Bootstrapping Code Translation with Weighted Multilanguage Exploration

2026-01-07

arXiv.org

AI Summary

This work addresses the challenges of multilingual code translation, which are primarily constrained by the scarcity of parallel data and imbalanced optimization across language pairs. The authors propose BootTrans, a novel approach that leverages test suites as verification oracles to enforce cross-lingual functional equivalence. BootTrans employs a dual-pool bootstrapping architecture—comprising a seed pool and an exploration pool—to iteratively expand high-quality training data in a guided manner. Additionally, it introduces a language-aware dynamic weighting strategy that adaptively adjusts training emphasis across different language pairs. Evaluated on the HumanEval-X and TransCoder-Test benchmarks, BootTrans substantially outperforms existing large language model baselines, achieving consistent and significant performance gains across all translation directions.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Code translation across multiple programming languages is essential yet challenging due to two vital obstacles: scarcity of parallel data paired with executable test oracles, and optimization imbalance when handling diverse language pairs. We propose BootTrans, a bootstrapping method that resolves both obstacles. Its key idea is to leverage the functional invariance and cross-lingual portability of test suites, adapting abundant pivot-language unit tests to serve as universal verification oracles for multilingual RL training. Our method introduces a dual-pool architecture with seed and exploration pools to progressively expand training data via execution-guided experience collection. Furthermore, we design a language-aware weighting mechanism that dynamically prioritizes harder translation directions based on relative performance across sibling languages, mitigating optimization imbalance. Extensive experiments on the HumanEval-X and TransCoder-Test benchmarks demonstrate substantial improvements over baseline LLMs across all translation directions, with ablations validating the effectiveness of both bootstrapping and weighting components.

Problem

• code translation
• parallel data scarcity
• optimization imbalance
• multilingual programming
• test oracles

Innovation

• bootstrapping
• multilingual code translation
• test oracle adaptation
• language-aware weighting
• execution-guided data expansion