Customer requests to a call center are subject to broad business and/or social situations like current promotions or government measures.
We have developed an algorithm to detect the latest trends when it comes to unfamiliar words and unusual expressions.
From a technical point of view, this algorithm is trained on raw data without any human supervision. In the case of a bot, we use the verbatims not understood by the bot to identify the words and phrases used. This can guide the bot designer towards new intentions or entities to be modeled. In the case of a semantic analysis of human-human conversations, this algorithm allows you to build classes that group together the phrases and expressions most used by your customers, giving you an in-depth view of their needs.