1 of 30

History, Present & Future of�WBStack & WBaas

Adam Shorland�@addshore

2 of 30

Who am I?

I have been involved in Wikipedia in some way since 2005.

I joined the Wikidata team back in 2013.

I’m currently the Wikidata & Wikibase Tech Lead.

I am on the board of the MediaWiki Stakeholders group.

I created WBStack.

Twitter: @addshore

Github: addshore

Wikimedia: Addshore

etc...

3 of 30

WBStack at a glance

Open-source & free WBaaS platform launched by me (as a volunteer)

Combines MediaWiki, Wikibase, Blazegraph Query Service, SPARQL UI, Elasticsearch and more within a single platform

Designed for quick start up rather than lots of customisation

Shared infrastructure → Very substantial infrastructure cost savings

4 of 30

How did we get here?

5 of 30

Wikidata (2012+)

6 of 30

The early days (2015-2019)

7 of 30

MWaaS

MWaaS notes started appearing on a train in 2015.

I believed that MediaWiki had great potential for people and organizations.

Between 2013 and 2015 I mainly experienced a collection of people and organizations that would struggle at the early stages of software setup & maintenance, instead of really getting to use the software.

I was dreaming of possible future solutions.

8 of 30

WikiHub

I continued to develop the idea in 2016

Started modeling a Github like structure, for personal and organizational use.

A connection to Wikibase still hadn’t been made, instead focusing on MediaWiki

9 of 30

Prototype thoughts

In 2017 I started thinking about how to implement something for people to see.

This would ultimately aim to be Wikimedia services and ecosystem, but for non Wikimedia specific use cases.

Wikibase first appeared in the notes.

I created the first git repo / code.

10 of 30

Federated Workshops

In 2018 we had a round of Wikibase workshops supported by the ERC (European Research Council)

I started focusing on the idea of WBaaS as a focus, rather than MediaWiki

More impactful for Wikibase, as the setup and maintenance is even harder.

Code switched from mwaas -> wbaas & prototyping began.

11 of 30

Proof of concept (WikWiki)

The end of 2018 saw an initial working proof of concept.

Shown to a small audience the interest continued to grow.

Similar architecture to what currently exists today.

12 of 30

WikWiki -> OpenCura

In 2019 I took a sabbatical & created most of what can be seen on WBStack today.

This sabbatical ended just before Wikidatacon 2019, so that was the next step!

13 of 30

WBStack onwards (2019+)

14 of 30

Wikidatacon 2019

Shown at Wikidatacon 2019 in its first version.

Demoed as 0 to Wikibase in under a minute.

Invitation codes started being given out.

Maintained solely by me in my spare time.

Infrastructure costs covered by Rhizome from the start.

15 of 30

2019 continued

Service modifications

  • Shape Expressions
  • Query Service Examples
  • Speed improvements
  • Stability improvements

Discussions around the future with WMDE

16 of 30

2020

Service modifications by me

  • MW&WB 1.33 -> 1.35
  • Wikibase Lexeme
  • More extensions, eg. Scritunto, Advanced Search, Echo & 18+ more
  • Dashboard for modifying site settings & more customizations
  • Cradle & Widar
  • Under the hood changes

Open Source all code & Infrastructure including docs

MVP for Wikibase as a service platform” landed on the WMDE development plan

“OpenCura has proven to make it easier for people to spin up a working Wikibase instance in minutes. We need to continue working on it, clarify its direction and build it out.”

17 of 30

2021

Service modifications by WMDE

  • Federated Properties initial version
  • Wikidata entity mapping
  • Wikibase Manifest
  • Elasticsearch

Fresh documentation of the infrastructure

Wikibase evaluation service (Wikibase as a Service MVP): Development” landed on the WMDE development plan

�We will build the Wikibase evaluation service, a way to instantly create a temporary Wikibase sandbox that will enable users to more quickly and easily evaluate if the software is a fit for their project. This is the MVP version of "Wikibase as a Service". The goal of this MVP is to learn about the technical requirements and challenges of hosted Wikibases so that we can make decisions around a long-term strategy.

18 of 30

2021 continued

WMDE has taken over the running costs of the WBStack infrastructure from Rhizome

WMDE has performed a deep dive into the future of WBaaS over the past 3 months (March - May)

WMDE is working on releasing an updated strategy for Wikibase around WBaaS.

WMDE has worked on WBStack code in a few different settings in 2021, and more resources are allocated for further work this year.

19 of 30

Usage

20 of 30

WBStack users

WBStack has ~100 unique users

15+ unique institutions has WBStack accounts (including 8 libraries)

12+ research teams are represented

13+ other organizations including small companies and non profit orgs

21 of 30

Differentiated by ease of use

WBstack makes Wikibase more accessible to those who will care for the data

Non-technical specialists can focus on getting the Wikibase going and worrying about the data, instead of worrying about running the software and keeping it going through time.

22 of 30

Example Sites

23 of 30

Sandboxes

  • While throwaway sandboxes are quickly deleted/abandoned, some sandboxes can be used for months or even years as places to experiment, learn, and innovate.
    • Research Ecosystem Sandbox (University of Oviedo)
    • LexBib (University of Basque)

24 of 30

Pilots/Proofs of Concept

  • Starting with a subset of a larger dataset or smaller scope, the goal is to prove Wikibase is the right fit for a particular project

25 of 30

“Permanent” Instances

  • WBStack is becoming established as research data infrastructure and is the permanent home to several Wikibases

26 of 30

The Future

27 of 30

Vision

WBaaS empowers research teams and institutions to co-create the world’s largest, publicly available knowledge graph.

Demolishing data silos, making it easier than ever to create, connect, and grow your own linked data

28 of 30

Invitations

I’ll be handing out another round of invitation codes today.

If you would like one please message me after this presentation.

Email or Twitter preferred.

29 of 30

Reading & Links

30 of 30

Questions?