查詢結果分析
來源資料
頁籤選單縮合
| 題 名 | 詞類原型範疇數據化初探--以副詞維度為例=A Preliminary Study on the Dataization of Prototype Categories of Word Classes: Taking Adverbs as an Example |
|---|---|
| 作 者 | 楊苛鑫; 莊會彬; | 書刊名 | 中國語文通訊 |
| 卷 期 | 104:1 2025.01[民114.01] |
| 頁 次 | 頁43-54 |
| 分類號 | 802.631 |
| 關鍵詞 | 大語言模型; 詞類; 原型範疇; 副詞; Large language models; Word class; Prototypes categories; Adverbs; |
| 語 文 | 中文(Chinese) |
| DOI | 10.29499/CrCL.202501_104(1).0003 |
| 中文摘要 | 學界很早就提出了從原型範疇的角度去分析詞類,如《漢語詞類劃分手冊》使用人工量表, 根據得分刻畫詞類範疇。隨著大語言模型的出現,學界呼籲將人工智能與語言研究相結合, 因此在研究中採用 TCBert 大模型對詞從副詞維度進行賦分,以得出不同詞在副詞維度的 隸屬度。同時以人工標注的 UD 樹庫為參考,基於概率配價模式理論,計算不同詞作狀語 的概率,從統計角度計算詞在副詞維度的隸屬度,為大模型與語言研究相結合提供思路與 探索。 |
| 英文摘要 | The approach of analyzing word classes from the perspective of prototype categories is well-established in the academic community. For example, the Chinese Word Class Classification Manual categorizes words based on the scores assigned manually through dedicated scales. With the rise of large language models (LLMs), there is growing interest in applying artificial intelligence to language research. In this study, we use the TCBert large model to score selected words, assessing their degree of adverbiality, instead of relying on manual scales. Additionally, we adopt the theory of Probabilistic Valency Pattern and utilize the manually annotated UD treebank to calculate the percentage of these words functioning as adverbs, thereby determining how strongly they belong to the adverb class. As an exploratory effort, it is hoped that this study can offer inspiration in applying LLMs in language research. |
本系統中英文摘要資訊取自各篇刊載內容。