Automated Fault Report Classification Using Semantic Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual sorting of large numbers of fault reports for mechanical assets is time-consuming and costly due to the need for manual matching of report components with asset lists, which often have thousands of nodes in a hierarchical structure.
Innovation Solution
A method that automatically classifies query strings by generating sets of query string-object name pairs based on closeness scores and semantic similarity, using approximate string matching and a semantic matching language model like BERT, to identify matching components in a list without a tree structure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual sorting and matching of fault reports with asset lists is performed, then accuracy in identifying components can be maintained, but time consumption and cost increase significantly
Solution Approach 1:
The patent replaces the manual mechanical sorting process with an automated computer-based system that uses natural language processing and machine learning algorithms to match fault report components with asset list components, eliminating manual labor while maintaining or improving matching accuracy
Solution Approach 2:
The patent introduces an automated classification system as an intermediary between fault reports and asset lists, using trained machine learning models to perform the matching function that previously required human operators, thereby resolving the contradiction between speed and accuracy
2Productivity
If automated classification using NLP and machine learning is implemented, then processing speed and efficiency improve, but system complexity increases
Solution Approach 1:
The patent applies preliminary action by training machine learning models in advance on labeled fault report data, so that when actual classification is needed, the pre-trained system can quickly and accurately process reports without requiring complex real-time decision-making logic
Solution Approach 2:
The patent changes parameters by transforming unstructured text data into structured feature representations that machine learning models can process, converting the classification problem into a mathematical optimization problem that can be solved efficiently
3Ease of operation
If hierarchical asset tree structures are maintained for organizing components, then asset organization and navigation are improved, but manual matching difficulty increases due to thousands of nodes
Solution Approach 1:
The patent extracts the essential matching information from fault reports (component names, descriptions, characteristics) and directly compares these extracted features with asset list components, bypassing the need for manual navigation through hierarchical tree structures
Solution Approach 2:
The patent creates a simplified representation or copy of the asset list components with key identifying features, allowing the system to perform matching operations on this simplified version rather than navigating the full hierarchical structure
Data Source
AI summary
A plurality of object names and a query string are defined. A first set that includes a plurality of first query string-object name pairs is generated based on the plurality of object names and the query string. A closeness score is determined for each first query string-object name pair in the first set. A second set that includes a plurality of second query string-object name pairs is generated, wherein the second query string-object name pairs are selected from the first set based on the closeness scores. A measure of semantic similarity is determined for each second query string-object name pair in the second set. A third set of third query string-object name pairs is defined, wherein the third query string-object name pairs are selected from the second set based on the measures of semantic similarity.