You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As requested by TACT members: this is a tracking issue that subsumes many, including new vs old confidence scoring, and new novelty factor scores, analysis of grouping. ARS calls Appraiser which calls other scoring systems as in slide 2 of O&O flow diagram.
It is actually a showstopper issue until the scores flow with reasonable certainty, guessing 90%.
At the moment, there are two suspects for slowness - networking issues that affect the Appraiser-ARS link, and PubMed Reference performance in returning recency results to Novelty scoring.
The text was updated successfully, but these errors were encountered:
Answer Appraiser has been updated to disable novelty scoring, as well as other performance enhancements. It is now returning the other scores within seconds instead of minutes. I say we run a few tests, but otherwise this is no longer an issue.
Just a note regarding the Publication Metadata API -- it was designed to meet specs provided by the UI team and therefore may not be suitable for the recency filter calculations if the issue is speed. A dramatic speedup would likely require a different underlying architecture (probably in-memory) and trimming down the returned data to only include only PMID and date published.
As requested by TACT members: this is a tracking issue that subsumes many, including new vs old confidence scoring, and new novelty factor scores, analysis of grouping. ARS calls Appraiser which calls other scoring systems as in slide 2 of O&O flow diagram.
It is actually a showstopper issue until the scores flow with reasonable certainty, guessing 90%.
At the moment, there are two suspects for slowness - networking issues that affect the Appraiser-ARS link, and PubMed Reference performance in returning recency results to Novelty scoring.
The text was updated successfully, but these errors were encountered: