Mojo inside Exasol database for User Defined Functions (UDF) - as a Script Languge Container

Mojo
1

Hello out here,

I have created a Mojo Script Language Container for the Exasol database for
creating fast(er) User Defined Functions. (UDF). You can find it here:

Exasol is available as a test bed for free on the Exasol homepage (Exasol Personal) or on
Docker Hub (Exasol Nano). Have fun playing around with it. Happy to receive feedback…

A blog article is in preparation…

2

Implementing Mojo inside Exasol sounds pretty intresting for someone interested on GeoSpatial Computing. I read you GitHub repo description and found some stuff that seem to align perfectly with someone on my grind. Spatial clustering groups geographic locations based on proximity. In relational databases, this is often done using a SET UDF (a custom aggregation that processes a group of rows and yields cluster assignments-if you’ve done this before :sweat_smile: ).

Common algorithms that are pretty good are none other than DBSCAN and K Medoids…

So my insterest is blatantly DBSCAN (Density-Based Spatial Clustering of Applications with Noise)

where Groups points that are packed closely together within a distance threshold (\\epsilon) and flags points in low-density regions so Mojo might be interesting for this. DBSCAN requires repeated distance evaluations between points. Running this in pure SQL requires expensive self-joins. Inside a Mojo SET UDF, points in a group are collected in memory, and Mojo’s native speed processes the spatial distance matrix in parallel across Exasol nodes. as for K medoids i havent experimented on that. i wonder if youve done any

3

No, I have not done any computation / use case you are describing. But it sounds interesting though…