User:Jean-Frédéric/WLMint input

Tech

About me

• fell into the cauldron of French monuments in 2009
  • kickstarting the related wiki-project on Wikimedia Commons
• spearheaded the inception of WLM in France in 2011
• Joined international team in July 2016 (Commons:Wiki_Loves_Monuments_2017/Organizers)
  • Involved in the Jury tools evaluation

• Adopted Monuments_database & ErfgoedBot in 2015
  • Daily uploads in 2016
• Adopted wikiloves stats tool from the pt community in 2017, after expanding support to Wiki Loves Africa

About the Monuments Database

See 2017 presentation: File:Wiki_Loves_Monuments_and_Wikidata_-_WikidataCon_2017.pdf

What worked well

• scaled fairly well (today: 117 datasets, 1.7 million monuments)
• worked on (!= for) all countries
• helpful abstraction layer
  • able to translate/reduce different data models (ie lists)
  • building block for other tools (API + DB)

What did not work so well

• Some assumptions did not always hold very well
  • Monuments have an ID (unless they don’t)
  • Lists might be duplicated between countries
• Data was reduced
• No support for sources

Patterns

• Flexible & agnostic tools
  • works for all countries in WLM ;
  • works for all Wiki Loves X

→ MonumentsDB is mostly for WLM, but also had WLPA

→ Email-notif is agnostic

→ Wikiloves was WLM/WLE, extended for WLPA, WSC, WLF, WLA…

• FLOSS-license, developed in the open, collaboration-enabled

• Run things on Wikimedia Cloud
  • Policies in place to protect tools (Abandoned tool policy, Right to fork policy)
  • You just have to count the dead tools / domains (see eg http://wikiwakacje.pl/mapa/ or 2014 tools)

• It’s dangerous to go alone: get a co-maintainers

Anti-patterns

• Duplication of effort (!= reinventing the wheel)
• operated outside of the Wikimedia Cloud

Cautionary tales

• Jury tools - Commons:Jury_tools & Commons:Jury_tools/WLM_jury_tool_requirements
• Stats tools - Commons:Wiki_Loves_Monuments_2014/Tools