LLM-Enhanced Framework for Building Domain-Specific Lexicon for Urban Power Grid Design DOI Creative Commons
Xu Yan, Tao Wang, Yang Yuan

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(8), P. 4134 - 4134

Published: April 9, 2025

Traditional methods for urban power grid design have struggled to meet the demands of multi-energy integration and high resilience scenarios due issues such as delayed updates terminology semantic ambiguity. Current techniques constructing domain-specific lexicons face challenges like insufficient coverage specialized vocabulary imprecise synonym mining, which restrict parsing capabilities intelligent systems. To address these challenges, this study proposes a framework lexicon based on Large Language Models (LLMs). The aim is enhance accuracy practicality through multi-level term extraction expansion. Initially, structured corpus covering national industry standards in field was constructed. An improved Term Frequency–Inverse Document Frequency (TF-IDF) algorithm, combined with mutual information adjacency entropy filtering mechanisms, utilized extract high-quality seed from 3426 candidate terms. Leveraging LLMs, prompt templates were designed guide incorporating self-correction mechanism verification mitigate errors caused by model hallucinations. This approach successfully built comprising core words 10,745 synonyms. average cosine similarity pairs reached 0.86, expert validation confirmed an rate 89.3%; text classification experiments showed that integrating dictionary classifier’s F1-score 9.2%, demonstrating effectiveness method. research innovatively constructs high-precision first time embedding domain-driven constraints workflows, solving problems expansion traditional methods, supporting development semantically systems smart design, significant practical application value.

Language: Английский

LLM-Enhanced Framework for Building Domain-Specific Lexicon for Urban Power Grid Design DOI Creative Commons
Xu Yan, Tao Wang, Yang Yuan

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(8), P. 4134 - 4134

Published: April 9, 2025

Traditional methods for urban power grid design have struggled to meet the demands of multi-energy integration and high resilience scenarios due issues such as delayed updates terminology semantic ambiguity. Current techniques constructing domain-specific lexicons face challenges like insufficient coverage specialized vocabulary imprecise synonym mining, which restrict parsing capabilities intelligent systems. To address these challenges, this study proposes a framework lexicon based on Large Language Models (LLMs). The aim is enhance accuracy practicality through multi-level term extraction expansion. Initially, structured corpus covering national industry standards in field was constructed. An improved Term Frequency–Inverse Document Frequency (TF-IDF) algorithm, combined with mutual information adjacency entropy filtering mechanisms, utilized extract high-quality seed from 3426 candidate terms. Leveraging LLMs, prompt templates were designed guide incorporating self-correction mechanism verification mitigate errors caused by model hallucinations. This approach successfully built comprising core words 10,745 synonyms. average cosine similarity pairs reached 0.86, expert validation confirmed an rate 89.3%; text classification experiments showed that integrating dictionary classifier’s F1-score 9.2%, demonstrating effectiveness method. research innovatively constructs high-precision first time embedding domain-driven constraints workflows, solving problems expansion traditional methods, supporting development semantically systems smart design, significant practical application value.

Language: Английский

Citations

0