课题基金基金详情
融合语言学特征的跨主题作文自动评分关键技术研究
结题报告
批准号:
61976062
项目类别:
面上项目
资助金额:
56.0 万元
负责人:
李霞
依托单位:
学科分类:
自然语言处理
结题年份:
2023
批准年份:
2019
项目状态:
已结题
项目参与者:
李霞
国基评审专家1V1指导 中标率高出同行96.8%
结合最新热点,提供专业选题建议
深度指导申报书撰写,确保创新可行
指导项目中标800+,快速提高中标率
客服二维码
微信扫码咨询
中文摘要
英文作文自动评分技术在教育领域具有广泛的应用和重要的研究价值。现有英文作文自动评分研究主要面向具有已评分训练数据的主题作文,对没有任何已评分训练数据的跨主题作文自动评分研究、尤其对适应于中国英语学习者作文的跨主题英文作文自动评分的相关研究还不多见。本项目将围绕跨主题英文作文自动评分中的关键技术开展研究,主要包括以下几个方面:(1)研究基于主题无关特征判别的端到端跨主题作文自动评分算法;(2)研究融合深度语言文本特征及特征选择的跨主题作文自动评分算法;(3)研究融合聚类方法的跨主题作文自动评分算法;(4)研究面向中国英语学习者特定模式的跨主题作文自动评分方法,包括:基于n-grams字符编码的跨主题作文自动评分算法、基于作文短语级比较的跨主题作文自动评分算法、融合题目信息及孪生网络结构的跨主题作文自动评分算法。研究内容同时将推动在中文跨主题作文自动评分等相关场景中的应用。
英文摘要
Automated Essay Scoring (AES) technology is widely used in the field of education, and has important application research value. The existing works of automated essay scoring are mainly for prompt-dependent scenes, in which sufficient pre-scored essays are needed for training the models. However, there are few works for cross-prompt scenes, in which we have no any pre-scored essays for training, especially for those essays written by Chinese English learners. Based on the needs of the cross-prompt automated essay scoring in the educational scene, the purpose of this project is to study and find the effective cross-prompt automated essay scoring algorithms and methods for general English essays scoring and English essays written by Chinese English learners. In this project, four aspects of research are to be studied: .1) As the distribution of essays from different prompts are different, we need to find a method to discern the prompt-independent feature for cross-prompt automated essay scoring. Thus, for the first, we will study an end-to-end neural-network based cross-prompt automated essay scoring algorithm based on prompt-independent feature discrimination..2) In order to make the scoring model has a better interpretability, the project intends to study the cross-prompt automated essay scoring algorithm based on Coh-Metrix and feature selection method..3) Since clustering algorithm performs well in unsupervised machine learning, we propose to use the clustering algorithm in cross-prompt automated essay scoring task. Based on the consistency of the prompt-independent features of essays from source prompt and target prompt, we intend to study the cross-prompt automated essay scoring method based on the sample metrics and clustering algorithm..4) On the base of the particularity of Chinese English learners' writing, this project proposes to study the effective cross-prompt automated essay scoring algorithms for Chinese English learners, including: cross-prompt automated essay scoring algorithm based on n-grams letters encoding, cross-prompt automated essay scoring algorithm based on text phrases level comparison and cross-prompt automated essay scoring algorithm based on topic information monitoring and Siamese Network.
期刊论文列表
专著列表
科研奖励列表
会议论文列表
专利列表
DOI:--
发表时间:2020
期刊:中文信息学报
影响因子:--
作者:李霞;马骏腾;覃世豪
通讯作者:覃世豪
DOI:10.1016/j.ins.2023.03.102
发表时间:2023-03
期刊:Inf. Sci.
影响因子:--
作者:Yongqiang Zheng;xia li;Jian-Yun Nie
通讯作者:Yongqiang Zheng;xia li;Jian-Yun Nie
DOI:10.1016/j.knosys.2020.106491
发表时间:2020-12-27
期刊:KNOWLEDGE-BASED SYSTEMS
影响因子:8.8
作者:Li, Xia;Chen, Minping;Nie, Jian-Yun
通讯作者:Nie, Jian-Yun
面向中国英语学习者的英文作文全自动评分及诊断反馈技术研究
  • 批准号:
    61402119
  • 项目类别:
    青年科学基金项目
  • 资助金额:
    24.0万元
  • 批准年份:
    2014
  • 负责人:
    李霞
  • 依托单位:
国内基金
海外基金