Wikidata Queries and AI LLMs: better dataviz with ChatGPT4O

Paintings owned by family and business connections to Presidents of the Swiss Art Dealers Association

The above graph, generated by a Wikidata query, can be made easier to understand with help from AI.

Steps:

1. Run Wikidata query: 

Short link to Wikidata Query: 

https://w.wiki/Cdse

2. Download  CSV from Wikidata Query 

3.  Upload into ChatGPT4o with the following prompt:

"please graph - pay attention to edges"

Here is a graph representing the relationships between Swiss Art Dealer Association Presidents, their business or family connections, and associated paintings. The edges are labeled to indicate the type of connection, and node colors differentiate between SwissArtpres (blue), connections (green), and paintings (orange). Let me know if you'd like further adjustments

chatGPT40 : "Here is the updated graph with adjusted label positions to minimize overlap. Labels are placed with a background for clarity, and the spacing between nodes has been increased to make the connections easier to read. Let me know if you need further refinements! "​

Art Market Network Analysis with Wikidata Sparql Queries and Beyond

What might replicable data pipeline from #Wikidata #KG Query to Data Frame to Network Visualisation of owners of artworks passing through a specific network look like?
Perls Family Network-Red
Links to other owners-Blue
Jupyter Notebook in Google CoLabs
👉https://t.co/2oyU0kmuqi pic.twitter.com/UdQhY8ZKVp

— Open Art Data | @openartdata.bsky.social (@OpenLinkArtData) August 21, 2024

 

What might replicable data pipeline from #Wikidata #KG Query to Data Frame to Network Visualisation of owners of artworks passing through a specific network look like?

In the example below, we look at 27 artworks that passed through one of the members of the Perls art dealing dynasty or one of their companies.

The starting point is a Wikdata Query to retrieve the artworks known to have been owned by one of the Perls family, as well as the other known owners of the same artworks.

The information is retrieved from Wikidata, loaded into a Data Frame, then visualised with MatPlotLib.

The code is saved in a Jupyter Notebook and Shared publicly via Google Colab.

Anyone with a Google Account should be able to run the code simply by clicking on the arrows to the left of each code cell.

Try it and let me know if it works.

https://colab.research.google.com/drive/1f7V2SMzxkCmt2lbCS3l4ulotqUGghNAm#scrollTo=0sPcg-gWZOo3

Perls Family Network-Red

Links to other owners-Blue

Jupyter Notebook in Google CoLabs

https://colab.research.google.com/drive/1f7V2SMzxkCmt2lbCS3l4ulotqUGghNAm#scrollTo=0sPcg-gWZOo3

Tracking Looted Art with Knowledge Graphs: A Wikidata Case Study

Art looting networks operate on many levels, many of them hidden, over long periods of time. The native graph function of Wikidata enhanced by federated queries can help track them.

Tracking Looted Art with Graphs

April 9, 2022, Laurel Zuckerman

Graphs and Networks in the Humanities 2022 Technologies, Models, Analyses, and Visualizations

6th International Conference, 3. – 4. February 2022, Online

The 6th international conference on Graphs and Networks in the Humanities took place from Thursday 3. February to Friday 4. February 2022 online, co-organized by scholars from the Huygens Institute (Royal Netherlands Academy of Arts and Sciences), the Academy of Sciences and Literature | Mainz, Vienna University, University of Leipzig, and the University Ca’ Foscari Venice

Paper: Tracking Looted Art with Graphs: A Case Study 

See also:

The Error is the Message: Extracting Insights from Deceptive Data for Nazi looted art

The Knowledge Graph Conference, 2023

VIDEO: 

https://youtu.be/WBMpZ3NDNRQ?si=wsFtV9wzBEghCSoB