Consider how to assess plain English

matt-dray / altcheckr

:sunrise_over_mountains: :white_check_mark: R package: assess image alt text on websites

https://matt-dray.github.io/altcheckr/

Other

7 stars 1 forks source link

Consider how to assess plain English #4

Closed matt-dray closed 4 years ago

matt-dray commented 4 years ago

Currently uses quanteda::textstat_readability() for readability (Flesch's Reading Ease Score), which is probably not the right measure for such short snippets of text.

Also gives results like 121 for 'W3C' and -106 for 'Web Accessibility Initiative'. ¯\_(ツ)_/¯

matt-dray commented 4 years ago

Maybe get the most frequently used n words in the English language and check for them in the alt text.

matt-dray commented 4 years ago

Current solution: Match to Charles Kay Ogden's Basic English. Closed with 2acf54fa578c0d3a5343183f77e85926f8c1a1af