Pith. sign in

REVIEW 3 major objections 4 minor 86 references

JieZi: A Large-Scale Expert-Audited Dataset and Benchmark for Ancient Chinese Character Exegesis

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that ancient Chinese character exegesis can be modeled as a four-level vision-language question-answering task, and that the JieZi-Dataset and JieZi-Bench resources make it the first such task to have a large-scale…

desk verdict Strong resource paper whose benchmark independence claim for early script stages does not hold up as stated; fine-tuning gains may be partly circular. read the letter →

arxiv 2608.11741 v1 pith:UC26MV3J submitted 2026-08-12 cs.CV cs.AI

Ran Li , Huiguo He , Jiahuan Cao , Junle Liu , Hiuyi Cheng , Lianwen Jin This is my paper · ORCID
classification cs.CVcs.AI
keywords AncientChineseCharacterExegesisvision-languagequestionansweringpaleographicdatasetmultimodallargelanguagemodelsdiachronicevolutionglyph-formanalysisbenchmarkexpertaudit
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper's central claim is that the scholarly interpretation of an ancient Chinese glyph—identifying it, analyzing its components, explaining its original meaning, and tracing how it changed across script periods—can be formalized as a four-level vision-language question-answering task (ACCE) and backed by the first large-scale resources for that task. The authors build JieZi-Dataset, with over 500K QA pairs and 130K glyph images spanning six script stages, using an expert-in-the-loop pipeline in which an LLM generates questions from expert-designed templates plus dictionary source text, and experts spot-check each stage. They build JieZi-Bench, roughly 8K QA pairs whose reference answers come from separate authoritative dictionaries and every one of which was manually verified by experts. Benchmarking current multimodal models, they report that the models do reasonably on basic identification and script classification but struggle on form analysis, meaning, and diachronic evolution, and that fine-tuning on JieZi-Dataset improves performance on all four levels. A careful reader would care because this turns a labor-intensive humanities skill into a measurable machine-learning task with a public dataset and a reproducible evaluation.

What carries the argument

The load-bearing object is the ACCE task decomposition itself: four progressive levels (L1 basic information, L2 glyph form, L3 meaning, L4 diachronic evolution) spanning ten subtasks (CHAR, SCRC, STRC, COMR, COMF, COMI, FORC, ORIM, COME, EVOI). The mechanism that carries the argument is the expert-in-the-loop generation pipeline: expert-designed QA templates plus dictionary source-text references constrain LLM generation so answers are grounded in verified metadata rather than parametric memory; stage-wise human verification (spot-checks for the training set, exhaustive checks for the benchmark) prevents error propagation. The named central entities are JieZi-Dataset, the 500K-pair training resource, and JieZi-Bench, the 8K-pair evaluation resource, and the separation of benchmark sources from training sources is what makes the fine-tuning gains interpretable.

What would settle it

Independently re-derive answers for a random sample of JieZi-Bench items: give three paleography experts who did not build the benchmark the same glyph images and the same four dictionaries, without showing the released answers, and measure agreement with the released references; low agreement (e.g., below roughly 90%) would directly undermine the benchmark reliability claim.

Watch

Extended reading notes

Core claim

The central discovery is that the full exegesis workflow—not just recognition—can be stated as a structured VQA task with four progressive levels: basic information (character identity and script type), glyph form (structure, components, their functions and interpretations, formation category), glyph meaning (original meaning), and diachronic evolution (component evolution and holistic explanation). To support it, the paper contributes two complementary resources: a large training set whose generation is constrained by expert templates and verified dictionary text to suppress hallucination, and a smaller benchmark whose reference answers are curated from four authoritative lexicographic works and exhaustively checked by experts. On that benchmark, the paper finds that general multimodal models score moderately on script classification and formation classification but drop sharply on component-level, semantic, and evolutionary reasoning; after fine-tuning on JieZi-Dataset, a small model surpasses much larger general models on structural parsing, and the largest fine-tuned model achieves the strongest results across nearly all subtasks. The paper concludes that domain-specific, expert-audited training data is the decisive bottleneck for computational paleography, not model scale.

Load-bearing premise

Everything rests on the assumption that every JieZi-Bench QA pair was truly checked and revised by human experts and that the four source dictionaries agree on the entries that were kept, so the reference answers are trustworthy ground truth.

