I’m not sure if this worked in previous versions or whether it’s supposed to work this way but I’ve noticed that it seems that the “Add Candidate” feature requires at least a whole word. What I would like to is to be able to select a sub-word and add it as a class candidate.
For example, suppose I had “users” in my text and I want to add “user”, which is a sub-word of “users”. Currently, the textual analysis UI allows me to select “users” and more and add it as a class candidate. However, if I select “user” so that the selection looks like [user]s, then I can’t seem to add this selection as a class candidate.
This is because textual analysis supports extracting candidate in word-base not character base. In other words you must select a complete word “users” to be a candidate (or several words), rather than part of a word. If you want to add “user”, please click on Add Candidate Class from the toolbar, select Add text as [type], then enter the word (e.g. “user”), and confirm.
In that case, is there a way to map an existing candidate class to a substring of text?
By mapping, do you mean that if you have extracted candidate class “users” (with ‘s’), you want when you select “users” in the candidate item pane, the word “user” will be identified and be highlighted along with “users” within the passage? (In short, you want a way to associate “users” with “user”, making textual analysis “knows” that “users” is equivalent to “user”?)
What you just mentioned is actually better than what I had originally thought but yes, that’s roughly what I was thinking of.
Sounds good. It should work like… to let user define alias for a candidate item. I will forward our conversation to our developers for further studies.