This is a demo of a text classifier which categorises user-input statements as a problem, a remedy, an outcome, or other.
Simply type an answer to the question, "and what would you like to have happen?", into the box and click analyze.
This is useful particularly in Clean Language Coaching, but has other applications such as customer service. It uses OpenAI's GPT-4o-mini and few-shot prompting. See their privacy policy for details.
If your results seem inaccurate, let me know by email, including your input, the result, and what you expected.
A problem is a statement that expresses dislike for the current or a future situation and contains no desire words. For example, "my boss is nagging me".
A remedy is a statement that expresses a desire for a problem to disappear. For example, "I want my boss to stop nagging me".
An outcome is a statement that expresses a desired future state without containing a problem. For example, "I want a promotion".
The difference between a remedy and an outcome is that a remedy usually results in an absence of something bad, an outcome results in the presence of something good.
The original article from 2006 by Penny Tompkins and James Lawley can be found here.