Intent classification for conversational AI
Conversational agents, such as Apple’s SIRI, Amazon’s Alexa, or the Google Assistant are capable of handling a broad range of requests. In the context of such conversational AI systems, the objective of this project is to develop a machine learned classifier that determines the underlying intent of user utterances.
The project involves the following specific tasks:
- Developing a categorization system of possible intents (e.g., requesting information, clarifying request, providing additional details, critiquing the system, stating/removing personal preferences, etc.)
- Devising appropriate classification methods; both traditional feature-based and neural (deep learning) approaches might be explored.
- Creating a data collection for training and evaluation.