Let's look at some of the challenges in the RB approach:
- It is not easy to mimic the behavior of a human.
- Selecting or designing architecture is the critical part of the RB system.
- In order to develop the RB system, you need to be an expert of the specific domain which generates rules for us. For NLP we need linguists who know how to analyze language.
- Natural language is itself a challenging domain because it has so many exception cases and covering those exceptions using rules is also a challenging task, especially when you have a large amount of rules.
- Arabic, Gujarati, Hindi, and Urdu are difficult to implement in the RB system because finding a domain expert for these languages is a difficult task. There are also less tools available for the described languages to implement the rules.
- Time consumption of human effort is too high. ...
