article by Grant Ronald, February 2021 (repost)
A key feature of Oracle Digital Assistant is the ability to construct your digital assistant from multiple skills. This concept of splitting software into more functionally manageable, or reusable, building blocks is a common technique in software development, whether it’s database tables, PL/SQL packages or Java classes.
However, even though the general concept is understood, the question as to what is the best approach for partitioning a digital assistant comes up frequently in a number of different guises: “How should I partition my digital assistant into skills”, “How many skills should I have”, or “Should I combine these two skills into one”.
Alas, there is no magic formula for calculating the answer, and more often the answer is “it depends”, but there are some key consideration to be aware of when making the decision.
This article explains how to how to partition your digital assistant projects into skills and how to make those decisions
Related Content
TechExchange: Training your NLP model – Best Practices In Writing Utterances
TechExchange – First Step in Training Your Bot
TechExchange: Implementing Varying Bot Responses Based on Intent Confidence
