There have been a lot of improvements to tracker since endlessm first looked at it.
Many of the improvements that have been discussed for Tracker3 have been around robustness, sandboxing custom ontologies etc.
One area that hasn’t been talked about much is the possibility of doing graphql quieries on remote endpoints.
E.g from garnacho - tracker3 sparql --remote-service https://query.wikidata.org/sparql -q “select ?item { ?item wdt:P31 wd:Q146 }”
Theres a lot of interesting stuff that can be done through tracker now that would benefit e.g an ostree based os.
Search syndication across containers being one that immediately jumps to mind, making use of stuff like sourcegraph and its graphql api to do complex search quieries from host for code that is hosted in multiple containers locally or even remotely.
Another area which could be interesting is integration with the endless encyclopedia so that you could perform search quieries on remote or local data and maybe formulate more complex queries from multiple datasets.
E.g show me musicians(wikidata) who share a birthday with Daniel(contacts)