Ask the Last Person Who Asked

Recipient-Side Evidence for Expert Finding on an Organisational Question Board

A paper from the School of Underlying Conditions adds recipient-side evidence to a document model of expertise, crediting the colleague who asked a question and accepted an answer as well as the colleague who wrote it, and evaluates the model on a geoscience consultancy's internal question board of 18,240 questions. The recipient-side model performs worst on offline replay, but in a fourteen-week randomised field trial on 1,362 new questions a blend of the two signals raised the accepted-answer rate from 68.7% to 73.6% and cut the median wait for a first answer from over three hours to under one. The usefulness of a received answer halves in about nineteen weeks, and answers relayed more than once lose fidelity to the original.

@misc{slop_q7hn3v,
  author       = {Ragnhild Uzoma and Vigdis Emenike},
  title        = {Ask the Last Person Who Asked: Recipient-Side Evidence for Expert Finding on an Organisational Question Board},
  year         = {2026},
  publisher    = {Slop University},
  doi          = {10.5555/slop.q7hn3v},
  url          = {https://slop.university/outputs/slop-paper-routing-a-question-q7hn3v/},
  version      = {1.0},
  note         = {Research paper},
}