Intent-Entity Prediction Model for Dynamic Chatbot Question Suggestions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing conversational AI technologies, such as chatbots, struggle to engage users in dynamic back-and-forth conversations due to the complexity of mimicking human agent behavior, often feeling limited to choose-your-own-adventure narratives.
Innovation Solution
A method for automatically generating question suggestions by predicting a next intent-entity sequence using a trained prediction model, which converts the predicted sequence into likely question suggestions to return to the user.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If complicated guided-paths are built to mimic human agent behavior, then the chatbot can converse more like a human, but the system becomes cumbersome and limited to choose-your-own-adventure narratives
Solution Approach 1:
The patent applies dynamics by transitioning from static, pre-defined conversation paths to a dynamic system that generates question suggestions in real-time based on the current conversation state. The model adapts its output according to the flowing context of the dialogue, making the conversation flexible without requiring complex predetermined paths.
Solution Approach 2:
The patent uses copying by training the model on historical human agent conversation data. The model learns to replicate human-like questioning patterns by copying from existing conversations, enabling it to generate appropriate question suggestions without needing to manually program complex conversation paths.
2Measurement precision
If large amounts of natural language data are processed to extract meaning, then the chatbot can understand user intent better, but the processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by pre-training the model on large datasets of natural language conversations before deployment. This preliminary training allows the model to learn intent recognition patterns in advance, so that during actual conversation, it can quickly process and understand user input without requiring extensive real-time computation.
3Ease of operation
If the chatbot follows predetermined conversation paths, then the conversation flow is structured, but the interaction feels limited and less engaging
Solution Approach 1:
The patent applies feedback by using the current conversation state as input to generate question suggestions. The system continuously monitors the dialogue context and adapts its suggestions based on what has already been exchanged, creating a responsive interaction that feels engaging while maintaining structured flow through the suggestion mechanism.
Data Source
AI summary
A method, computer system, and a computer program product for automatically generated question suggestions is provided. The present invention may include generating an intent-entity sequence representative of a current chat conversation with a user. The present invention may also include predicting, using a trained intent-entity prediction model, a next intent-entity sequence based on the generated intent-entity sequence representative of the current chat conversation. The present invention may further include converting the predicted next intent-entity sequence into at least one likely question suggestion to return back to the user.