What question did this study set out to answer?

This research aims to systematically evaluate the capabilities of MLLMs on oracle bone inscriptions, a challenging cultural heritage artifact.

July 5, 2026npj Heritage ScienceOpen Access

Structure–semantic dual-alignment benchmarking of large language model capability on ancient oracle bone inscriptions

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial evaluates MLLM performance on ancient oracle bone inscriptions, suggesting significant capability differences.

This research aims to systematically evaluate the capabilities of MLLMs on oracle bone inscriptions, a challenging cultural heritage artifact.
Developed the Dual-Alignment Oracle Character Reasoning Benchmark (DAOCR-Bench) for evaluation.
Conducted experiments on 18 multimodal large language models under various oracle bone conditions.
Assessed structural perception, semantic grounding, reconstruction, and consistency judgment.
GPT-4o achieved the highest comprehensive score in the evaluation.
Substantial performance differences were observed across the models under both standardized and degraded conditions.

Wong et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark