Paper analysis




= motivation, setting up the context

  • explain what’s being done and why
  • how are SE biased, what effect?
  • compares previous research, nothing really focuses on the biases from psychology in the context of search

= hooks for the story, show you know other research

  1. overview over all the work
  2. goes into detail for each of them
  3. ties them together
  4. how to build on this?
  5. Focus on: Health & yes-no questions


  1. describes retrospective survey on biased beliefs in web search
  2. how people switch from yes to uncertain to no before and after a web search
  3. provides reasons

Questions, answers, judgements

  1. crowd-sourcing of SERPs obtained from yes/no Bing queries (= 2% of all queries)
  2. filtered out the non-medical questions
  3. doctors annotated the results as true/false → ground truth
  4. caption + landing page → system truth
  5. → 600 SERPs as a dataset
  • yes more often than no
  • yes ranked higher than no (65%)
  • → system bias

Biased SERPs & Search results

  • more likely to click on yes
  • caption: 2-4x more likely to click on yes
  • more likely to skip “no”