QuestioningAuthority::LDF
This sofware providies bindings from Questioning Authority to the linked data caching fragment server for fast query of RDF-based authorities.
What Does This Do?
This gem offers a caching version of the Questioning Authority interface relying on the ActiveTriples LDF caching service. A set of linked data "authorities" are exposed through the Questioning Authority API, and cached for fast recall using the Linked Data Fragments Caching Server. Models on the Hydra/Rails side are provided to handle caching and easy presentation of labels.
Both the QA and LDF APIs are available in mountable forms, so they can run directly in your Hydra application---see below for details.
Why use this?
The main reason you may be interested in this software is to handle local caching of linked data references in your metadata.
Upcoming features include:
- Support for
Qa::Authority::Base#all
for all LDF authorities through Hydra Core Partial Collection View style paging. - A default search, accessing only cached items.
How Does it Work?
[TK] Architecture Overview.
Mounting and configuring the LDF caching server
The LDF caching server can run as a mounted application within Rails. To use the caching server, add the following to your Gemfile
:
gem 'qa-ldf', '~>0.3.0'
# for now use the active branch of the linked data caching fragment server
gem 'ld_cache_fragment', github: 'ActiveTriples/linked-data-fragments', branch: 'feature/multi-dataset'
Then you need to mount the application to an unused route:
# config/routes.rb
Rails.application.routes.draw do
# ...
mount LinkedDataFragments::CacheServer::APPLICATION => '/ldcache'
# ...
end
And add an ldf.yml
configuration.
# config/ldf.yml
development:
uri_endpoint: 'http://localhost:3000/ldcache/{?subject}'
uri_root: 'http://localhost:3000/ldcache'
cache_backend:
provider: 'repository'
test: &TEST_
uri_endpoint: 'http://localhost:3000/ldcache/{?subject}'
uri_root: 'http://localhost:3000/ldcache'
cache_backend:
provider: 'repository'
production:
uri_endpoint: 'http://localhost:3000/ldcache/{?subject}'
uri_root: 'http://localhost:3000/ldcache'
cache_backend:
provider: 'marmotta'
You should now be able to access the ldcache endpoint at http://example.com/ldcache
. That address will return a dataset description for the root dataset. Queries sent to that endpoint access the transparent cache. For example:
http://example.com/ldcache/?subject=http://id.loc.gov/authorities/subjects/sh2004002556
will pass the request through to the Library of Congress on the first request, subsequent requests are retrieved from the cached.
Finally, configure Qa::LDF
to aim at the cache server:
Qa::LDF.configure!(endpoint: LinkedDataFragments::Settings.uri_root)
In a Hydra app, the best place to put this is in an initializer, e.g. config/initializers/hydra_config.rb
.
Datasets
The cache server uses RDF Datasets to manage sperate cache spaces. The root dataset is at the server root (http://example.com/ldcache
, above), and additional datasets are available at any dataset/*
path following from the root. The Qa::LDF
authorities are configured to use independent datasets, so each authority corresponds to a cache space of its own. Each dataset provides its own description at its base path, and is queryable with ?subject=
queries as demonstrated above.
Selecting a Triplestore Backend
The caching server provides several backends for persistent storage, we recommend using the Marmotta backend, but a Blazegraph backend (untested with this gem) is also provided. Read the ActiveTriples::LinkedDataFragments documentation on backend configuration for more.
LDF caching as an external service.
The LDF caching server can run independently from your Hydra application as a lightweight, generic Rack application, or as a standalone Rails app. You may want to deploy the server in this way so it can run on separate hardware, or to segregate Linked Data "follow your nose" network traffic.
The simplest way to get a working application that will run on any Rack-compatible server is to create a ./config.ru
containing:
# ./config.ru
require 'ld_cache_fragment/cache_server'
run LinkedDataFragments::CacheServer::APPLICATION
With this in your working directory, you can run $ rackup
to launch a basic server. More robust deployments are possible using servers like Puma.
