DITING: A Multi-Agent Evaluation Framework for Benchmarking Web Novel Translation
This work addresses the problem of evaluating translation quality in web novels for researchers and developers, though it is incremental as it builds on existing LLM-based translation methods.
The authors tackled the lack of effective benchmarks for evaluating machine translation of web novels by introducing DITING, a comprehensive framework with expert-annotated data, and found that Chinese-trained LLMs like DeepSeek-V3 outperform larger foreign models in delivering faithful translations.
Large language models (LLMs) have substantially advanced machine translation (MT), yet their effectiveness in translating web novels remains unclear. Existing benchmarks rely on surface-level metrics that fail to capture the distinctive traits of this genre. To address these gaps, we introduce DITING, the first comprehensive evaluation framework for web novel translation, assessing narrative and cultural fidelity across six dimensions: idiom translation, lexical ambiguity, terminology localization, tense consistency, zero-pronoun resolution, and cultural safety, supported by over 18K expert-annotated Chinese-English sentence pairs. We further propose AgentEval, a reasoning-driven multi-agent evaluation framework that simulates expert deliberation to assess translation quality beyond lexical overlap, achieving the highest correlation with human judgments among seven tested automatic metrics. To enable metric comparison, we develop MetricAlign, a meta-evaluation dataset of 300 sentence pairs annotated with error labels and scalar quality scores. Comprehensive evaluation of fourteen open, closed, and commercial models reveals that Chinese-trained LLMs surpass larger foreign counterparts, and that DeepSeek-V3 delivers the most faithful and stylistically coherent translations. Our work establishes a new paradigm for exploring LLM-based web novel translation and provides public resources to advance future research.