Hybrid Neural Network System for Accurate Natural Language Question Answering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human-computer interaction systems, particularly automatic question-answering and natural language dialog systems, face challenges in providing accurate and knowledge-based responses due to issues like natural language polysemy and ambiguity, leading to low accuracy in answering user questions.
Innovation Solution
A neural network system is developed that converts natural language questions into question vectors, retrieves relevant knowledge answers from a knowledge base, calculates similarity using intermediate result vectors, and generates fact-based natural language answers by combining deep learning technologies with a knowledge base-based question-answering system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a deep neural network-based natural language dialog system is used to generate semantically relevant replies, then the system can have meaningful conversations with users, but the correctness of reply sentences cannot always be ensured due to model limitations
Solution Approach 1:
The patent merges a deep neural network-based natural language dialog system with a knowledge base-based automatic question-answering system. The dialog system handles semantic understanding and natural language generation, while the question-answering system ensures factual correctness by retrieving verified information from the knowledge base. This combination allows the system to maintain conversation fluency while improving reply accuracy.
Solution Approach 2:
The patent introduces an intermediate module that acts as a mediator between the dialog system and the knowledge base. This module determines when to query the knowledge base based on the user's question, retrieves relevant factual information, and integrates it with the dialog system's responses. The intermediary ensures that factual queries are answered with verified information while maintaining the natural flow of conversation.
2Device complexity
If only automatic question-answering or natural language dialog is considered separately, then the system design is simpler, but the accuracy of reply to user questions is low
Solution Approach 1:
The patent combines two previously separate systems - automatic question-answering and natural language dialog - into a unified hybrid system. The question-answering component provides factual accuracy by querying the knowledge base, while the dialog component ensures natural language fluency. This integration achieves high accuracy in question-answering while maintaining conversational capability, resolving the trade-off between system complexity and performance.
Data Source
AI summary
A method and neural network system for human-computer interaction, and user equipment are disclosed. According to the method for human-computer interaction, a natural language question and a knowledge base are vectorized, and an intermediate result vector that is based on the knowledge base and that represents a similarity between a natural language question and a knowledge base answer is obtained by means of vector calculation, and then a fact-based correct natural language answer is obtained by means of calculation according to the question vector and the intermediate result vector. By means of this method, a dialog and knowledge base-based question-answering are combined by means of vector calculation, so that natural language interaction can be performed with a user, and a fact-based correct natural language answer can be given according to the knowledge base.