漢學家共同作者邀請

主旨： 邀請共同作者：道教 LLM 評測框架白皮書（不涉經費）

{{稱呼}}{{老師}}　道鑒：

冒昧致函。在下劉啟泳，台灣道士、劉厝派傳承，目前主持「[鼎稔道學館](https://lius.cc)」這個開放道教知識基礎建設專案。

近日我們釋出全球第一個開源的**道教專業大語言模型 Daoism-Qwen3.5-9B**，連同 95,919 條目的知識庫與公開 RAG API，三件套全部 Apache 2.0 / CC0 開源。詳情見：

長期關注 {{您讀過他/她的代表作}}，覺得您對 {{一句話原因}} 的觀點，正是我們現在最需要的學術嚴謹度。

正事： 我們已釋出《DaoEval 白皮書 v0.1》——首個開源的道教知識 LLM 評測框架（500 黃金樣例、ABCD 信賴度分級、底本欄位與 ID 規範），已透過 Zenodo 取得 DOI [10.5281/zenodo.20248697](https://doi.org/10.5281/zenodo.20248697) 永久存檔。

誠摯邀請您擔任**共同作者**；若您希望承諾更輕，亦可僅列為**致謝顧問**。**不涉及任何經費**，無金錢往來，純學術合作。

您的貢獻：

對等回報：

整體預估佔用您 **2-3 小時**，分散在 4 週內。

為何我們認為值得： 現有評測（MMLU / CMMLU / C-Eval）對道教知識覆蓋極淺、且多為選擇題。DaoEval 是首個結構化開源框架，涵蓋「文獻溯源、跨派詮釋、儀式準確性」三大面向。

若此議題與您當前研究方向相合，方便的話可寄上白皮書 v0.1 PDF，並安排 30 分鐘 Zoom 細談。

順頌

道安

劉啟泳　謹啟

鼎稔道學館（lius.cc）維護者

[email protected]

**Subject:** Invitation to co-author an open Daoism-LLM evaluation paper (no funding required)

Dear Professor {{TITLE}} {{LAST_NAME}},

I'm writing as the maintainer of [lius.cc](https://lius.cc) (鼎稔道學館 / Dingren Daoxue Lab), a Taiwan-based open Daoism knowledge initiative. We have just released **Daoism-Qwen3.5-9B** — the first open-source large language model specialized for Taoist knowledge — together with a 95,919-node curated knowledge base and a public RAG API. All three artefacts are Apache 2.0 / CC0.

I have followed {{THEIR_RELATED_WORK}} with great interest, and your perspective on {{ONE_LINE_REASON}} is exactly the kind of academic rigour our preprint needs.

**The ask:** We have just released the **DaoEval Whitepaper v0.1** — an evaluation framework for Daoist-knowledge LLMs (500 gold-standard examples across 8 categories, ABCD confidence grading, base-text & ID conventions). It is archived on Zenodo with DOI [10.5281/zenodo.20248697](https://doi.org/10.5281/zenodo.20248697). We would be honoured to invite you as a **co-author** (or, if you prefer lower commitment, as an **acknowledged advisor**) on the v0.2 update and the subsequent journal version.

We **do not request funding from your side**. Your contribution:

In return:

Expected engagement: 2–3 hours total over 4 weeks.

**Why this matters:** Existing evaluation suites (MMLU, CMMLU, C-Eval) cover Daoism thinly and only via multiple-choice. DaoEval is the first structured open framework covering scripture provenance, cross-sect interpretation, and ritual fidelity.

If aligned with your interests, I'd be happy to send the v0.1 preprint PDF and arrange a 30-min Zoom.

With gratitude,

**Liu, Chi-Ying** 劉啟泳

Maintainer · Dingren Daoxue Lab (lius.cc)

[email protected]