--- title: "一种改进的汉语语义角色分类体系与标注实践 — 宋衡 2023" description: "Chinese semantic role labeling lacks a common role inventory." authors: ["宋衡", "曹存根", "王亚", "王石"] year: 2023 venue: "中文信息学报 (Journal of Chinese Information Processing)" source: "http://jcip.cipsc.org.cn/article/id/zwxxxb_3455" licence: "Free access (jcip.cipsc.org.cn), no open licence stated — readonly; © 中文信息学报" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/papers/song-2023-fine-grained-annotated/" --- [Elamur Research](https://md.research.elamur.ai/) · [this page in HTML](https://research.elamur.ai/free/papers/song-2023-fine-grained-annotated/) [home](https://md.research.elamur.ai/) / [papers](https://md.research.elamur.ai/free/papers/) / [2023](https://md.research.elamur.ai/free/papers/2023/page/7/) / 一种改进的汉语语义角色分类体系与标注实践 # 一种改进的汉语语义角色分类体系与标注实践 **Authors:** 宋衡, 曹存根, 王亚, 王石 · **Year:** 2023 · **Venue:** 中文信息学报 (Journal of Chinese Information Processing) · **Licence:** Free access (jcip.cipsc.org.cn), no open licence stated — readonly; © 中文信息学报 **Links:** [Publisher's page](http://jcip.cipsc.org.cn/article/id/zwxxxb_3455) · [PDF](http://jcip.cipsc.org.cn/cn/article/pdf/preview/zwxxxb_3455.pdf) ## Summary Chinese semantic role labeling lacks a common role inventory. Existing taxonomies differ in how many peripheral roles they define and what those roles mean. The authors found these gaps while analysing their Chinese Semantic Corpus (CSC), which contains web text on cars, digital products, news and encyclopedia entries. Inconsistent annotations there also led to errors in FSTD semantic grammars. They propose a revised taxonomy with three parts: - **Pivot roles:** taken from Lu Chuan and Wang Ya. - **Principal peripheral roles:** the main peripheral layer. - **Auxiliary peripheral roles:** mark who gains or loses in an event, and can be added on top of a principal role. Each principal role comes with explicit decision criteria and example sentences to reduce subjective judgement. The paper also gives consistency rules, including: - semi-automatic identification of pivots; - handling of compound verbs as pivots; - guidance for easily confused roles. The authors annotated CSC under the new scheme, checked inter-annotator agreement, and compared the results with Lu Chuan-based labels. The finer distinctions matter for web text analysis, knowledge extraction and commonsense acquisition. ## Key points - **Peripheral role inventory:** Lu Chuan's 8 first-level classes are extended with a new Descriptive class, giving 32 second-level principal peripheral roles. - 19 roles are reused unchanged. - 6 are new: Hiring Agent, Resulting Agent, Change, Attribute, Property, Attribute-value. - 7 are redefined: Patient, Entity, Origin, Destination, Target, Reliance, Location. - **Auxiliary and pivot roles:** the auxiliary roles are Benefactive and Malefactive. Pivot roles follow a 26-category scheme. - **Annotated data:** the English abstract reports 9,550 sentences with 9,423 pivot, 29,142 principal peripheral and 3,745 auxiliary peripheral roles. The Chinese introduction instead gives 9,541 sentences for CSC. - **Errors in the earlier corpus:** CSC had labeling error rates of about 17.0% for 客事 and 14.86% for 受事. - **Model comparison and outlook:** a Bi-LSTM+CRF model compares the dataset with the Chinese Proposition Bank. In the comparison with Lu Chuan's scheme, 26.37% of Change labels had been Causer (致事). The authors plan to release CSC later. