Ask HN: Best way to approach entity resolution

Hello HN! We're currently working on creating a domain-specific directory of businesses and their properties, and we are running into issues keeping track of the various sources of data that we use to build the directory, especially manual edits we start doing ourselves (which is the primary value addition in some ways).

We are also trying to come up with models that can resolve these entities at various steps of the pipeline, and while the system works, I was curious if there's any prior infrastructure/architecture design info available (or even any packages for that matter) that can help with this process (which is to track how our pipeline evolves, and to be able to trace the history of why an entity was tagged as it was in the end).

We're essentially trying to tackle to the same problems Google Maps and Yelp have solved, but I'm curious if we have to lay out the groundwork from scratch or if there are any resources out there that I might not be aware of.