://linkeddatabook.com/editions/1.0/#htoc84)

So how will Linked Data systems work?

http://linkeddatabook.com/editions/1.0/#htoc84

This is all the way it should work

It’s going to be a while before we get there

Big effort needed to start connecting up these data sets

Data sets and tools will get better as we start using the data in this way and demand more

The library community can help to shape this evolution, but only if we fully understand and engage with the assumptions and mechanisms in play in the Linked Data community

Reality check


This will be hard. But that won’t stop us.

We need to redefine our baseline.

We need to rethink what new models mean for us.

And, most importantly, we need to put the right people in positions to work through these issues and get the details settled.

A sea-change? Most definitely.


We got this.

[email protected]

@jenlrile

These presentation slides: http://www.jennriley.com/presentations/alaannual2015/riley-garden.pptx

Thank you!


Education

Designing the Garden: Getting Grounded in Linked Data