[TK] LDF caching as a Rails app
Authority Sources
LCNAF
The Library of Congress Name Authority File
FAST
Faceted Application of Subject Terminology
Configuring authorities
[TK] Models.
Forms
Autocomplete handling for Questioning Authority is provided by Sufia/Hyrax. To this handling, we add a custom dropdown for selecting from multiple authorities in a field.
Hyrax Autocomplete
You've got a Work
in Hyrax and it has a Creator
field. Wouldn't it be nice to have autocomplete on that form with a controlled vocabulary? Hyrax has this functionality, but you need to set some things up first.
The most basic way a field is associated with a particular autocomplete data source is with a data attribute on the element. The data-autocomplete-url
attribute stores a path to the data source. To add one to an existing field create a partial for the field in app/views/records/edit_fields/
.
In the partial, specify the data source:
<%=
f.input key,
as: :multi_value,
input_html: {
class: 'form-control',
data: { 'autocomplete-url' => "/authorities/search/loc/names",
'autocomplete' => key }
} ,
required: f.object.required?(key) %>
Hyrax comes with Questioning Authority which provides some RESTful endpoints that you can use as autocomplete sources. Checkout the QA README for a list of the authorities that it comes with (you can even make your own).
When the Hyrax form for your Work
loads the autocomplete JavaScript is initialized for certain fields. For the Creator
field that we created a partial for, all we needed to was add the data attribute and the autocomplete is activated.
If the field is not one of the default fields, you'll need to run this to activate autocomplete on the field:
var Autocomplete = require('hyrax/autocomplete');
var autocomplete = new Autocomplete({
"autocompleteFields":
["subject","language","creator","based_near","work"]
});
$('.multi_value.form-group').manage_fields({
add: function(e, element) {
autocomplete.fieldAdded(element)
}
});
autocomplete.setup();
You just need to pass the fields when creating an Autocomplete instance.
Autocomplete in Hyrax currently uses jQuery UI Autocomplete. Hyrax stores the jQuery UI Autocomplete options and source in ES6 classes. Unless you need to make specific query that requires you to change the options and source, the JS will use a default query from default.es6. This should just work with the QA vocabs.
The autocomplete for the Work
field requires different a different source and options so it has a different class: work.es6.
After creating your own class, you will need to import it and an additional case to the autocomplete method in autocomplete.es6
Adding Controlled Vocabulary Dropdown Options
To create your own dropdown with authorities you extend the QASelectService
class. By convention, we place this in app/services
.
# app/services/name_authorities.rb
class NameAuthorities < QaSelectService
def initialize
super('names')
end
end
Initialize the service with the name of a YAML file names.yml
that contains a list of authorities:
terms:
- id: /authorities/search/loc/names
term: LOC Names
active: true
- id: /authorities/search/assign_fast/all
term: FAST
active: true
Then in the partials for the input field at app/views/records/edit_fields/_creator.html.erb
:
<% name_authorities = Hyrax::NameAuthorities.new %>
<%=
f.input key,
as: :multi_value,
input_html: {
class: 'form-control',
data: { 'autocomplete-url' => name_authorities.select_active_options.first.last,
'autocomplete' => key } } ,
required: f.object.required?(key) %>
<%= f.input key, collection: name_authorities.select_active_options, label: false %>
Implementing Custom Authorities
Custom (local) authorities can be included by placing a YAML configuration in config/authorities
. This configuration should specific a list of terms, and give an id
, term
name, and active
flag for each of them. Terms can be set inactive
to disable them.
# config/authorities/moomin_valley.yml
terms:
- id: http://example.com/moomin
term: 'Moomin'
active: true
- id: http://example.com/moomin_papa
term: 'Moomin Papa'
active: true
- id: http://example.com/moomin_mama
term: 'Moomin Mama'
active: true
- id: http://example.com/snorkmaiden
term: 'Snorkmaiden'
active: true
These terms are now available from Questioning Authority at `/authorities/search/local/moomin_valley'.
License
This software is available under the Apache 2.0 license.