融合篇章成分识别的中文记叙文篇章结构测评  

Evaluation of Chinese Narrative Text Structure by Integrating Text Component Identification

作　　者：王晓艺 王锦丞 刘杰 WANG Xiaoyi;WANG Jincheng;LIU Jie(China Language Intelligence Research Center,Capital Normal University,Beijing 100048,China;School of Literature,Capital Normal University,Beijing 100048,China;Beijing Institute of Control and Electronics Technology,Beijing 100038,China;School of Information Science,North China University of Technology,Beijing 100144,China)

机构地区：[1]首都师范大学中国语言智能研究中心,北京100048 [2]首都师范大学文学院,北京100048 [3]北京控制与电子技术研究所,北京100038 [4]北方工业大学信息学院,北京100144

出　　处：《小型微型计算机系统》2025年第1期55-63,共9页Journal of Chinese Computer Systems

基　　金：国家科技创新2030-“新一代人工智能”重大项目(2020AAA0109700)资助;国家自然科学基金项目(62076167)资助;国家语委十四五规划项目(YB145-16)资助。

摘　　要：篇章结构合理性是作文自动评测中的重要考量维度.当前针对记叙文篇章结构合理性自动评估的研究还处于起步阶段.该文通过与语文教学专家共同制定记叙文篇章结构评价标准与篇章成分标注规范,构建了具有一定规模的中文记叙文篇章结构测评语料库.基于该语料库,该文提出了融合篇章成分识别的记叙文篇章结构测评模型.模型利用深度学习算法和注意力机制从单词、句子、段落3个层次学习文章特征,从而提取篇章结构的重要信息,最后通过融合识别的篇章成分结果进行结构合理性评分.利用构建的记叙文篇章结构语料库进行实验,结果表明,该文提出的模型准确率达到79.6%,优于现有工作和基线模型.The rationality of the text structure is an important dimension in automatic essays scoring.The research on automatic evaluation of the text structure in narrative essay is still in its infancy.In this paper,by working with experts to formulate text structure evaluation criteria and text component labeling specifications,a large-scale Chinese narrative text structure evaluation corpus is constructed.Based on this corpus,this paper proposes an evaluation model of narrative text structure that integrates text component identification.The model uses deep learning algorithm and attention mechanism to learn text features from the three levels of words,sentences and paragraphs,so as to extract important information about the text structure,and finally score the structural rationality by integrating the identified text components.The experimental results on the constructed narrative text structure corpus show that the accuracy of the proposed model is better than the existing work and baseline models,reaching 79.6%.

关 键 词：篇章结构合理性 语料库 作文自动评分 层次注意力网络 BERT 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]