arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.09540cs.CLcs.AI
收藏

利用ChatGPT为作文及短形式建构作答评分

Using ChatGPT to Score Essays and Short-Form Constructed Responses

Performance Assessment Analytics, LLC(绩效评估分析有限责任公司)

Mark D. Shermis

AI总结:

本研究探究ChatGPT大语言模型的作文与短作答评分能力,对比ASAP竞赛的人机评分准确率,发现其梯度提升模型部分数据集表现接近人类但整体不稳定,可补充人工评分但需优化以适配高利害评估。

摘要:

本研究旨在确定ChatGPT的大语言模型能否达到ASAP竞赛中人工评分与机器评分的准确率水平。研究聚焦于多种预测模型,包括线性回归、随机森林、梯度提升(gradient boost)与boost模型。研究采用二次加权kappa(QWK)指标,将ChatGPT的表现与人类评分者的结果进行对比评估。结果显示,尽管ChatGPT的梯度提升模型在部分数据集上取得的QWK值接近人类评分者,但整体表现不稳定,且通常低于人工评分水平。研究强调需对模型做进一步优化,尤其是在处理偏差、保障评分公平性方面。尽管存在这些挑战,ChatGPT仍展现出评分效率方面的潜力,尤其是经过领域特定微调后。研究得出结论:ChatGPT可作为人工评分的补充,但需进一步开发才能在高利害评估中可靠应用。未来研究应提升模型准确率、解决伦理考量问题,并探索将ChatGPT与实证方法结合的混合模型。

英文摘要:

This study aimed to determine if ChatGPT's large language models could match the scoring accuracy of human and machine scores from the ASAP competition. The investigation focused on various prediction models, including linear regression, random forest, gradient boost, and boost. ChatGPT's performance was evaluated against human raters using quadratic weighted kappa (QWK) metrics. Results indicated that while ChatGPT's gradient boost model achieved QWKs close to human raters for some data sets, its overall performance was inconsistent and often lower than human scores. The study highlighted the need for further refinement, particularly in handling biases and ensuring scoring fairness. Despite these challenges, ChatGPT demonstrated potential for scoring efficiency, especially with domain-specific fine-tuning. The study concludes that ChatGPT can complement human scoring but requires additional development to be reliable for high-stakes assessments. Future research should improve model accuracy, address ethical considerations, and explore hybrid models combining ChatGPT with empirical methods.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