govdirectory / website Goto Github PK

• citizen's initiatives
• FOI requests #17
• email
• edit Wikidata

It's referred to in both the navbar #42 and the footer #35.

We might need to move the project from Wikidata:Gov_directory to Wikidata:WikiProject_* as the templates seem to have hardcoded the WikiProject part but I'm not sure.

https://www.wikidata.org/wiki/Template:Participants
https://www.wikidata.org/wiki/Template:Ping_project

We should have all current Swedish agencies in the data (not necessarily with all social media accounts). Currently courts and agencies abroad (mostly embassies) are missing.

Example: https://w.wiki/3fCd

It's probably something in the BIND(IF(BOUND(?quality), IF(?quality = wd:Q28378282, True, False), False) AS ?verified) line.

Displaying if accounts are verified is something that could help the user trust the data.

On Wikidata this is recorded with the qualifier has quality (P1552) with the value verified account (Q28378282).

Plausible is a "lightweight and open-source website analytics tool. No cookies and fully compliant with GDPR, CCPA and PECR. Made and hosted in the EU", perhaps we can use that to get some insights in how the website gets uesd?

Example on a real cloud hosted version.

The 404 page should be "smart" meaning that it should be able to recognize if there is a QID in the URL and if so show a usable message and provide a link to Wikidata because the item might have been on govdirectory.org or might one day be there.

Similarly, to how each country page works by turning .safeName into a query path this approach could be used to do so in a loop for each country in sitemap.xml.

I got a notification that this organization is eligible for GitHub Sponsors. Perhaps we should apply when we have something viewable?

Aim: To be more relevant than Wikipedia's first paragraph(and the bonus of being CC0).

Label@en -> " is a " P:basic form of government -> ...

Notes:

• maybe it can be implemented in a single SPARQL query with CONCAT etc?

We should define some threshold for how good the data of a country must be before we publish it on the website. This should then be clearly communicatied on https://www.wikidata.org/wiki/Wikidata:Gov_Directory#Add_your_country so that any volunteer know exactly what they need to do before a country gets added.

It would be nice if it was possible to filter/select all the countries in the European Union and similar.

Possible and interesting things to filter on would be:

• Unions (like the EU)
• Continents (like Africa)
• Alliances (like NATO)
• Geographic selections (like Scandinavia)
• Cultural regions (like the Balkans)

United Kingdom.

Originally posted by @Abbe98 in #19 (comment)

Deadline July 31st.

https://form.typeform.com/to/NwzIwcie

https://github.com/RichardLitt/standard-readme seems like a fairly straightforward spec that we could easily follow, especially when starting early like this.

This might be harder for countries than for individual agencies since many countries have several claims on P31. Sweden for example have four claims. It might be easier to just do a listing similar to what we display on a country page on Wikidata.

Originally posted by @Ainali in #26 (comment)

Similar to countries (in #26) we should also have logic for the department pages.

To make it easier to "start" a country on Wikidata, we could create a template that could be preloaded with a button like this:

<inputbox>
type=create
prefix=Wikidata:Gov Directory/
placeholder=country name
buttonlabel=Add a new country
preload=Wikidata:WikiProject Govdirectory/Country template
</inputbox>

but first we have to create /country template that can be preloaded.

Currently, the website uses the issuer property to retrieve the platform name from the identifier property but often and rightfully the issuer points to the company running the platform. Causing bugs like Facebook being called "Facebook Inc" or TikTok being named ByteDance.

An option to the issuer property is to use P:subproperty of->Q:website username->P:of->"platform", it has the downside that its more expensive to retrieve and that it uses the questionable "of" pattern. Anyway, we would need to add the data for each identifier property.

Make sure best practices are followed.

Look at things like

• general meta tags
• open graph
• twitter cards
• app icons and colors

Can we use the qLabel library to get multilingual values in the interface (similar to how the Wikimania menu works? This could be a step towards full multilinguality.

The UN Classification of the Functions of Government would not only be good for us to classify agencies, but would be great for the Wikidata community to be able to create comparative queries. Let's make a property proposal for it.

We should also have a generic comment somewhere to direct people with general questions to discussions/Wikidata.

Originally posted by @Abbe98 in #10 (comment)

This also came up as an idea after interviewing. We should possibly have separate entry points if someone wants to discuss the data or if someone wants to discuss the govdirectory website.

govdirectory.org will be statically generated whose data could behind Wikidata's. As a user, I want my edits to show up directly on govdirectory.org.

Requirements:

• Do not use SPARQL (to avoid its lag too)
• The UI should not "jump" or behave oddly when new data is pulled in which might happen later than the initial page rendering

Implementation notes:

• Create a static list of Wikidata properties that only updates each time the site is built (using SPARQL/Snowman)
• To avoid a "jumpy" page we might consider adding the new data in an expandable section for "recent additions" or similar which would also make potential vandalism less apparent.

Out of scoop:

• For the initial implementation leave out updated data(just support additions) we need it later on but let's first see how this works.
• Pull only online account properties (descriptions/website etc can be supported later)

• Do a country have very few government agencies?
• Are a country's government agencies well or easily modeled in Wikidata?
• Maybe all government agencies of a country have an exclusive external identifier?

etc

https://en.wikipedia.org/wiki/Category:Government_agencies_by_country

We should provide a schema.org representation of each agency to make it easier for others to reuse the data.

