Model Deployment and Training

We can classify the overall process into three stages:

1. Data Ingestion
2. Build Model
3. Deploy Model

For Data Ingestion of emotion, OCI Data Science jobs pull the chat data using the reports from Oracle B2C Service cloud. This chat data is tagged with emotion using EmoRoBERTa model and the result is stored into the object storage for emotions.

The following diagram shows the process of model deployment and training.

In case of supervisor ask model, OCI jobs pull the data from a csv file and store the same into object storage for supervisor ask.

Build and Deploy Model stage is a data science job which takes the data from the object storage and cleans the data before it can be used for creating and training the model. The job creates and deploys a custom model with training data. Once the model is deployed, the job creates a model endpoint which can be accessed by the OCI users.

For more information on creating and training a custom model, refer to Oracle Cloud Infrastructure Documentation.