Graph Question Answering Using Multi-Model Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning technologies lack the capability to effectively answer visual or symbolic graph-based questions using digital images or symbolic descriptions in conjunction with natural language, particularly in contexts like electrical circuit inspection and map-based navigation.
Innovation Solution
A computer-implemented method and device for machine learning that processes digital images or symbolic graph descriptions, utilizing multiple models for semantic segmentation, optical character recognition, and answer-set programming to determine expressions associating nodes and edges, and generate answers to natural language questions based on graph content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If current machine learning technologies are used, then processing of simple data is possible, but answering visual or symbolic graph-based questions cannot be effectively performed
Solution Approach 1:
The system segments the graph question answering task into three distinct models: a first model for recognizing graph structures and determining expressions associating edges with nodes, a second model for optical character recognition to associate names with nodes, and a third model for processing natural language questions. This segmentation allows each model to specialize in a specific aspect, improving overall effectiveness while maintaining adaptability across different graph types and questions.
2Measurement precision
If multiple models are used for processing, then accuracy in graph question answering is improved, but system complexity increases
Solution Approach 1:
The system merges three specialized models into a unified graph question answering system. The first model processes graph structures, the second model handles text recognition and node name association, and the third model processes natural language questions. These models work together in an integrated manner, with their combined outputs feeding into a cohesive decision-making process, thereby achieving high accuracy while managing complexity through systematic integration.
3Productivity
If automated inspection of electrical circuits is implemented, then manufacturing efficiency is improved, but requirement for specialized graph processing capability arises
Solution Approach 1:
The graph question answering system is designed with universal applicability across multiple domains. The same three-model architecture can process electrical circuit graphs for manufacturing inspection, map graphs for navigation, and other graph-based structures. The system's ability to handle different graph types and questions through its versatile model combination enables it to serve multiple functions, including automated electrical circuit inspection, thereby improving manufacturing efficiency while adapting to specialized processing requirements.
Data Source
AI summary
A device and computer-implemented method for machine learning. The method includes: providing an input, in particular a digital image or a symbolic description of a graph, wherein the input includes a first node, a first name, a second node, and a second name, and an edge between the first node and the second node; determining, with a first model an expression that associates the edge with the first node and the second node; determining with a second model an expression that associates the first node with the first name and an expression that associates the second node with the second name; providing a question that includes the first name and the second name; determining, with a third model, depending on the question an expression that includes the first name and the second name, and determining an answer to the question depending on the expressions.