0.9 Bayes adjustment
Philosophy
Everyone knows that Oscar is a grouch, we appreciate him for the surly fellow he is accept his curmudgeonly ways. moreso than the world-weary Bert, or the free-spirited Earnie.
o assume that we incorporate his feedback in a slightly differently fashion
Naturally you would take the words of Eric Cartman with a grain of salt,
____________________________
Accounting for tendencies of each individual user to be more or less generous in their critiques relative to their peers, our system adjusts the relative weight of each review per bayesian inference. Meaning that exceptional reviews count for more when coming from a typically reserved user versus a more lavish one. It also means that critical reviews will count for less when coming from bearish users, rather than bullish ones.
Addressing Bias
_________________________
State of the Art:
Simplistic review and rating systems (ie yelp, airbnb, amazon product, etc) fail to take into account the predilections and biases of reviewers; giving equal weight to the opinion of any reviewer who cares to make public their opinion. Those reviewers who are most vocal, or most extreme with their opinions influence the overall rating of a product/service far more than those users without extreme opinions or sense of self importance. By design, those simplistic reviewing/rating systems show preference to the most opinionated users, and those prone to making extreme or exaggerated claims.
Our solution: Bias adjustment
User reviews are weighted as per expected behaviour using a bayesian logic prior to being incorporated into a person/product/service’s rating. When past reviews indicate a bias, their trend toward bearish or bullishness is mitigated by an adjustment to the weight of their review.
EG: a person who consistently rates their experiences lower when the issuer has trait X, that review will carry less weight.
Justification: -people aint perfect, we neither need nor expect them to be. What we do expect, and all we ask, is for people to show up willing to engage with one another. This cannot be stressed enough: who you are, as you are, is enough. Who you are now is exactly the sort of person we want, we need, to make this thing work. You, with your flaws, your shortcomings, we can adjust for them, you’re welcome here. You dont have to be perfect, you dont have to do everything right.
____________________
The relative weight of a members review is adjusted per Bayesian inference. -Meaning that exceptional reviews count for more when coming from a typically reserved member versus a more lavish one. It also means that critical reviews will count for less when coming from bearish members, rather than bullish ones.
Examples
lorem
Last updated
