K-Trans introduces a novel approach that significantly improves LLM performance when translating code within complex repository contexts, addressing a critical gap in industrial applications.

Key Innovations:

• Leverages triple knowledge augmentation to navigate complex dependencies in repository-level code
• Outperforms standard LLM code translation techniques in real-world environments
• Addresses the translation gap between function-level and repository-level code contexts
• Enhances industrial applicability of LLM-based code translation systems

Engineering Impact: This research enables more reliable code migration between programming languages in enterprise settings, where code rarely exists in isolation and depends on the broader repository context.

Enhancing LLM-based Code Translation in Repository Context via Triple Knowledge-Augmented