Investigation recommended vs not recommended.
On what data is the investigation "recommended" "not recommended" labels determined? Trying to make sense of all the data and not seeing why some come up as one and not the other.
3 replies
-
Hi Dwaldron - sorry for the delayed response, I've been waiting to confirm an answer for you.
The recommendation is based on our prediction algorithm which is trained off of hundreds of linguistic features. There are a few features that are easier to understand and are contained in the report, such as readability, avg sentence length, phrases per sentence, etc. Many of the linguistic features included in the algorithm are not easy to interpret by humans which is why they are not included in the report, and that is why you may have some troubles seeing the differences when looking at the report. An example of one of these features is n-grams.
It is always worthwhile looking at the report, regardless of the label. The new beta view also surfaces known cheating methods and these will be made even easier to identify when the new dashboard is released in August. -
The issue is I am looking at examples where it says recommended, and examples where it is not and being unsure what to look for in the rest of the data when the names match up and the sentence complexity etc looks very similar. So I am wondering the best way to proceed or what issues were flagged.