Open Repositories 2015 - Fedora Interest Group 3B: Hydra and Others

Learning to use Fedora 4 as a BIBFRAME Linked Data Platform

Slide of 24
seconds left

Introduction 1/24

Introduction

My name is Jeremy Nelson and I am the Metadata and Systems Librarian at Colorado College, a private liberal arts college in Colorado Springs, Colorado.

At 24 slides in 7 minutes pace, please email me any questions at:
jeremy.nelson@coloradocollege.edu
Twitter: @jermnelson

Books plugs

My book, Becoming a Lean Library is being published this fall by Chandos Publishing.

I am currently writing my second book, Mastering Redis, for Packt Publishing with a projected publication date of early 2016.

Traditional Push Processes 6/24

Traditional Push Processes

TL&DR - "We already know what you want and this is our product pitch"

What do we mean by push?

With it's origins in early modern industrial companies, best epitomized by Ford Motor Company, Push is about the ability of large companies and organizations to accurately predict and mold demand of customers and organize their operations around those predictions.

Characteristics:

Top down design
Centralized control
Procedural & Tightly coupled
Marketing-focused

BIBFRAME Works & Instances 9/24

BIBFRAME Works & Instances

Using the Semantic Server has allowed me to experiment with different Linked-Data vocabularies but given my interest on library data, I've been mostly interested in two; schema.org and BIBFRAME, with most of my development work for the past few years focusing on the BIBFRAME, the Library of Congress replacement for MARC.

In the BIBFRAME Datastore, RDF graphs for Works and Instances are broken out into subject graphs that are ingested into Fedora, Fuseki, and then indexed in Elasticsearch. Work is defined as "As an abstract representation of the conceptual essence of a resource". Here is partial example of BIBFRAME Work for Lewis Carroll's Alice in Wonderland:

Work URI	Predicate	Object
http://catalog.coloradocollege.edu/3417443	rdf:type	bf:Text bf:Work
	bf:authorizedAccessPoint	"Carroll, Lewis, 1832-1898. Alice's adventures in wonderland :and Through the looking-glass"
	bf:creator	http://catalog.coloradocollege.edu/3417443person6
	bf:language	http://id.loc.gov/vocabulary/languages/eng

BIBFRAME Instance are

Instance URI	Predicate	Object
http://catalog.coloradocollege.edu/3417443instance13	rdf:type	bf:Instance bf:Monograph
	bf:providerStatement	"New York : Illustrated Editions Co. 19--?"
	bf:instanceOf	http://catalog.coloradocollege.edu/3417443

BIBFRAME Annotations & Authorities 10/24

BIBFRAME Annotations & Authorities

BIBFRAME Annotations are resources that assert additional information for other BIBFRAME Annotations. These annotations may or may not be encoded in the source MARC21.

Annotation URI	Predicate	Object
http://catalog.coloradocollege.edu/3417443CoverArt	rdf:type	bf:CoverArt bf:Annotation
	bf:coverArt	http://catalog.coloradocollege.edu/3417443instance13
	bf:coverArtFor	http://catalog.coloradocollege.edu/3417443instance13

BIBFRAME Authorities are resources that represent concepts or things that have key relationships with other BIBFRAME Resources.

Authority URI	Predicate	Object
http://catalog.coloradocollege.edu/3417443person6	rdf:type	bf:Person
	bf:authorizedAccessPoint	Carroll, Lewis, 1832-1898.

Elasticsearch 12/24

Elasticsearch

For search and access we are using Elasticsearch using modified JSON-Linked Data that is serialized from the BIBFRAME entity stored in Fedora. To link Fedora and Elasticsearch we are using the fedora:uuid as the document id and the Fedora URL.

