Machine learning is the operating process of an Artificial Intelligence (AI) or simple algorithmic system, whereby the process is equipped with a learning system, via the analysis of data (e.g., email exchanges, past customer complaint files, etc.).
At dydu, the bot administrator creates the knowledge base from scratch, so it’s not really machine learning. Our solution also doesn’t learn alone, we use a supervised learning mode.
Supervised learning here means that the action of a human, in this case the bot’s administrator, is required to ensure that the bot understands your users. We use supervised learning modules. These modules aim to help the bot administrator manage and improve the business knowledge articles by providing suggestions for improvements according to matching, suggestions for knowledge that is too close, quality alerts, audit reports, etc.
Bot management by a team of operational staff, with a well-defined scope, allows to properly manage the answers provided to users and to obtain a high level of user satisfaction from the start, unlike machine learning that requires a lot of data, and therefore dialogues, so that the bot can learn by itself.