Abstract
This study describes a method for quantifying human language requirements for design modifications using a Large Language Model (LLM) and reflects those quantified values in real-time on design drawings. To achieve this, we propose an information integration process for quantification, and a design instruction sentence construction method for prompting. The experimental results demonstrate that the LLM generates quantified values, and scatter plot and variance analyses confirm the reliability of results obtained. A case study involving integration of the LLM with 3D modeling software demonstrated effective real-time model modification. Therefore, this study proposes and validates a method for LLM utilization in design tools, thus demonstrating the potential for further advancement of these tools.
| Translated title of the contribution | A Study on the Quantification of Quantitative Adjective Expressions Using LLM |
|---|---|
| Original language | Korean |
| Pages (from-to) | 581-591 |
| Number of pages | 11 |
| Journal | Transactions of the Korean Society of Mechanical Engineers, A |
| Volume | 48 |
| Issue number | 9 |
| DOIs | |
| State | Published - 2024 |
Keywords
- Design Automation Tool
- Information Integration Prompting
- Large Language Model
- NLP Application
- Quantification