When performing feature extraction, it can be advantageous to log many features in one request. As expected, with a large number of requests, the number of features computed times the number of results dominates the server-side performance. But then for a large number of results (100s or 1000s) 10s or 100s of ms can be spent on receiving the data. Every byte counts
If I'm just getting a doc id and the [features] back, I believe the order in the [features] value corresponds to the order I stored the features (as that's a flat list). Is there an option instead of title=12.34,body=5.12,recency=251.1 To just get this back without the labels? 12.34,5.12,251.1 I noticed dense vs sparse as an option here https://solr.apache.org/guide/solr/latest/query-guide/learning-to-rank.html This helps shave a tad bit of time, but it seems we could do a bit better eliminating the feature labels? Is there some way to do this I'm missing? Thanks! -Doug