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* Fields: [Linguistics & languages](https://md.research.elamur.ai/free/shelves/linguistics/) · [Natural language processing](https://md.research.elamur.ai/free/shelves/nlp/) ## Cite The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @article{2023, title = {{一种改进的汉语语义角色分类体系与标注实践}}, author = {宋衡 and 曹存根 and 王亚 and 王石}, journal = {中文信息学报 (Journal of Chinese Information Processing)}, year = {2023}, url = {http://jcip.cipsc.org.cn/article/id/zwxxxb_3455} } ``` - [← previous ニューラル記号推論における推論過程の教示方法](https://md.research.elamur.ai/free/papers/aoki-2023-how-teach-reasoning-process/) - [next → 与知识相结合的提示学习研究综述](https://md.research.elamur.ai/free/papers/bao-2023-prompt-learning-combined/) - [↑ back to papers of 2023](https://md.research.elamur.ai/free/papers/2023/page/7/) ## References 7 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. [ 1] WAITE W.Seman t i cana l s i s[M ].Be r l i n:Sp r i nge r y , Pr e s s 1974. [find](https://scholar.google.com/scholar?q=%5B+1%5D+WAITE+W%2ESeman+t+i+cana+l+s+i+s%5BM+%5D%2EBe+r+l+i+n%3ASp+r+i+nge+r+y+%2C+Pr+e+s+s+1974%2E) 2. [ 2] 刘亚慧,杨浩苹,李正华,等 . 一种轻量级的汉语语义 角色标注规范 [ J]. 中 文 信 息 学 报,2020,34( 4):10 - 20. [find](https://scholar.google.com/scholar?q=%5B+2%5D+%E5%88%98%E4%BA%9A%E6%85%A7%2C%E6%9D%A8%E6%B5%A9%E8%8B%B9%2C%E6%9D%8E%E6%AD%A3%E5%8D%8E%2C%E7%AD%89+%2E+%E4%B8%80%E7%A7%8D%E8%BD%BB%E9%87%8F%E7%BA%A7%E7%9A%84%E6%B1%89%E8%AF%AD%E8%AF%AD%E4%B9%89+%E8%A7%92%E8%89%B2%E6%A0%87%E6%B3%A8%E8%A7%84%E8%8C%83+%5B+J%5D%2E+%E4%B8%AD+%E6%96%87+%E4%BF%A1+%E6%81%AF+%E5%AD%A6+%E6%8A%A5%2C2020%2C34%28+4%29%3A10+%2D+20%2E) 3. [ 3] MÀRQUEZL,CARRERAS X,LITKOWSKIK C, e ta l.Seman t i cr o l el abe l i ng:ani n t r oduc t i ont ot he spe c i a li s sue[ J].Compu t a t i ona lL i ngu i s t i c s,2008,34 ( 2):145 - 159. [find](https://scholar.google.com/scholar?q=%5B+3%5D+M%C3%80RQUEZL%2CCARRERAS+X%2CLITKOWSKIK+C%2C+e+ta+l%2ESeman+t+i+cr+o+l+el+abe+l+i+ng%3Aani+n+t+r+oduc+t+i+ont+ot+he+spe+c+i+a+li+s+sue%5B+J%5D%2ECompu+t+a+t+i+ona+lL+i+ngu+i+s+t+i+c+s%2C2008%2C34+%28+2%29%3A145+%2D+159%2E) 4. [ 4] 朱晓亚 . 现代汉语句模研究[M]. 北京:北京 大 学 出 版 社,2010. [find](https://scholar.google.com/scholar?q=%5B+4%5D+%E6%9C%B1%E6%99%93%E4%BA%9A+%2E+%E7%8E%B0%E4%BB%A3%E6%B1%89%E8%AF%AD%E5%8F%A5%E6%A8%A1%E7%A0%94%E7%A9%B6%5BM%5D%2E+%E5%8C%97%E4%BA%AC%3A%E5%8C%97%E4%BA%AC+%E5%A4%A7+%E5%AD%A6+%E5%87%BA+%E7%89%88+%E7%A4%BE%2C2010%2E) 5. [ 5] 袁毓林 . 基于认知的 汉 语 计 算 语 言 学 研 究 [M]. 北 京: 北京大学出版社,2008. [find](https://scholar.google.com/scholar?q=%5B+5%5D+%E8%A2%81%E6%AF%93%E6%9E%97+%2E+%E5%9F%BA%E4%BA%8E%E8%AE%A4%E7%9F%A5%E7%9A%84+%E6%B1%89+%E8%AF%AD+%E8%AE%A1+%E7%AE%97+%E8%AF%AD+%E8%A8%80+%E5%AD%A6+%E7%A0%94+%E7%A9%B6+%5BM%5D%2E+%E5%8C%97+%E4%BA%AC%3A+%E5%8C%97%E4%BA%AC%E5%A4%A7%E5%AD%A6%E5%87%BA%E7%89%88%E7%A4%BE%2C2008%2E) 6. [ 6] 鲁川 . 知识 工 程 语 言 学 [M]. 北 京:清 华 大 学 出 版 社, 2010. [find](https://scholar.google.com/scholar?q=%5B+6%5D+%E9%B2%81%E5%B7%9D+%2E+%E7%9F%A5%E8%AF%86+%E5%B7%A5+%E7%A8%8B+%E8%AF%AD+%E8%A8%80+%E5%AD%A6+%5BM%5D%2E+%E5%8C%97+%E4%BA%AC%3A%E6%B8%85+%E5%8D%8E+%E5%A4%A7+%E5%AD%A6+%E5%87%BA+%E7%89%88+%E7%A4%BE%2C+2010%2E) 7. [ 7] 刘茂福,胡慧 君 . 基 于 认 知 与 计 算 的 事 件 语 义 学 研 究 [M]. 北京:科学出版社,2013. [find](https://scholar.google.com/scholar?q=%5B+7%5D+%E5%88%98%E8%8C%82%E7%A6%8F%2C%E8%83%A1%E6%85%A7+%E5%90%9B+%2E+%E5%9F%BA+%E4%BA%8E+%E8%AE%A4+%E7%9F%A5+%E4%B8%8E+%E8%AE%A1+%E7%AE%97+%E7%9A%84+%E4%BA%8B+%E4%BB%B6+%E8%AF%AD+%E4%B9%89+%E5%AD%A6+%E7%A0%94+%E7%A9%B6+%5BM%5D%2E+%E5%8C%97%E4%BA%AC%3A%E7%A7%91%E5%AD%A6%E5%87%BA%E7%89%88%E7%A4%BE%2C2013%2E)