Currently WDS only accepts one relevancy training rating per query, meaning if a user rates a particular answer and some time later, they rate it again, their earlier rating gets overwritten. This works for an operating model where only one person or group, e.g. an SME or data scientist manages the training of WDS, but it means that we can't allow users to do the training themselves as each person would be overwriting someone elses training.
To account for this, I suggest a mode in WDS that allows relevancy training to be weighted so that each additional rating can either incrementally increase or decrease the value of the rating applied.
Why is it useful?
|Who would benefit from this IDEA?|
How should it work?