LLM-Based Issue Identification for Remote Technical Support
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online communication systems for technical support in industrial environments face challenges in accurately identifying and resolving issues due to incomplete feature descriptions, language ambiguities, lack of technical expertise, and inadequate documentation, leading to lengthy problem-solving processes and potential equipment malfunctions.
Innovation Solution
A computer-implemented method using a Large Language Model (LLM) to translate non-formal user descriptions and additional information like images or videos into unambiguous technical language, leveraging a knowledge graph to identify issues and generate solutions, which can be further translated back into colloquial language for user understanding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If non-formal language descriptions are used by non-experts, then ease of operation is improved, but loss of information and measurement precision deteriorate
Solution Approach 1:
A Large Language Model (LLM) acts as an intermediary between the non-expert user and the technical support system. The LLM translates colloquial descriptions into precise technical language, bridging the gap between user convenience and information accuracy without requiring the user to have domain expertise
Solution Approach 2:
The patent replaces the mechanical process of manual technical translation with an AI-based language model. Instead of requiring human experts to interpret non-formal descriptions, the LLM automatically performs the translation from colloquial to technical language, improving both speed and consistency
2Measurement precision
If detailed technical descriptions are required, then measurement precision is improved, but ease of operation deteriorates
Solution Approach 1:
The system enables non-expert users to independently describe issues in their own words without requiring them to learn technical terminology. The LLM handles the conversion to precise technical descriptions, allowing users to self-serve while maintaining high accuracy
3Adaptability or versatility
If multiple languages are supported for communication, then adaptability is improved, but loss of information deteriorates due to translation ambiguities
Solution Approach 1:
The LLM serves as a sophisticated translation intermediary that not only translates between languages but also preserves technical meaning. It converts multi-lingual colloquial descriptions into standardized technical language, maintaining precision across language barriers
4Reliability
If on-site experts are deployed for issue resolution, then reliability is improved, but loss of time and productivity deteriorate
Solution Approach 1:
The system enables automated self-diagnosis and self-resolution by non-expert users through the LLM-guided process. Users can independently identify and resolve common issues without waiting for expert deployment, significantly reducing downtime while maintaining reliable resolution for actionable issues
Solution Approach 2:
The system performs preliminary issue analysis and generates potential solutions before expert intervention is needed. By pre-processing the issue description and generating actionable recommendations, the system resolves many issues automatically, reserving expert resources only for complex cases
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates a Computer-implemented method for the identification of an issue in a technical system with the steps: a) entry as a first information a description of the issue in non-formal language (12), and b) entry as a second information at least one picture, photo, or video of at least a presumably affected part of the technical system (11), c) analysis of the second information (20), and amending context of the technical system, d) generating output in technical unambiguous language (23) by a LLM (10), e) using output for evaluation of the to be identified issue (30, 40), f) generation of at least one possible solution for issue in technical language (41) and g) entry of possible solution and conversion by the LLM (10) generating an output (50).