A 17-metric testing framework for an LLM-RAG tourism system shows model upgrades improve response length and complexity, not semantic quality, while extreme temperature and top-p settings degrade output.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Quality Assurance for LLM-RAG Systems: Empirical Insights from Tourism Application Testing
A 17-metric testing framework for an LLM-RAG tourism system shows model upgrades improve response length and complexity, not semantic quality, while extreme temperature and top-p settings degrade output.