Editorial extensions

If this is right

  • JieZi-Bench gives the field a common yardstick: any future model can be scored on the same four levels and ten subtasks, with explicit splits for unseen characters, unseen glyphs, and unseen components.
  • Fine-tuning on JieZi-Dataset improves every subtask, so teams without access to huge general-purpose corpora can build competitive paleography models from this single public resource.
  • Because structural parsing holds up even when character identification fails on unseen glyphs, the data appears to teach transferable knowledge of visual form rather than rote character-to-label mapping; this points toward architectures that separate form analysis from identification.
  • Current models' lowest scores concentrate in component interpretation, original meaning, and evolution, which singles out the exact tasks where future data collection and modeling effort should focus.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper's claims, the four-level decomposition looks portable to other undeciphered or under-resourced scripts: identification, form analysis, meaning, and diachronic change are general questions for cuneiform, Egyptian, and Maya writing, though those fields would need equivalent authoritative dictionaries to reproduce the pipeline.
  • The finding that form-based reasoning survives recognition failure suggests a two-stage design—glyph-agnostic structural analysis feeding a separate identification stage—might be more effective for rare glyphs than end-to-end generation, a hypothesis the paper's data could be used to test.
  • The validated automated metrics (BERTScore and an LLM judge with human-correlation around 0.6 to 0.7) could serve as cheaper evaluation tools for open-ended humanities VQA beyond ancient Chinese, saving human annotation effort in benchmark construction.
  • The paper's paraphrase test covers only two rewritten questions per instance; a stronger paraphrase attack with more diverse rewrites would clarify how much of the fine-tuning gain is genuine glyph-grounded exegesis versus template pattern matching.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The manuscript introduces Ancient Chinese Character Exegesis (ACCE), a four-level vision-language question answering task that spans basic character identification, glyph-form analysis, meaning exegesis, and diachronic evolution. The authors construct two resources: JieZi-Dataset, a roughly 500K QA-pair training set with about 130K glyph images across six script stages, built from the modern etymological dictionary Hanzi Yuanliu Dazidian through OCR, LLM-based structurization, and template-guided VQA generation with stage-wise expert spot-checking; and JieZi-Bench, an approximately 8K QA-pair benchmark on 1,024 images, claimed to be fully expert-verified and curated from four lexicographic sources held separate from the training data. The paper benchmarks a range of commercial and open MLLMs and reports that fine-tuning Qwen3.5-2B/4B/9B on JieZi-Dataset improves performance on all ACCE levels, with additional controls for unseen characters, unseen glyphs, in-context learning, retrieval-augmented generation, data scaling, and question paraphrasing.

Significance. If the benchmark-independence issue identified below is resolved, this is a substantial contribution to computational paleography and multimodal understanding. The paper formalizes a scholarly workflow that prior datasets reduce to recognition, provides the first multi-script, multi-level VQA training resource at this scale, and evaluates with several carefully designed controls: unseen-character and unseen-glyph splits, comparison against ICL and RAG baselines, a paraphrase-robustness test, and human-expert validation of both the LLM-as-a-judge protocol and BERTScore. The public release of code and data and the explicit reporting of verification error rates are strengths. The main empirical claim, that domain-specific fine-tuning improves ACCE performance, is plausible and mostly supported by the reported experiments; the weakest load-bearing point is the provenance and independence of the JieZi-Bench reference answers for early script stages.

major comments (3)
  1. [Section 4.1 and Section 4.3] The provenance description for JieZi-Bench is internally inconsistent. Section 4.1 states that JieZi-Bench is sourced solely from Kangxi Dictionary, Shuowen Jiezi, Shuowen Jiezi Zhu, and Revised Mandarin Chinese Dictionary, while Section 4.3 and Fig. 3 state that the benchmark spans all six script stages, including Oracle Bone, Bronze, and Warring States. Shuowen Jiezi (ca. 100 CE) cannot contain oracle-bone forms, which were not archaeologically recognized until 1899, and none of the four cited works is a systematic source for Bronze or Warring States component functions or diachronic trajectories. The manuscript must state, per script stage, where the benchmark glyph images and the L2-L4 reference answers actually come from. As written, the claim that the benchmark was held separate from the training data is not established for a substantial subset of the benchmark items.
  2. [Section 5.3 and Section 6] Because JieZi-Dataset is built from Hanzi Yuanliu Dazidian, a modern etymological dictionary that synthesizes contemporary paleographic scholarship, and because early-script reference answers in JieZi-Bench must, per the previous comment, be expert-written from the same modern scholarship, the unseen-character and unseen-glyph splits do not by themselves rule out memorization of etymological analyses rather than transferable exegetical reasoning. The central fine-tuning conclusion would be materially strengthened by (i) reporting the provenance of each benchmark answer as taken from a pre-modern lexicographic source versus written by experts from modern paleographic literature, (ii) measuring textual or semantic overlap between benchmark reference answers and Hanzi Yuanliu Dazidian entries, and (iii) re-running the headline fine-tuning comparison on the subset of benchmark items whose reference answers can be traced to sources independent of the training dictionary.
  3. [Section 4.2 and Section 4.3] The claim that JieZi-Bench is entirely expert-curated is not backed by reproducibility statistics for the verification process. The paper reports error rates found during correction but does not state how many experts participated, what their paleographic qualifications were, how often experts disagreed, or how disagreements were resolved. Given that the paper itself acknowledges that scholarly consensus varies for early scripts and that even Shuowen Jiezi can completely misinterpret an etymology (Fig. 6), reporting inter-annotator agreement on a sample of the benchmark would materially support the benchmark-reliability claim.
minor comments (4)
  1. [Section 4.1] The empirical token-length threshold of 200 used to select benchmark entries should be justified, and the sensitivity of the benchmark composition to this threshold should be reported.
  2. [Table 3] Several generalization cells have very small sample sizes (e.g., Bronze UC n=26, Bronze UG n=87); confidence intervals or exact tests should be reported before drawing conclusions from differences between unseen-character and unseen-glyph conditions.
  3. [Abstract and Section 5.3] The phrase substantially improves performance across all four levels is broadly supported, but the L3 ORIM gains for the 2B and 9B models are modest (+3.4 and +2.0 in Table 2); the claim could be nuanced to avoid overstating the effect on original-meaning tasks.
  4. [Section 5.3 and Table 2] The near-zero baseline scores for Qwen3.5-4B are unusual and the copying-bias explanation is plausible, but the paper should clarify whether this anomalous behavior was observed consistently across all inference settings and whether the other model families exhibit any similar instability.