Relevant types:
https://schema.org/Organization
https://schema.org/ContactPoint <- maybe not for accounts but for email/FOI?

Some ideas:

Link social media accounts as schema:sameAs together with WP/WD and official website.
schema:additionalType -> P31
schema:url -> our URL

Other relevant properties:
schema:name
schema:description
schema:alternateName
schema:email
schema:logo

Make a short video, probably shorter than 3 minutes that informes the vision and mission of the project, summarizes what we are currently doing and how people can help.

Some data(countries now, subjects later) is reused in a lot of places especially for the various URLs/"safeNames". Currently, safeNames are duplicated in the countries configurations and in the countries individual queries. Duplicating one value is fine but we also need other values for various future tasks.

We could maybe return the country QIDs in the individual country queries and include them in the countries config SPARQL which we then can use as a map to retrieve the other values(using a template taking a QID).

Anyway I think this should not be solved upstream, as it's a result of Wikidata's data model rather than the nature of RDF.

Our second sprint occurs during Wikimania which have a hackathon on Friday, August 13. Do we want to use that opportunity to get some help from the Wikimedia community?

Check where they are running and what they are doing. Especially look if they use the verified status.

Right now we have them on our concept board, but we should make them publicly available so that anyone who wants to contribute gets a better understanding of what research we have done already.

Related to #55 but implementation can be done separately.

Icons as follows: F Y T I (+6)
Meaning Facebook, YouTube, Twitter, Instagram, and 6 additional ones.

Pesudo-codeish:

$results = query social-accounts <org>
$preview_count := 0
for $account in $results:
  if $account.something == $youtube:
    $preview_count += 1
    <i>...
  else if $account.something2 == $facebook:
    $preview_count += 1
  ...

$additional_accounts_count := len $results - $preview_count
if $additional_accounts_count > 0:
  <i>$additional_accounts_count

Known issue: if an org has two Facebook accounts this would render two Facebook logos.

Perhaps we can even try the brand new yaml based forms?

https://www.youtube.com/watch?v=qQE1BUkf2-s

What kinds of issues do we want templates for?

First guess is:

• Bug report (with options that can sort out content issues nicely)
• Feature request

Let's put together a high level roadmap of what we want to do and in roughly what order we want to it in. We can document it and ROADMAP.md and refer to it from both the readme and contributing file.

so that I can get an insight if there is a story there.

For example, compare how many municipalities in one country that have Twitter accounts with the municipalities in another country.

We have a rough draft that should be added to the layout template. Current design:

@Abbe98 Should this wait until we have something to show on the country and subject pages or should I add it with only the first two links?

At one point we will hit Github Pages 1GB hosting limit so we would need to move, Netifly, AWS S3, etc are good technical chooses. But for ethical and cost reasons it would be neat if we could use WM Cloud with a custom domain (I think Wikidocumentaries does this, although it does not need as much space).

This could also be useful for data dump hosting #24.

As a data journalist/archivist I will ingest and filter the data in my own tooling.

Task:

Generate data dumps using the snowman application/sparql-results+json cache by turning it into CSV or another format more common than SPARQL resultsets.

We want data to be filterable on what type of agency it is. Environmental protection agencies will be the one to use in the demo.

• Figure out how to model this type of data in Wikidata
• Add the data for all countries
  • Give instructions to Wikimedians for Sustainable Development

To enable filtering of what kind of agency it is like we need in #18 we need a way to model it in Wikidata. There are a number of possible solutions.

• Use P101 (field of work) (like this)
• Use P921 (main subject)
• Create a new property based on the UN Classification of the Functions of Government

The last option is probably the best one, and the first one feels like the second best choice.

Generate the CSV from the HTML table on the client to save disk space and bandwidth. Only generate it when the button is clicked.

English speaking so that we have the chance to understand if we are getting sensible results when iterating.

59 sovereign states with English as official language.

When lists are long, like on a country or topic page it would be nice to be able to filter on just a single platform. For example, on the page for Sweden or the page for Environmental protection, I could select to only see Twitter accounts or YouTube accounts.

The results views will have one row for each organization where we want to display some icons to indicate if an organization has an account on one of the more common platforms. We probably want to support this for about 3-5 platforms which we need to pick.

Some canidates:

• Facebook
• YouTube
• Instagram
• LinkedIn
• WeChat
• Telegram
• Twitter
• TikTok

Wikimedia CEE Online Meeting 2021 has a deadline on August 31

We have a rough basic footer we iterated on a bit, add it to the layout.

Mandatory property: object has role unofficial/official

Related Links:
handlingar.se/
https://www.foia.gov/?id=15505805-6876-4fc8-9101-fb45c5ff32d0&type=agency
https://www.wikidata.org/wiki/Property:P8167

We heard in some of our user interviews that it would be good to get some sort of activity indicator on accounts so that it is easier to figure out with whom an online dialogue could get started.

• How to determine any account is active?
• On which platforms can we get info like this?
• How and where should we display this?