Determining context and intent in omnichannel communications using machine learning based artificial intelligence (AI) techniques

A method for determining intent in a digital conversation, comprising:

• receiving, by a processor, data associated with a conversation with a user via a communication channel, the data usable for training at least one model to determine intent in the conversation with the user;
• selecting, by the processor, unlabeled data from the data associated with the conversation with the user, the selected unlabeled data having probability of impacting the at least one model, wherein selecting the unlabeled data comprises:
• sampling, using one or more first auxiliary models, the unlabeled data based on the probability that the unlabeled data has on impacting a model performance of the at least one model,
• providing, by the processor, a hierarchical multi-intent data labeling framework for labeling a representative subset of data from the data associated with the conversation with the user and the selected unlabeled data;
• generating, by the processor for the labeled representative subset of data from the data associated with the conversation and the selected unlabeled data based on the hierarchical multi-intent data labeling framework, at least one primary intent and one or more layers of sub-intents for each of the at least one primary intent;
• training, by the processor, the at least one model based on the labeled representative subset of data and a plurality of parameter choices;
• retaining, by the processor, at least one candidate model set from the at least one trained models based on a global metric performance;
• generating, by the processor, a production-ready model based on the retained at least one candidate model set;
• and
• deploying, by the processor, the production-ready model for intent determination for at least the conversation with the user.

A non-transitory computer-readable storage medium having an executable stored thereon, which when executed instructs a processor to perform the following:

• receiving data associated with a conversation with a user via a communication channel, the data usable for training at least one model to determine intent in the conversation with the user;
• selecting unlabeled data from the data associated with the conversation with the user, the selected unlabeled data having probability of impacting the at least one model wherein selecting the unlabeled data comprises:
• sampling, using one or more first auxiliary models, the unlabeled data based on the probability that the unlabeled data has on impacting a model performance of the at least one model,
• providing a hierarchical multi-intent data labeling framework for labeling a representative subset of data from the data associated with the conversation with the user and the selected unlabeled data;
• generating, for the labeled representative subset of data from the data associated with the conversation and the selected unlabeled data based on the hierarchical multi-intent data labeling framework, at least one primary intent and one or more layers of sub-intents for each of the at least one primary intent;
• training the at least one model based on the labeled representative subset of data using a plurality of parameter choices;
• retaining at least one candidate model set from the at least one trained models based on a global metric performance;
• generating a production-ready model based on the retained at least one candidate model set;
• and
• deploying the production-ready model for intent determination for at least the conversation with the user.