The idea is that we could say some rows are more important than others.
So for example:
text,isnegative,importance
You suck,1,0.9
This kind of seems bad,1,0.5
The importance column could give the model a "score" on how important this row is. We might score some rows as being "worse" or "more important" in creation of the model.
Context: https://docs.microsoft.com/en-us/dotnet/api/microsoft.ml.trainers.linearsvmtrainer.options.exampleweightcolumnname?view=ml-dotnet
The idea is that we could say some rows are more important than others.
So for example:
The
importance
column could give the model a "score" on how important this row is. We might score some rows as being "worse" or "more important" in creation of the model.