Circularity Check

1 steps flagged · score 5.0 of 10

For OBI, Bronze, and Warring States items the four named benchmark sources cannot supply the L2–L4 reference answers, so the claimed “held separate / preventing data leakage” guarantee fails by construction and part of the fine-tuning gain is measured against the training knowledge base itself; UC/UG splits and later-script items preserve independent content.

  1. other [Sec. 4.1 (Data Sources), Sec. 4.3 (Data Statistics and Fig. 3), Fig. 6]
    "JieZi-Bench is sourced from four classical and modern lexicographic works: Kangxi Dictionary, Shuowen Jiezi, Shuowen Jiezi Zhu, and Revised Mandarin Chinese Dictionary. These dictionaries are selected because their explanations are mutually verifiable and cover characters absent from the training data, preventing data leakage. … Both resources span all six script stages rather than concentrating on a single type, with Seal and Bronze scripts constituting a significant proportion."

    Shuowen Jiezi (ca. 100 CE) predates the 1899 discovery of oracle-bone script, so OBI and most Bronze/Warring States L2–L4 content cannot come from the four cited dictionaries; the paper itself notes Shuowen “completely misinterpret[ed]” the etymology of 射 (Fig. 6). Since JieZi-Bench spans all six script stages, its early-script reference answers must have been authored from modern paleographic knowledge — the same tradition, and largely the same modern etymological dictionary ([21]), that produced JieZi-Dataset’s 500K QA pairs. The “held separate…preventing data leakage” guarantee therefore fails by construction for that subset, and Sec. 4.3 concedes only 12.5% of benchmark characters are entirely unseen.

full rationale

The paper’s derivation chain is otherwise self-contained: JieZi-Dataset is built from Hanzi Yuanliu Dazidian [21] plus ACCP [52] and MegaHan97K [66] images; JieZi-Bench from Kangxi, Shuowen Jiezi, Shuowen Jiezi Zhu, and Revised Mandarin Chinese Dictionary with exhaustive expert verification; and fine-tuning is scored on the held-out benchmark using BERTScore and an LLM-as-a-judge protocol that are validated against human expert ratings (Tabs. 8–9). No load-bearing self-citation chain is present: self-citations (ACCP, MegaHan97K, C3-Bench, MCS-Bench, TongGu-VL, EVOBC) function as data sources or related work rather than as justification for the central claim, and no uniqueness theorem is imported from the authors’ prior work. The paraphrase-robustness test (Tab. 13), the unseen-character and unseen-glyph splits (Tab. 3), and the observation that CHAR collapses while COMF/COMI hold on unseen glyphs provide genuine, non-circular transfer evidence; for Seal, Clerical, and Regular items the four dictionaries are legitimate external sources independent of the training dictionary. The one substantive circularity is the early-script portion of JieZi-Bench: because the cited dictionaries cannot contain OBI/Bronze/Warring States exegesis, the benchmark’s claimed separation from the modern etymological knowledge used to build the training data is unestablished for that subset, so a share of the reported fine-tuning gains measures consistency with the same knowledge base that produced the training QA pairs rather than validation against independent scholarly authority. This is partial circularity of the evaluation claim, not of dataset construction; the paper’s own Fig. 6 (Shuowen “completely misinterpret[ed]”) and the 12.5% unseen-character statement are in-paper evidence of the gap.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claims rest on domain assumptions about the validity of the task decomposition, the authority of the source dictionaries, and the quality of expert verification. No fitted numerical parameters or invented physical entities support the result; the only listed free parameter affects benchmark composition, not the experimental conclusion.

