3rd panel listing all sentences. User can "label" them by selecting positive examples, rules can be trained based on those examples, user can see which other sentences these rules would trigger, then they can also label those sentences to mark that they are positive or negative examples (thereby increasing the amount of labeled data, one sentence at a time), and can choose to accept rules, reject rules, or "train rules" (i.e. refine them), thereby refining the rule system as they go along
3rd panel listing all sentences. User can "label" them by selecting positive examples, rules can be trained based on those examples, user can see which other sentences these rules would trigger, then they can also label those sentences to mark that they are positive or negative examples (thereby increasing the amount of labeled data, one sentence at a time), and can choose to accept rules, reject rules, or "train rules" (i.e. refine them), thereby refining the rule system as they go along