Hello!
I was wondering if there can be cases where the marked words occur more than once in the context paragraphs. This question arose from the example that was provided in the competition overview where the words 'population' and 'people' occur twice (context 1). However, the same example is present in the trial data but the second 'population' is missing and the sentence containing the second 'people' is also missing.
Example Context (from overview):
Disease also kills off a lot of the gazelle population. There are many people and domesticated animals that come onto their land. If they pick up a disease from one of these domesticated species they may not be able to fight it off and die. Also, a big reason for the decline of this gazelle population is habitat destruction. People go out and cut down the branches of the trees that these gazelles need to feed from.
Example Context (from trial data):
Disease also kills off a lot of the gazelle population. There are many people and domesticated animals that come onto their land. If they pick up a disease from one of these domesticated species they may not be able to fight it off and die. Also, a big reason for the decline of this gazelle is habitat destruction.
Posted by: kanishka @ Aug. 16, 2019, 4:07 p.m.Hi,
Well spotted :)
I manually modified that context (and the woman-task one) for the trial data since that was just some examples to see the format and how the final dataset will look like.
The final dataset won't contain context where the word appears more than once.
That's something we didn't control for in our pilots and came up as a lesson learned.
However I guess it would be better if we can get other interesting examples from the pilots where that doesn't happen, we will do that when we add the rest of the languages.
For moment just be reassured you won't find that problem in the final dataset.
Thanks!
Posted by: csantosarmendariz @ Aug. 16, 2019, 4:55 p.m.Perfect, thanks! And thank you for organizing this task :)
Posted by: kanishka @ Aug. 16, 2019, 5:15 p.m.No worries, thanks for looking into it!
Carlos