free parameters (1)
  • Token-length threshold for JieZi-Bench selection = 200
    Chosen empirically in Section 4.1 to retain the top 20% most informative dictionary entries; affects the composition of the benchmark but not the central experimental conclusion.
assumptions (4)
  • domain assumption The four-level decomposition of exegesis matches real scholarly practice.
    Section 3 defines ACCE from this decomposition; if it is not representative, the benchmark measures the wrong construct.
  • domain assumption The four lexicographic works used for JieZi-Bench provide correct, mutually consistent ground truth.
    Section 4.1 states the dictionaries are authoritative and mutually verifiable; the benchmark reference answers are curated from them.
  • domain assumption Qualified experts performed the claimed exhaustive verification of all 8K benchmark QA pairs.
    Section 4.2 and B.4 assert exhaustive expert verification; this is self-reported and cannot be independently confirmed.
  • domain assumption LLM-based extraction and QA generation with spot-checking yields training data with approximately 1-3% error rates.
    Section B.4 reports error rates from sampled audits; the unexamined 90% is assumed to have similar quality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of JieZi: A Large-Scale Expert-Audited Dataset and Benchmark for Ancient Chinese Character Exegesis." pith.science (2026). https://pith.science/paper/UC26MV3J

@misc{pith2026260811741,
  author       = {Pith},
  title        = {Pith review of: JieZi: A Large-Scale Expert-Audited Dataset and Benchmark for Ancient Chinese Character Exegesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UC26MV3J}},
  note         = {Machine review of arXiv:2608.11741}
}
read the original abstract

The scholarly exegesis of ancient Chinese characters demands integrating visual observation, linguistic analysis, and historical context. However, existing computational approaches focus narrowly on subtasks such as character recognition and retrieval, lacking the structured datasets and benchmarks required for comprehensive scholarly analysis. To address this limitation, we introduce Ancient Chinese Character Exegesis (ACCE), a vision-language question answering (VQA) task that models the scholarly exegesis process. ACCE is organized into four progressive levels: basic character identification, glyph-form analysis, meaning exegesis, and diachronic evolution analysis. To support this task, we construct two complementary resources. JieZi-Dataset is the first large-scale, expert-audited VQA training dataset for ACCE, comprising over 500K QA pairs. It is constructed via a pipeline that reduces factual errors by constraining generation with expert-designed templates and source-text references. Human verification is further applied at each key stage to ensure scholarly accuracy. JieZi-Bench is an evaluation benchmark aligned with the exegesis process, constructed and verified by human experts to ensure evaluation reliability. It consists of four levels with reference answers curated from authoritative lexicographic works held separate from the training data. Experiments on multimodal large language models show that current models perform well on basic identification but struggle with glyph analysis, semantic reasoning, and diachronic understanding. Fine-tuning on JieZi-Dataset substantially improves performance across all four levels. Code and dataset are available at https://github.com/Ran00w/JieZi.

Figures

Figures reproduced from arXiv: 2608.11741 by the authors.

Figure 1
Figure 1. Comparison between the Exegesis task and the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Data generation pipeline of JieZi-Dataset and JieZi-Bench. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Distribution of glyph stages in JieZi-Dataset and [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Distribution of token lengths across metadata en [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Examples of comprehensive scholarly exegesis in [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 8
Figure 8. Figure 8: Examples of responses for different tasks from [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 7
Figure 7. Figure 7: Examples of Question-Answering pairs across the [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 9
Figure 9. Figure 9: Prompt template used for structured metadata extraction from dictionary entries. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Prompt template for VQA generation. The prompt [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: Prompt template for evaluating fact alignment [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 13
Figure 13. Figure 13: An example of anomalous output from few-shot [PITH_FULL_IMAGE:figures/full_fig_p017_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

86 extracted references · 62 canonical work pages

  1. [1]

    Anthropic. 2026. Claude Opus 4.6 System Card. https://www-cdn.anthropic. com/14e4fb01875d2a69f646fa5e574dea2b1c0ff7b5.pdf

    2026

  2. [2]

    Yannis Assael, Thea Sommerschield, Alison Cooley, Brendan Shillingford, John Pavlopoulos, Priyanka Suresh, Bailey Herms, Justin Grayston, Benjamin May- nard, Nicholas Dietrich, et al. 2025. Contextualizing ancient texts with generative neural networks.Nature645, 8079 (2025), 141–147

    2025

  3. [3]

    Yannis Assael, Thea Sommerschield, Brendan Shillingford, Mahyar Bordbar, John Pavlopoulos, Marita Chatzipanagiotou, Ion Androutsopoulos, Jonathan Prag, and Nando De Freitas. 2022. Restoring and attributing ancient texts using deep neural networks.Nature603, 7900 (2022), 280–283

    2022

  4. [4]

    Nija Babu and A Soumya. 2019. Character recognition in historical handwritten documents–a survey. In2019 international conference on communication and signal processing (ICCSP). IEEE, 0299–0304

    2019

  5. [5]

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, et al. 2025. Qwen3-VL Technical Report. arXiv:2511.21631 [cs.CV] https://arxiv.org/abs/2511.21631

    arXiv 2025

  6. [6]

