Forgive minor mistakes setting about h5p-essay

[otacke]

Hi @Jarvil!

There are plenty of ways to assert the similarity of two strings. I created the function areSimilar in the TextUtilities library. It computes the so called Damerau-Levenshtein distance which essentially counts the number of changes (delete a character, add a character, exchange a character, swap two adjacent characters) needed to transform one string into the other - those count is considered as the number of mistakes. In consequence, the algorithm is not tailored to a particular language but will do the same for any language.

Also, as always, H5P content types are not tailored to a particular host system. It doesn't matter if you use moodle, WordPress, Drupal or something else - they always work the same.

Cheers,
Oliver

[Jarvil]

Hello,

So it compares the characters in a keyword and how much they wary. And you will get full points even if you have 9 wrong charcaters so number of mistakes doesn't count

[otacke]

Hi @Jarvil!

Yes, you are correct. If someone wrote "desoxyribonucliec acid" instead of "deoxyribonucleic acid" this would still count as a valid answer. The correct answer deoxyribonucleic acid has more than 9 characters and there are just two changes needed:

1. Remove the extra s
2. Swap i and e

In an essay scenario that's "stupidly" based on detection of keywords, it doesn't make much sense to punish spelling IMHO, but the person who wrote "desoxyribonucliec acid" probably meant the correct thing in a larger context and should get full score. It's not a dictation.