Why don’t you provide a score on a fixed or floating-point scale for your sentiment?


tl;dr: because we believe it’s a gimmick without much utility.

The proponents of assigning a score claim that there’s a difference between “best food ever eaten” and “good meal”. It might be the case if we just look at one particular review, however:

  • the gradation of the superlatives are highly subjective. Is “wonderful” better than “excellent”? Is “excellent” the same as “great”? Is “good” the same as “pretty good”?
  • the choice of the adjectives is often influenced by the culture of the reviewer. British “could be improved” may mean “horrible” in another culture, and vice versa.
  • with multiple aspects, the results become much harder to read, integrate, and aggregate properly

We believe that increasing the variety of aspects is better aligned with the actual business needs.