    Françoise Bottéro. 1996. Review of The Origin and Early Development of the Chinese Writing System, by William G. Boltz.Journal of the American Oriental Society116, 3 (1996), 574–577. https://doi.org/10.2307/605196

    work page doi:10.2307/605196 1996

  7. [7]

    Shuowen Jiezi

    Françoise Bottéro and Christoph Harbsmeier. 2008. The "Shuowen Jiezi" Dictio- nary and the Human Sciences in China.Asia Major(2008), 249–271

    2008

  8. [8]

    ByteDance Seed. 2026. Seed2.0 Model Card. https://seed.bytedance.com/seed2

    2026

Show all 86 references
  1. [9]

    ByteDance Seed Team. 2026. Seed 2.0 Official Launch. Official blog post. https://seed.bytedance.com/en/blog/seed2-0-%E6%AD%A3%E5%BC%8F% E5%8F%91%E5%B8%83 Accessed: 2026-04-09

    2026

  2. [10]

    Jiahuan Cao, Yang Liu, Peirong Zhang, Yongxin Shi, Kai Ding, and Lianwen Jin. 2025. TongGu-VL: Advancing Visual-Language Understanding in Chinese Classical Studies through Parameter Sensitivity-Guided Instruction Tuning. In Proceedings of the 33rd ACM International Conference ...

    2025

  3. [11]

    Jiahuan Cao, Yongxin Shi, Dezhi Peng, Yang Liu, and Lianwen Jin. 2024. C3Bench: A Comprehensive Classical Chinese Understanding Benchmark for Large Lan- guage Models. arXiv:2405.17732 [cs.CL] https://arxiv.org/abs/2405.17732

    2024 arXiv

  4. [12]

    Diego Chapinal-Heras and Carlos Díaz-Sánchez. 2023. A review of AI applications in Human Sciences research.Digital Applications in Archaeology and Cultural Heritage30 (2023), e00288

    2023

  5. [13]

    Zijian Chen, Tingzhu Chen, Wenjun Zhang, and Guangtao Zhai. 2024. OBI- Bench: Can LMMs aid in study of ancient script on oracle bones?arXiv preprint arXiv:2412.01175(2024)

    2024 arXiv

  6. [14]

    Yang Chi, Fausto Giunchiglia, Chuntao Li, and Hao Xu. 2024. Ancient Chinese Glyph Identification Powered by Radical Semantics. InFindings of the Association for Computational Linguistics: ACL 2024, Lun-Wei Ku, Andre Martins, and Vivek Srikumar (Eds.). Association for Computati...

    2024 doi

  7. [15]

    2017.The Oxford handbook of classical Chinese literature (1000 BCE-900 CE)

    Wiebke Denecke, Wai-yee Li, and Xiaofei Tian. 2017.The Oxford handbook of classical Chinese literature (1000 BCE-900 CE). Oxford University Press

    2017

  8. [16]

    Xiaolei Diao, Daqian Shi, Jian Li, Lida Shi, Mingzhe Yue, Ruihua Qi, Chuntao Li, and Hao Xu. 2023. Toward zero-shot character recognition: a gold standard dataset with radical-level annotations. InProceedings of the 31st ACM International Conference on Multimedia. 6869–6877

    2023

  9. [17]

    Google Gemini Team. 2025. Gemini 2.5: Pushing the Frontier with Advanced Rea- soning, Multimodality, Long Context, and Next Generation Agentic Capabilities. arXiv preprint arXiv:2507.06261(2025). https://arxiv.org/abs/2507.06261

    2025 arXiv

  10. [18]

    Google DeepMind. 2025. Gemini 3 Flash Model Card. https://storage.googleapis. com/deepmind-media/Model-Cards/Gemini-3-Flash-Model-Card.pdf

    2025

  11. [19]

    Google DeepMind. 2026. Gemini 3.1 Flash-Lite Model Card. https://deepmind. google/models/model-cards/gemini-3-1-flash-lite

    2026

  12. [20]

    Google DeepMind. 2026. Gemini 3.1 Pro Model Card. https://deepmind.google/ models/model-cards/gemini-3-1-pro

    2026

  13. [21]

    2023.Hanzi Yuanliu Dazidian [Dictionary of Chinese Character Etymology]

    Yankui Gu. 2023.Hanzi Yuanliu Dazidian [Dictionary of Chinese Character Etymology]. Language and Culture Press, Beijing, China

    2023

  14. [22]

    Haisu Guan, Jinpeng Wan, Yuliang Liu, Pengjie Wang, Kaile Zhang, Zhebin Kuang, Xinyu Wang, Xiang Bai, and Lianwen Jin. 2024. An open dataset for the evolution of oracle bone characters: EVOBC.arXiv preprint arXiv:2401.12467 (2024)

    2024 arXiv

  15. [23]

    Patrick Heinrich et al. 2020. Language modernization in the Chinese character cultural sphere: China, Japan, Korea and Vietnam. InThe Cambridge handbook of language standardization. Cambridge University Press, 576–596

