COVID-19 Open-Source Helpdesk

Knowledge Graph to fight COVID-19

The “GraphHackers, Let’s Unite to Help Save the World — Graphs4Good 2020” initiative is looking for volunteers to join the effort.

We have set up a project covid-19-community to crowdsource heterogeneous data about COVID-19 and represent the data in a Knowledge Graph.

We use Jupyter Notebooks, the Neo4j Graph Database, and Binder to incubate new data and new analyses of these data to gain new insights into COVID-19. This is an excellent opportunity for data science students to apply their skills.

We need help with a few Jupyter Lab related issues, e.g., a plugin to display and interactively explore property graphs.

4 Likes

This is very exciting. I was thinking along very similar lines.

however, I come with a background in wikidata / wikibase. Do you know if there is a frontend (web-based) for neo4j’s knowledge graph?

I think Wikibase / Wikidata would have a huge potential here and one could do nice visualizations with SPARQL queries.

4 Likes

Additionally it’s easy to set up using docker-compose: https://github.com/wmde/wikibase-docker

We might be able to get some infrastructure from wikimedia for this effort… and they have a huge telegram chat room where we could ask for help.

1 Like

I’m totally in favor to setup up a centralised project where anyone can start open sourcing datasets. In the future can be useful for other areas not only for covid19 stuff.

The Neo4j browser can be used as a front end. Here is an example:
https://diseaseknowledgebase.etriks.org/protein/browser/

Wikibase/Wikidata sounds interesting, too. Is there an example of visualization?

yes! check out the examples here (this will render a map of all hospitals worldwide, but if you click the “Examples” button there are many other examples, some with visualizations).

One can use SPARQL to generate maps, graphs, tables …

https://query.wikidata.org/sparql?query=%23Map%20of%20hospitals %23added%202017-08 %23defaultView%3AMap SELECT%20*%20WHERE%20{ %20%20%3Fitem%20wdt%3AP31%2Fwdt%3AP279*%20wd%3AQ16917%20. %20%20%3Fitem%20wdt%3AP625%20%3Fgeo%20. }

The other good thing about wikibase is that it’s built for large-scale collaboration … but I haven’t ever tested neo4j so can’t really give an informed opinion.

Can you double-check that URL? It doesn’t work for me. -Peter

Sorry!

Here is a list with all examples (note the Try It button below) https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries/examples

Here is another nice example with hospitals in paris:

https://w.wiki/L6d

These links hopefully work :slight_smile:

pinging @jtp , may be able to help on the jupyterlab issues.

1 Like

Thanks @adrinjalali for the ping!

For reference, there is also some discussion happening in this issue: https://github.com/covid-19-net/covid-19-community/issues/1