基于多特征融合和机器学习理论的英语机考作文自动评分系统  

An Automatic Scoring System of English Machine Test Composition Based on Multi-feature Fusion and Machine Learning Theory

作　　者：李俊娴 LI Junxian(School of Railway Transportation, Shaanxi College of Communication Technology, Xi’an 710018, China)

机构地区：[1]陕西交通职业技术学院轨道交通学院,陕西西安710018

出　　处：《微型电脑应用》2021年第10期157-160,共4页Microcomputer Applications

摘　　要：传统作文自动评分系词汇相似度计算结果影响系统评分逻辑,导致评价不准确,为此设计了多特征融合和机器学习理论的英语机考作文自动评分系统。在硬件设计方面设计了控制电路实现对评分系统的全局控制。软件设计方面采用多特征融合技术跟踪英文词汇,并计算目标词汇和评分标准词汇之间的相似度;利用机器学习中的逻辑回归算法,设计作文自动评分逻辑。通过对比实验,对未跑题、跑题两类英语作文进行自动评分,测试结果表明,这一评分结果较准确,关键词汇识别较精准,具有优越性。In traditional automatic scoring system,lexical similarity calculation results affect the scoring logic of the system,resulting in inaccurate evaluation.Therefore,an automatic scoring system for English computer-based examination based on multi-feature fusion and machine learning theory is designed.In hardware design,the control circuit is designed to realize the global control of the scoring system.In terms of software design,multi-feature fusion technology is used to track English vocabulary,and the similarity between target vocabulary and scoring standard vocabulary is calculated;the logic regression algorithm in machine learning is used to design the automatic scoring logic of composition.The results show that the score of this study is more accurate and the key words recognition is more accurate,which verifies the superiority of this study.

关 键 词：多特征融合 自动评分系统 机器学习理论 余弦相似度 英语机考作文 

分 类 号：H315[语言文字—英语]