some kind of comparison suggestion

[cheapsteak]

check out https://www.npmjs.com/package/sqlite

maintained a list of packages and then a pairings table that recorded the number of times a pair of packages were searched for together you could query the pairing table for things searched for together and take any that turn up with a high enough pairing factor

much thanks to @Haegin

For instances of when (A, B, C) are already being compared, and someone adds E, bump the pairing factors for A:E, B:E, C:E

Separate db file to store raw event logs for potential future processing

For future (out of scope for now)

• can potentially use derived columns to calculate the number of times the two items were paired up out of the total pairs for each item
• weigh recent entries higher than older ones

---

Also worth investigating (thanks to @kieran) Dynamic Synonyms and Search Analytics with ES; the specific plugin seems to be ruby/rails, but I assume it's an ES thing and not a ruby thing

dynamic reindexing seems potentially problematic; not necessary, will want an event log anyway, hourly (or even weekly) reindexing should be fine