Oracle Digital Assistant Explained PART 2: A Deep Dive into Oracle Digital Assistant's Trainer HT and Trainer TM

 Hello there fellow readers.

In my previous blog post about Oracle Digital Assistant, I have explained in the simplest possible way about basic concepts such as skills, Intents, Entities, and Utterances. In this blog post, we are diving deep into the training models that are incorporated into Oracle Digital Assistant.

In Oracle Digital Assistant we have 2 training models: "Trainer HT" and "Trainer TM".

In today's digital landscape, conversational AI plays a pivotal role in enhancing customer engagement and streamlining business processes. Oracle Digital Assistant stands at the forefront of this evolution, offering advanced tools such as Trainer HT and Trainer TM to train and refine the capabilities of digital assistants. These tools are instrumental in improving the accuracy and effectiveness of conversational AI, ultimately leading to a more seamless and intuitive user experience.

A Subtle Background:

Trainer HT is a linguistic-based model provided by Oracle Digital Assistant. As it is a basic model that does not have Machine Learning intervention, it requires only 4-5 utterances to get trained. It does not take much of a time to get trained, hardly 5-6 seconds are enough for it to get trained. 

Whereas, Trainer TM is a Machine learning-based model. It uses an 80-20 train-test split algorithm. It typically means that 80% of the utterances are used for training and the rest 20% goes for testing.

Trainer HT and Trainer TM are integral components of Oracle Digital Assistant, offering advanced training capabilities to enhance conversational AI. While Trainer HT focuses on entity extraction for precise data handling, Trainer TM improves intent recognition for a more holistic understanding of user inputs. By leveraging these tools, businesses can create more intuitive and effective conversational experiences, ultimately driving greater customer satisfaction and loyalty.

What should we use?

As an Oracle Digital Assistant Developer, it is not recommended to use Trainer HT. As we are dealing with machine learning in the coming years, it is better to opt for Trainer TM. There is a specific reason for that too.

Whenever we provide utterances for the digital assistant to get trained, it is important to keep in mind that Conversational AI learns by the examples, not by the heart. Whatever examples you feed to it, it will act accordingly. Once you have trained the example utterances on the basis of intent and the messages you have anticipated for the task, the simple linguistic model (Trainer HT) will not be able to classify the messages outside the scope of the training set provided.

Keeping this blog short and sweet, this was an information post for the new beginners as expert Oracle Digital Assitant Developers might relate it somewhere and somehow. Until then Bye, Ciao!!