Automated Search Pattern Creation from Knowledge Bases
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning algorithms require a large quantity of training data and significant effort for training and fine-tuning, making it difficult to automatically recognize errors in medical technology systems and provide quick solutions.
Innovation Solution
A method for automated creation of search patterns and training datasets using a knowledge database, such as the Siemens Knowledge Base, to enable targeted searching of protocol files and automatic problem analysis or solution derivation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning algorithms are used for automated error recognition in medical technology systems, then problem analysis speed is improved, but the requirement for large quantities of training data and significant training effort increases
Solution Approach 1:
The patent applies preliminary action by pre-processing unstructured knowledge base entries into structured search patterns before they are needed for error recognition. The system extracts key concepts, error conditions, and solutions from unstructured text and images in advance, transforming them into machine-learning-ready formats. This eliminates the need for extensive training data collection and model training at runtime, as the knowledge is already prepared and structured for immediate use in automated diagnostic workflows.
2Quantity of substance
If traditional manual error analysis methods are used, then training data requirements are reduced, but problem analysis time and manual effort increase significantly
Solution Approach 1:
The system applies self-service by enabling automated extraction and structuring of knowledge from unstructured sources without requiring manual annotation or extensive human intervention. The processing unit automatically parses knowledge base entries, identifies key concepts, and generates search patterns autonomously. This self-organizing capability allows the system to prepare training data and diagnostic tools without significant manual effort, thereby reducing both training data preparation time and operational analysis time.
3Ease of manufacture
If unstructured knowledge base entries are used directly, then data preparation effort is reduced, but machine learning model training effectiveness decreases
Solution Approach 1:
The patent introduces an intermediary processing layer that transforms unstructured knowledge base entries into structured search patterns. This intermediary structure serves as a bridge between the raw unstructured data and the machine learning model, organizing key concepts, error conditions, and solutions into a standardized format. This intermediate structuring maintains ease of data preparation from unstructured sources while simultaneously improving model training effectiveness by providing consistent, organized input data that the model can process efficiently.
Data Source
AI summary
One or more example embodiments of the present invention describes a method for automated creation of search patterns on the basis of entries in a knowledge database. One or more example embodiments of the present invention further describes a search pattern dataset, a device for automated creation of search patterns, a problem solving method, a creation method, a training dataset and a device system.