Linked Data for Librarians

by Seth van Hooland and Ruben Verborgh

Part 2 – Module 4: Decentralization and federation

Linked Data for Librarians

by Seth van Hooland and Ruben Verborgh

Part 2 – Module 4: Decentralization and federation

Institute of Museum and Library Services Drexel University College of Computing & Informatics

Why are we publishing
Linked Data on the Web?

Why are we publishing
Linked Data on the Web?

Why are we publishing
Linked Data on the Web?

Linked Data is decentralized.
There’s no single source of truth.

[screenshot of “The Night Watch” on the Europeana website]
[screenshot of “The Night Watch” on Google]

Aggregators merge multiple collections
into a single centralized view.

Are we publishing Linked Data
only for the happy few?

Building better clients starts
with building better servers.

Web APIs essentially need
what Linked Data did for data.

Evaluation a query over a federation of sources shows the power of Linked Data.

Linked Data supports federation well,
but not all Web APIs are ready.

Self-assessment 1: decentralization

How does Linked Data help decentralization?

  1. It has a flexible data model.
    • No; although true, this is not directly related to decentralization.
  2. It uses URIs.
    • Yes: URIs help establish shared concepts across sources.
  3. It does not require a single source of truth.
    • Yes: this allows multiple sources to contain partial data.

Self-assessment 2: aggregators

What is a possible role for aggregators?

  1. They should be the sole publishers of metadata.
    • No, this would give several problems, such as update frequency.
  2. Aggregators can facilitate exploration across datasets.
    • Yes, since they have knowledge about how different datasets connect.
  3. They can guarantee metadata quality.
    • No, automated aggregation comes with quality challenges.

Self-assessment 3: Web APIs

What is a problem with current Web APIs?

  1. They often are highly custom implementations.
    • Yes, this makes integration costly.
  2. They do not expose all data.
    • No, this is not a general problem.
  3. They only support specific use cases.
    • Yes, APIs are often tailored to specific scenarios.

Self-assessment 4: federation

Federated querying means:

  1. evaluating a query over multiple sources
    • Yes, this is the definition.
  2. evaluating a query on the Web
    • No, non-federated queries can also happen on the Web.
  3. evaluating a long-running query
    • No, while federated queries often take more time, it is not an inherent part of federation.