    2020

  16. [24]

    Zhikai Hu, Yiu-ming Cheung, Yonggang Zhang, Peiying Zhang, and Pui-ling Tang. 2024. Component-level oracle bone inscription retrieval. InProceedings of the 2024 International Conference on Multimedia Retrieval. 647–656

    2024

  17. [25]

    Hanqi Jiang, Yi Pan, Junhao Chen, Zhengliang Liu, Yifan Zhou, Peng Shu, Yiwei Li, Huaqin Zhao, Stephen Mihm, Lewis C Howe, et al. 2024. OracleSage: Towards unified visual-linguistic understanding of oracle bone scripts through cross- modal knowledge fusion.arXiv preprint arXiv...

    2024 arXiv

  18. [26]

    Runhua Jiang, Yongge Liu, Boyuan Zhang, Xu Chen, Deng Li, and Yahong Han

  19. [27]

    2024.Ultralytics YOLO11

    Glenn Jocher and Jing Qiu. 2024.Ultralytics YOLO11. https://github.com/ ultralytics/ultralytics

    2024

  20. [28]

    David N Keightley. 1996. Art, ancestors, and the origins of writing in China. Representations56 (1996), 68–95

    1996

  21. [29]

    Guolong Lai. 2019. On [Can] and [Xie]: Two Different Approaches to the Interpre- tation of Ancient Chinese Characters, Form-Oriented and Integrated Phonology- Form-Semantics.Bulletin of the Jao Tsung-I Academy of Sinology6, 1 (2019), 187–224

    2019

  22. [30]

    Bang Li, Donghao Luo, Yujie Liang, Jing Yang, Zengmao Ding, Xu Peng, Boyuan Jiang, Shengwei Han, Dan Sui, Peichao Qin, et al. 2024. Oracle bone inscriptions multi-modal dataset.arXiv preprint arXiv:2407.03900(2024)

    2024 arXiv

  23. [31]

    Bang Li, Jing Yang, Yujie Liang, Xiaobin Hu, Zengmao Ding, Xu Peng, Shengwei Han, Peichao Qin, Donghao Luo, Taisong Jin, et al. 2026. OBIMD: A Multi-modal Dataset for Contextual Interpretation of Oracle Bone Inscriptions.Scientific Data (2026)

    2026

  24. [32]

    Jing Li, Xueke Chi, Qiufeng Wang, Dahan Wang, Kaizhu Huang, Yongge Liu, and Cheng-Lin Liu. 2024. A comprehensive survey of oracle character recognition: challenges, benchmarks, and beyond. arXiv:2411.11354 [cs.CV] https://arxiv. org/abs/2411.11354

    2024 arXiv

  25. [33]

    Yang Liu, Jiahuan Cao, Hiuyi Cheng, Yongxin Shi, Kai Ding, and Lianwen Jin

  26. [34]

    Minesh Mathew, Dimosthenis Karatzas, and C.V. Jawahar. 2021. DocVQA: A Dataset for VQA on Document Images. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (W ACV). 2200–2209

    2021

  27. [35]

    Ian R. McKenzie, Alexander Lyzhov, Michael Martin Pieler, Alicia Parrish, Aaron Mueller, Ameya Prabhu, Euan McLean, Xudong Shen, Joe Cavanagh, Andrew George Gritsevskiy, Derik Kauffman, Aaron T. Kirtland, Zhengping Zhou, Yuhui Zhang, Sicong Huang, Daniel Wurgaft, Max Weiss, Al...

    2023

  28. [36]

    Sewon Min, Xinxi Lyu, Ari Holtzman, Mikel Artetxe, Mike Lewis, Hannaneh Hajishirzi, and Luke Zettlemoyer. 2022. Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?. InProceedings of the 2022 Conference on Empirical Methods in Natural Language Processing

    2022

  29. [37]

    Catherine Olsson et al. 2022. In-context Learning and Induction Heads. Trans- former Circuits

    2022

  30. [38]

    OpenAI. 2026. GPT-5.4 Thinking System Card. https://deploymentsafety.openai. com/gpt-5-4-thinking/gpt-5-4-thinking.pdf

    2026

  31. [39]

    Kaixin Peng, Mengyang Zhao, Haiyang Yu, Teng Fu, and Bin Li. 2025. Inter- pretable Oracle Bone Script Decipherment through Radical and Pictographic Analysis with LVLMs.arXiv preprint arXiv:2508.10113(2025)

    2025 arXiv

  32. [40]

    Runqi Qiao, Qiuna Tan, Guanting Dong, Minhui Wu, Jiapeng Wang, Yifan Zhang, Zhuoma GongQue, Chong Sun, Yida Xu, Yadong Xue, et al . 2025. V-Oracle: Making progressive reasoning in deciphering oracle bones for you and me. In Proceedings of the 63rd Annual Meeting of the Associa...

