I am working on mismo, a python lib for entity resolution. I am exploring using kuzu for the clustering stage.
In particular, one of the largest single value-adds for mismo would be a better interface in a python notebook environment for visualizing, debugging, and querying the entity graph (assuming that mismo has already done the initial blocking and matching stages of entity resolution). As a python lib for data engineers, the most valuable form factor for mismo's UI would be in the form of jupyter widgets, for display/interaction in notebooks. I already have this half-baked implementation for viewing clusters using ipycytoscape, but it really could be a lot better.
How motivated would you be to massage your code so that it was re-usable from a jupyter-widget environment? I could maybe work on the translation/wrapping code that made your core work within a jupyter-widget, but this core would have to be packaged in some sane way.
There are probably a lot more problems to think about here as well. For example, how hard would it be to switch to the backend being the python kernel instead of your wasm/express backend you have here?
Anyway, just wanted to throw this out there and get a sense for the appetite.
I am working on mismo, a python lib for entity resolution. I am exploring using kuzu for the clustering stage.
In particular, one of the largest single value-adds for mismo would be a better interface in a python notebook environment for visualizing, debugging, and querying the entity graph (assuming that mismo has already done the initial blocking and matching stages of entity resolution). As a python lib for data engineers, the most valuable form factor for mismo's UI would be in the form of jupyter widgets, for display/interaction in notebooks. I already have this half-baked implementation for viewing clusters using ipycytoscape, but it really could be a lot better.
How motivated would you be to massage your code so that it was re-usable from a jupyter-widget environment? I could maybe work on the translation/wrapping code that made your core work within a jupyter-widget, but this core would have to be packaged in some sane way.
There are probably a lot more problems to think about here as well. For example, how hard would it be to switch to the backend being the python kernel instead of your wasm/express backend you have here?
Anyway, just wanted to throw this out there and get a sense for the appetite.