Chatbot Self-Learning for Failed Counseling Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Chatbots struggle to provide answers to all user inquiries, especially for new problems, as they require manual administrator-generated responses, leading to increased administrative effort and reduced customer satisfaction.
Innovation Solution
An electronic apparatus and method that identifies failed counseling histories, acquires answers by searching counseling manuals and external search engines, and generates dialogue scenarios to automatically respond to user queries, reducing the need for manual administrator intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a chatbot uses manual administrator-generated answers for user queries, then answer accuracy is improved, but administrative workload increases and response time delays occur
Solution Approach 1:
The chatbot system performs self-learning by automatically analyzing failed counseling histories, extracting user queries, searching for answers from manuals and search engines, and generating dialogue scenarios without administrator intervention. This self-service mechanism reduces administrative workload while maintaining answer quality through automated validation processes.
Solution Approach 2:
The system proactively identifies failed counseling cases and automatically generates answers before administrators need to intervene. By performing preliminary analysis and answer generation, the system reduces response time and prevents backlog accumulation, thereby improving both accuracy and administrative efficiency.
2Reliability
If a chatbot manually processes each new user query, then customer service quality is maintained, but response time increases
Solution Approach 1:
The system automatically analyzes failed counseling histories and generates answers in advance before user queries are submitted. This preliminary processing eliminates waiting time for administrator responses while maintaining service quality through systematic answer verification and dialogue scenario generation.
Solution Approach 2:
The system continuously monitors counseling outcomes, identifies failed cases, and uses this feedback to improve future responses. By implementing a closed-loop feedback mechanism, the chatbot learns from past failures and automatically adjusts its answering strategy, maintaining high service quality while reducing response time.
3Adaptability or versatility
If a chatbot increases domain conversation topics and data, then conversation capability is improved, but system complexity increases
Solution Approach 1:
The chatbot automatically expands its knowledge base by analyzing failed counseling cases, extracting new topics and queries, and generating corresponding dialogue scenarios. This self-driven expansion mechanism increases conversation capability without requiring complex manual configuration or system redesign.
Solution Approach 2:
The system dynamically adapts its conversation capabilities by continuously learning from new user queries and failed counseling cases. Rather than statically pre-configuring all possible topics, the system evolves its knowledge base in real-time, maintaining versatility while managing complexity through adaptive learning processes.
Data Source
AI summary
An electronic apparatus includes a memory, and a processor connected to the memory and configured to control the electronic apparatus, where the processor is configured to identify a plurality of counseling end histories among a plurality of counseling histories stored in the memory, each of the plurality of counselling end histories indicating that a counselling has failed to answer a user query, based on a ratio of the plurality of counseling end histories to the plurality of counseling histories being equal to or greater than a threshold ratio, identify a first user query based on the plurality of counseling end histories, and based on a number of second user queries having a similarity to the identified first user query being equal to or greater than a first threshold number, acquire an answer to the first user query.