Shradha26 commented on issue #12553: URL: https://github.com/apache/lucene/issues/12553#issuecomment-1741079621
Thanks, Mike! > What do you mean by an aggregation group? Is this like counting documents that are either red or blue? Yes, exactly. > Do we need to do the low-level changes to support these new user features? For some of the ideas in here, yes. For example, the idea about support adding data for groups (ordinals) would require us to add some [new behaviours for the taxonomy index](https://github.com/apache/lucene/pull/12337) - especially around updates to the index. I've also mentioned how Faceting implementations currently are not generic in that any improvement to IntTaxonomyFacets has to be reimplemented for FloatTaxonomyFacets. As for the new features, it may be tricky to implement them over the existing Faceting implementation because it would require changing some assumptions. For an aggregation, I think we'll want to implement two concepts - an `accumulator` - where we keep a record of the running value for an aggregation & a `reducer` - which modifies the accumulator. In Lucene's faceting, an accumulator doesn't exist as a first class object; rather it is a feature of a class that performs the aggregation. This makes it difficult to define new aggregations - for example, Lucene's accumulators are defined in each aggregator class (this is any concrete child of `Facets` class) and assume there's only one value per-label, per-aggregation - like sum/count. But if were to implement other aggregations like average or median, we'd want to maintain multiple values per accumulator and then finally collapse them in some manner. Right now, we'd need to implement a new child of the `Facets` class, which doesn't isolate the reduction behaviour or allow generic accumulators to be passed in. > I wonder if we could organize them into a plan with dependencies and priorities. Yes, I think that's a great idea. I'll try to organise the ideas in a dependency graph. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
