Consider using Jupyter notebooks for some series of connected queries

[Daniel-Mietchen]

i.e. feeding the results of a query into later queries, saving intermediate results locally.

The WDQS allows that to some extent by way of subqueries, but their complexity is limited by the timeout threshold.

Example chain:

• get most popular topics of scientific papers
• for the top ones, check the most cited papers
• aggregate across
  • institutions
    • educational
    • research
• funders
• years
• prizes
• locations
• academic fields

etc.

[Daniel-Mietchen]

Here's a discussion re embedding Jupyter notebooks: https://groups.google.com/forum/#!msg/jupyter/P9j1GZCgZt0/kJmY6oUuBwAJ

[Daniel-Mietchen]

Here's a blog post about using Jupyter for Wikidata SPARQL queries:

• http://www.snee.com/bobdc.blog/2018/08/pipelining-sparql-queries-in-m.html .
  It cites several related ones, and some of the associated code sits at https://github.com/bobdc/misc/tree/master/JupyterSPARQL .

[fnielsen]

I do not know if people is aware of it, but we have another example here: https://github.com/fnielsen/scholia/blob/master/notebooks/investigation-of-text-to-q-text-setup.ipynb

[Daniel-Mietchen]

Yeah, I should have linked to this earlier - thanks!

[Daniel-Mietchen]

Some more on how we could use Jupyter:

• via iframes, much like the current Scholia — example
• via the SPARQL kernel for Jupyter — example
• via Gastrodon — example

[Daniel-Mietchen]

Here's a daily updating Jupyter notebook with Wikimedia analytics:
https://analytics.wikimedia.org/datasets/wikipedia-android-app-reports/suggested-edits-v1.html .

Seen via this tweet

[Daniel-Mietchen]

See also #303 on building sequential queries.