    2025

  33. [41]

    Qwen Team. 2026. Qwen3.5: Towards Native Multimodal Agents. https: //qwen.ai/blog?id=qwen3.5

    2026

  34. [42]

    2020.A phonological history of Chinese

    Zhongwei Shen. 2020.A phonological history of Chinese. Cambridge University Press

    2020

  35. [43]

    Yongxin Shi, Chongyu Liu, Dezhi Peng, Cheng Jian, Jiarong Huang, and Lianwen Jin. 2023. M5HisDoc: A large-scale multi-style Chinese historical document analysis benchmark.Advances in Neural Information Processing Systems36 (2023), 78483–78495

    2023

  36. [44]

    Adam D Smith. 2017. Early Chinese manuscript writings for the name of the Sage Emperor Shun, and the legacy of Warring States-period orthographic variation in early Chinese received texts.Early China40 (2017), 63–88

    2017

  37. [45]

    Thea Sommerschield, Yannis Assael, John Pavlopoulos, Vanessa Stefanak, An- drew Senior, Chris Dyer, John Bodel, Jonathan Prag, Ion Androutsopoulos, and Nando De Freitas. 2023. Machine learning for ancient languages: A survey. Computational Linguistics49, 3 (2023), 703–747

    2023

  38. [46]

    Ken-ichi Takashima. 2021. Some methodological issues in reading oracle-bone inscriptions: In particular reference to the Huayuanzhuang Locus East Collection. Bulletin of Chinese Linguistics14, 1 (2021), 1–41

    2021

  39. [47]

    Kimi Team, Tongtong Bai, Yifan Bai, Yiping Bao, S. H. Cai, Yuan Cao, Y. Charles, H. S. Che, Cheng Chen, et al . 2026. Kimi K2.5: Visual Agentic Intelligence. arXiv:2602.02276 [cs.CL] https://arxiv.org/abs/2602.02276

    2026 arXiv

  40. [48]

    V Team, Wenyi Hong, Wenmeng Yu, Xiaotao Gu, Guo Wang, Guobing Gan, Haomiao Tang, Jiale Cheng, Ji Qi, Junhui Ji, et al. 2025. GLM-4.5V and GLM-4.1V- Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Li, et al. Learning. arXiv:2507.01006 [cs.CV] https:...

    2025 arXiv

  41. [49]

    UNESCO. [n. d.]. Chinese Oracle-Bone Inscriptions. https://www.unesco.org/ en/memory-world/chinese-oracle-bone-inscriptions

  42. [50]

    C. J. van Rijsbergen. 1979.Information Retrieval. Butterworths, London

    1979

  43. [51]

    Mei Wang and Weihong Deng. 2022. Oracle-MNIST: a realistic image dataset for benchmarking machine learning algorithms.arXiv preprint arXiv:2205.09442 (2022)

    2022 arXiv

  44. [52]

    Pengjie Wang, Kaile Zhang, Xinyu Wang, Shengwei Han, Yongge Liu, Lianwen Jin, Xiang Bai, and Yuliang Liu. 2024. Puzzle Pieces Picker: Deciphering Ancient Chinese Characters with Radical Reconstruction. InDocument Analysis and Recognition – ICDAR 2024 (Lecture Notes in Computer...

    2024 doi

  45. [53]

    Pengjie Wang, Kaile Zhang, Xinyu Wang, Shengwei Han, Yongge Liu, Jinpeng Wan, Haisu Guan, Zhebin Kuang, Lianwen Jin, Xiang Bai, et al. 2024. An open dataset for oracle bone character recognition and decipherment.Scientific Data 11, 1 (2024), 976

    2024

  46. [55]

    Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, et al . 2025. InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Effi- ciency.arXiv preprint arXiv:2508.18265(2025)

    2025 arXiv

  47. [56]

    Jason Wei, Najoung Kim, Yi Tay, and Quoc V Le. 2023. Inverse Scaling Can Become U-Shaped. InEMNLP

    2023

  48. [57]

    Yuting Wei, Yuanxing Xu, Xinru Wei, Simin Yang, Yangfu Zhu, Yuqing Li, Di Liu, and Bin Wu. 2024. AC-EVAL: Evaluating Ancient Chinese Language Understand- ing in Large Language Models. InFindings of the Association for Computational Linguistics: EMNLP 2024, Yaser Al-Onaizan, Mo...

    2024

  49. [58]

    Crispin Williams. 2014. Scribal variation and the meaning of the houma and wenxian covenant texts’imprecation ma yi fei shi.Early China37 (2014), 101–179

    2014

  50. [59]

    Qiu Xigui. 1985. On the Methods of Studying Ancient Chinese Script.Early China11 (1985), 301–316

    1985

  51. [60]

    Wen Xing. 2011. Paleographic, Historical, and Intellectual History Approaches to Warring States Manuscripts Written on Bamboo Slips: A Review Article.Early China33 (2011), 233–262. https://doi.org/10.1017/S0362502800000298

    2011 doi

  52. [61]

    Tai-Ling Yuan, Zhe Zhu, Kun Xu, Cheng-Jun Li, Tai-Jiang Mu, and Shi-Min Hu

  53. [62]

    Xiang Yue, Yuansheng Ni, Kai Zhang, Tianyu Zheng, Ruoqi Liu, Ge Zhang, Samuel Stevens, Dongfu Jiang, Weiming Ren, Yuxuan Sun, et al. 2024. MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGI. InProceedings of the IEEE/CVF Conference...

