"Show me the poem about Knecht Ruprecht, Santa's little helper." Answer: "Unfortunately I was unable to find Ruprecht in your contacts." Oops! In this dialog with a machine there is still a problem with "natural language understanding" (NLU).
"Do you have iPhone X in stock?" – "Yes," answers the chatbot. Next question from the customer: "And what does it cost me?" If the digital assistant answers this correctly as well, it is probably more than just a search engine: it is equipped with "natural language understanding" – the ability to understand natural human language. Because the detached pronoun "it" cannot be easily understood by a simple machine, unless it also factors in the first question.
"Nuance," "IPSoft," and "7" are some of the best known software vendors that teach chatbots to understand and thus reveal the real intention behind user input. This is far more than just reacting to individual keywords. The objective is clear: users should be able to chat naturally. To make this possible, the developers feed the chatbot on the NLU platform with vocabulary, focusing on use cases. Another prerequisite is databases in which the various meanings of words are stored.
In summary: NLU is the component of an artificial intelligence application that teaches machines the meaning of words and sentences so that they understand them.