I have a task where I have a context, then a question about that context, and multiple options, and instead of picking just one answer, I need to rank the options from most to least plausible. What’s the right ontology for this? And what type of project should I create for that ontology?
Hey @pedram ,
I would recommend using a classification to do the ranking, if you have n elements just have n number of classification.
Regardless if the n is variable across your data rows, you can instruct the labeler to leave it blank if not required.
If you are planning to use the Multi modal chat editor we have a dedicated ranking tool to help.
@PT That’s what I was thinking. The only thing is that the dragging feature for ranking in the chat evaluation is a bit more user-friendly and avoids creating multiple (n) classification objects. Is there any way we can set up a custom sample in the editor and then use the ranking feature for that custom editor? For example, create an HTML sample in my editor like the following:
Context: this is a sample context
Question: this is a question about context
Options:
A: option A
B: option B
C: option C
D: option D
and then have a ranking feature to rank options A to D based on a criterion we define (e.g., most plausible to least plausible).