Sonal

SQL based blocking and distance functions

4

What if we could take sql from say a dbt model or otherwise and use that for our model training - blocking as well as similarity? Then non Java programmers...

Right now there is a lot of repeat code in the labeller and update labeller classes - for ex execute method could be in one place. Code is pretty hard...

make the interactive learner color coded

1

It will improve readability quite a bit if - headers were bold - options yes/no etc were color coded

We can build a cli that can filter and show results to the user, much like the labeller

Running training when dataset only has matches/non matches or limited samples throws errors. We should instead inform the user about this so they can add training samples.

3

Reported by Luke from Databricks [zingg_Dec21_0823_log4j-active (1).txt](https://github.com/zinggAI/zingg/files/7761163/zingg_Dec21_0823_log4j-active.1.txt) [zingg_Dec21_0823_sdtderr.txt](https://github.com/zinggAI/zingg/files/7761167/zingg_Dec21_0823_sdtderr.txt)

We have to figure out an error code/error and information reporting framework so that jobs orchestrated through dags like Airflow etc can be handled gracefully. Check best practises here.

Online Arguments creator through JSON Editor?

1

Explore if we can provide an easy way to build the json args - can we use something like https://github.com/json-editor/json-editor ?

descriptive text and word vector/n grams kind of data may need different kinds of blocking. evaluate this

putting this out there as a thought,

Now that the documents are in a better consumable shape, we should add the case studies