Ryan LeveringFeb 7, 2024

There are many days that I get disheartened about graphs being too complex to use as an interchange format. My world (structured data at Google) revolves around graphs but humans just have a lot of problems reasoning about higher dimensional data. Sometimes I think trees are about as far as the average person can handle intuitively.

I can easily point to all the advantages in composability, lack of redundancy, etc. but if you can't picture it easily in your mind it makes it hard to work with.

Show threadktk
@rrlevering It's not just that humans struggle with this; even the best-designed user interfaces are tree-based at their core. The real trick is to map the crucial segments of a graph for user interaction onto a tree structure without cutting the links to the original graph. This approach simplifies use and manipulation in a tree format before the changes are seamlessly integrated back into the graph.

I wrote about some experiments we do here, a more detailed version of this & some code will be released soon-ish https://www.linkedin.com/pulse/transforming-knowledge-graph-based-information-adrian-gschwend-ue3le/
Transforming and Visualizing Knowledge Graph-Based Information

At Zazuko, we're deeply invested in understanding how individuals prefer to interact with knowledge graph-based information. While viewing data "as a graph" might resonate with those in the RDF/graph sphere, it's probably not the ideal approach for the majority of enterprise users.

Feb 8, 2024 at 3:34pmWeb