    2024

  54. [63]

    Biao Zhang, Zhongtao Liu, Colin Cherry, and Orhan Firat. 2024. When Scaling Meets LLM Finetuning: The Effect of Data, Model and Finetuning Method. In ICLR

    2024

  55. [64]

    Chongsheng Zhang, Ruixing Zong, Shuang Cao, Yi Men, and Bofeng Mo. 2020. AI-Powered Oracle Bone Inscriptions Recognition and Fragments Rejoining. InProceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence (IJCAI-20). 5309–5311. https://doi.org/...

    2020 doi

  56. [65]

    Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q Weinberger, and Yoav Artzi

  57. [66]

    the human back/spine

    Yuyi Zhang, Yongxin Shi, Peirong Zhang, Yixin Zhao, Zhenhua Yang, and Lian- wen Jin. 2025. MegaHan97K: A large-scale dataset for mega-category Chinese character recognition with over 97K categories.Pattern Recognition167 (2025), 111757. JieZi: A Large-Scale Expert-Audited Data...

    2025

  58. [72]

    Please regenerate the same JSON completely, and it must be strictly parseable JSON

    Extra system prompt used in retry with larger max_tokens Your previous output may have been truncated due to length limits, causing incomplete JSON. Please regenerate the same JSON completely, and it must be strictly parseable JSON. To reduce length, output compact JSON (no in...

  59. [73]

    Output only one valid JSON object, and you may include an extra field `next_index` (integer)

    Extra system prompt used in paged fallback You must use paged output mode. Output only one valid JSON object, and you may include an extra field `next_index` (integer). Starting from index {start_index}, output at most {page_size} `meaning_chunks` items in the original numbere...

  60. [74]

    Output only one JSON object

  61. [75]

    Use exactly these fields: glyph type, character formation method, structure, special structure, components, original meaning, diachronic glyph evolution

  62. [76]

    glyph type

    The field "glyph type" must be consistent with the sample label "{glyph type}"

  63. [77]

    structure

    "structure" and "components" must follow the actual form in the current image stage

  64. [78]

    original meaning

    "original meaning" and "diachronic glyph evolution" must be grounded in context, especially historical divergence, later forms, and simplification merges

  65. [79]

    original meaning

    For simplified-form images, modern mainstream meaning is allowed for "original meaning", but the source chain must be explicit in "diachronic glyph evolution"

  66. [80]

    If weak hints conflict with the image, trust the image

  67. [81]

    original meaning

    "original meaning" must align with the candidate list first; only minor paraphrase is allowed when strongly supported

  68. [82]

    structure

    If "structure" is "single component", "special structure" must be a short description; otherwise it must be an empty string

  69. [83]

    historical character shape evolution

    If the same component appears multiple times in different positions, split it into separate keys with position suffixes. Recommended internal order: - Identify the glyph stage. - Identify the actual structure in this stage. - Identify visible components and their functions. - ...

  70. [84]

    Rely SOLELY on the image; no external knowledge

  71. [85]

    Never hallucinate unsupported details

  72. [86]

    Output exactly ONE raw JSON object (NO markdown, NO comments, NO extra text)

  73. [87]

    构件": { "构件名

    Prioritize valid fields, aligned content, and conservative analysis. … # Required JSON Format 构件: Object of 1-4 key components. Keys: Single characters/radicals (use brief desc ONLY if unnameable). Values: Objects containing exactly: 功能: String or Array of: 表意, 表音, 符号. 演变类型: C...

  74. [1617]

    https://doi.org/10.18653/v1/2024.findings-emnlp.87

    2024 doi

  75. [2019]

    A large Chinese text dataset in the wild.Journal of Computer Science and Technology34, 3 (2019), 509–521

    2019

  76. [2020]

    InInternational Con- ference on Learning Representations

    BERTScore: Evaluating Text Generation with BERT. InInternational Con- ference on Learning Representations

  77. [2023]

    In Proceedings of the 31st ACM International Conference on Multimedia

    OraclePoints: A Hybrid Neural Representation for Oracle Character. In Proceedings of the 31st ACM International Conference on Multimedia. 7901–7911. https://doi.org/10.1145/3581783.3612534

  78. [2025]

    InProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

    MCS-Bench: A Comprehensive Benchmark for Evaluating Multimodal Large Language Models in Chinese Classical Studies. InProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 10435–10492

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.