Here is an Elastic Search search result for a BIBFRAME Work:

{
  "_id": "1d27422d-1cc3-4f65-a405-6fe53b7590f8",
  "_index": "bibframe",
  "_score": 1.3480984,
  "_source": {
    "bf:authorizedAccessPoint": [
      "Mark Twain journal."
    ],
    "bf:title": [
      "Mark Twain journal."
    ],
    "fedora:created": [
      "2015-05-19T19:59:57.326000+00:00"
    ],
    "fedora:hasLocation": [
      "http://localhost:8080/rest/80/8b/59/46/808b5946-e689-41c6-bdbb-592f0cba5fd5"
    ],
    "fedora:lastModified": [
      "2015-05-19T20:00:04.528000+00:00"
    ],
    "fedora:uuid": [
      "1d27422d-1cc3-4f65-a405-6fe53b7590f8"
    ],
    "type": [
      "bf:Work"
    ],
    "work_suggest": {
      "input": [
        "Mark Twain journal."
      ],
      "output": "Mark Twain journal.",
      "payload": {
        "id": "1d27422d-1cc3-4f65-a405-6fe53b7590f8"
      }
    }
  },
  "_type": "Work"
}

Warning! Do not leave your Elasticsearch port open to the wider web. Bad things can happen!

Apache Fuskei 13/24

Apache Fuskei

A proper SPARQL endpoint is essential for working with Fedora as a Linked Data Platform. The current version of the catalog pull platform uses Apache Foundation Fuskei project for SPARQL queries. The primary use of SPARQL is for ingestion and reporting needs for the project.

For example, when ingesting BIBFRAME Works, Titles, Person, and Organization, we are doing a rough deduplication based on the following entity properties: bf:authorizedAccessPoint, bf:identifierValue, bf:classificationNumber, bf:label, bf:titleValue. We use a SPARQL query similar to this:

PREFIX bf: 
PREFIX rdf: 
PREFIX xsd: 
SELECT ?subject 
WHERE {
   ?subject bf:authorizedAccessPoint "Mark Twain journal."^^xsd:string .
   ?subject rdf:type bf:Work .
}

Introduction 1/24

Introduction

Books plugs

Obligatory S.H.I.E.L.D. Reference 2/24

Obligatory Agents of S.H.I.E.L.D. Reference

So when I heard this session's title* …

… my thought went to this television show:

This isn't Yet Another Hydra Presentation®

BUT

I am using a Python module called Falcon for building REST APIs

hmm …but no, too militaristic for my tastes

*variation of a joke overheard at this year's Code4Lib

Language, Frameworks, Meh! 3/24

Language, Frameworks, Meh!

My {insert-programming-whatever} is better than … your {insert-technology} sucks!

Blah

Have you ever had a conversation with a friend or a stranger about your work and you've had to explain what is a:

You know what? Me too …

What do people (maybe) understand?

Catalog Pull Platform 4/24

Catalog Pull Platform

Connecting People and Institutions with Algorithms 5/24

Connecting People

to their own and external institutions

to algorithms that impact their work & their lives

and most importantly to other People

Traditional Push Processes 6/24

Traditional Push Processes

TL&DR - "We already know what you want and this is our product pitch"

What do we mean by push?

Characteristics:

Pulling Requirements 7/24

Pulling Requirements

Semantic Server API to Linked Data Platform 8/24

Semantic Server API to Linked Data Platform

What is?

Decoupling & Abstraction

BIBFRAME Works & Instances 9/24

BIBFRAME Works & Instances

BIBFRAME Annotations & Authorities 10/24

BIBFRAME Annotations & Authorities

Fedora 4 for Named Graphs 11/24

Fedora 4 for Named Graphs

Sources for BIBFRAME RDF Graphs

Elasticsearch 12/24

Elasticsearch

Apache Fuskei 13/24

Apache Fuskei

Falcon REST Services 14/24

Using Falcon for REST Services

BIBFRAME Datastore 15/24

BIBFRAME Datastore

BIBFRAME Datastore Services API 16/24

BIBFRAME Datastore Services API

TIGER Catalog 17/24

TIGER Catalog

http://catalog.coloradocollege.edu

BIBCAT 18/24

BIBCAT

http://bibcat.org

Islandora OpenBadges 19/24

Islandora eBadges

BIBCAT Reporting Module 20/24

BIBCAT Reporting Module

Colorado College Library Website 21/24

Colorado College Library Website

i.e. You're Website IS is your Catalog

Why? Why not us! 22/24

Why? Why not us?

Participating 23/24

Participating

Thank-you 24/24

Thank-you!!