Dynamic Context Quick Replies for Chatbot Intent Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online chat systems face challenges in accurately predicting customer intents due to varied message formats, leading to inefficiencies in real-time feedback and document retrieval.
Innovation Solution
A dynamic conversation context-driven automated quick reply generator system using a chatbot and AI server to analyze user messages, provide matching intents, and retrieve or summarize documents, with constraints on reply number and length.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated intent prediction is used to reduce service cost and improve scalability, then productivity is improved, but measurement precision deteriorates due to difficulty in accurately predicting customer intents from varied message formats
Solution Approach 1:
The patent introduces a human service representative as an intermediary to review and correct automated intent predictions. The system flows customer messages through an automated intent predictor first, then presents uncertain predictions to human representatives for verification. This intermediary layer allows the system to maintain high scalability through automation while preserving measurement precision through human oversight on edge cases.
Solution Approach 2:
The system applies automated intent prediction selectively rather than universally. It uses automated prediction for clear, straightforward messages but transitions to human review for ambiguous or complex cases. This partial application of automation balances productivity gains from automated processing with measurement precision maintained through human judgment when needed.
2Speed
If real-time chat conversation is implemented to provide immediate feedback, then speed is improved, but device complexity increases due to need for automated conversation management and document retrieval systems
Solution Approach 1:
The patent segments the chat system into distinct functional modules: an intent predictor module, a document retrieval module, and a service representative interface module. Each module handles a specific aspect of the conversation process independently. This segmentation allows real-time feedback speed to be maintained through specialized optimization of each module while managing device complexity through modular architecture that makes the system easier to maintain and scale.
3Loss of time
If chatbot is used to automate conversation and reduce waiting time, then loss of time is reduced, but reliability deteriorates when chatbot cannot accurately understand varied customer message formats
Solution Approach 1:
The patent implements a feedback mechanism where service representatives review chatbot responses and intent predictions, providing corrections that are fed back into the system. This feedback loop continuously improves the chatbot's understanding of varied customer message formats. The system maintains reduced waiting time by having the chatbot handle initial processing while using feedback to progressively improve reliability without increasing customer wait time.
Data Source
AI summary
A system and a method for responding to user messages during an online chat conversation include: a computing system including a processor executing computer-readable instructions stored in a storage device; a chatbot; an AI server; and a network connection operatively connecting a user device to the computing system. The computing system performs steps of: establishing a chat conversation between the chatbot and the user device; receiving a chat message from the user device, the chatbot then sending chat history contextual data to the AI server and requesting quick replies from the server; the server responding to the request by sending to the chatbot at least one of the quick replies including a matching intent; and the chatbot sending to the user device the quick replies; and the quick replies being displayed by the user device on a display screen visual representation of a portion of the chat conversation.