Compound Chatbot with Overtrained and General Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In complex cloud-based computing environments, technical support personnel often lack expertise and skills to provide accurate and timely answers to technical questions, leading to excessive delays, especially when customers are in different time zones, due to the limitations of existing AI-based customer support chatbots that provide generic responses.
Innovation Solution
A method utilizing a compound chatbot with a strategically overtrained neural network and a non-overtrained neural network, where the overtrained network provides specific answers within a predetermined confidence level, and the non-overtrained network provides more general answers when the overtrained network is uncertain, supported by a deep learning recurrent neural network with Long Short-Term Memory (LSTM) cells, and an automatic messaging system to notify expert teams for precise answers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single overtrained neural network is used to provide specific answers, then answer precision is improved, but the system cannot handle queries outside its training domain
Solution Approach 1:
The support system is segmented into multiple specialized neural networks, each trained on specific technical domains. Instead of one general network, the system divides functionality into domain-specific experts (e.g., network infrastructure, security, applications) that can be selectively activated based on query type, resolving the contradiction between precision and versatility.
Solution Approach 2:
The system creates a universal support platform that integrates multiple specialized neural networks behind a single interface. The router component enables the system to function as both a specific domain expert (when routing to specialized networks) and a general support system (when handling diverse query types), achieving multi-functionality that resolves the precision-versatility tradeoff.
2Reliability
If multiple specialized neural networks are deployed, then query accuracy is improved, but system complexity increases
Solution Approach 1:
A router neural network serves as an intermediary component that receives queries and determines which specialized neural network should handle them. This mediator layer manages the complexity of having multiple networks by providing a unified interface and intelligent routing logic, allowing the system to maintain high accuracy while controlling architectural complexity through centralized coordination.
Solution Approach 2:
The system dynamically selects and activates only the relevant neural network for each specific query rather than maintaining all networks in active state. This dynamic approach optimizes resource utilization and reduces operational complexity while preserving the accuracy benefits of specialized networks when needed.
3Loss of time
If technicians rely on their own knowledge and available knowledge-based systems, then response time is reduced, but answer accuracy decreases due to knowledge gaps
Solution Approach 1:
The neural network support system enables technicians to obtain accurate answers independently without needing to consult subject matter experts or search extensive documentation. The system provides self-service capabilities where technicians can query the AI assistant and receive precise, context-aware responses immediately, eliminating both time delays and knowledge gaps simultaneously.
4Measurement precision
If expert teams are contacted for difficult questions, then answer accuracy is improved, but response time increases due to availability constraints
Solution Approach 1:
The system creates AI-based copies of expert knowledge that are available 24/7 without fatigue or availability constraints. Instead of relying on actual human experts who may be unavailable, the neural networks capture and replicate expert problem-solving patterns, providing accurate answers immediately at any time, thus eliminating the time penalty associated with expert availability while maintaining answer quality.
Data Source
AI summary
An example system and method facilitate enabling precise answers to technical support questions pertaining to a given computing environment, e.g., a cloud-based enterprise computing environment. The example method includes using an overtrained recurrent neural network employing Long Short-Term Memory (LSTM) cells to selectively answer specific questions by providing precise answers to technical support questions. A second neural network that is not overtrained can provide more generalized answers when a confidence measurement of an answer of the first neural network falls below a predetermined threshold. Furthermore, expert staff, e.g., developers and/or engineers that may be able to more precisely answer a specific question may be shown the answer. Subsequent expert answers or modifications to existing answers may be used to further refine the neural networks, e.g., via periodic supervised learning.