ALL-IN-ONE: Multi-Task Learning BERT models for Evaluating Peer Assessments

Jia, Qinjin; Cui, Jialin; Xiao, Yunkai; Liu, Chengyuan; Rashid, Parvez; Gehringer, Edward F.

doi:10.48550/arxiv.2110.03895

Cited by 7 publications

(4 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, LLMs have been used to generate children's narratives [4,35], some of which have even been sold publicly [64]. In a different research trajectory, several scholars have used LLMs to create intelligent learning partners capable of collaborating with humans [47], providing feedback [49] and encouraging students [24,88]. One common application involves employing LLMs as a conversational partner in written or oral form, such as in the context of task-oriented dialogues that offer language practice opportunities [24].…”

Section: Using Large Language Models For Child-facing Conversational ...mentioning

confidence: 99%

Mathemyths: Leveraging Large Language Models to Teach Mathematical Language through Child-AI Co-Creative Storytelling

Zhang,

Liu,

Ziska

et al. 2024

Proceedings of the CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Figure 1: Interaction between a child and Mathemyths: demonstrating the system's ability to teach mathematical language through child-AI co-creative storytelling. Mathemyths provides open-ended questions to solicit how the child wishes the story should progress, on-the-fly feedback to acknowledge the child's responses, and co-creative story continuation with in-context explanations of math words. When the child needs additional support to continue the story, Mathemyths offers scaffolding through follow-up questions and "hint & rephrase" strategies.

show abstract

Section: Using Large Language Models For Child-facing Conversational ...mentioning

confidence: 99%

Mathemyths: Leveraging Large Language Models to Teach Mathematical Language through Child-AI Co-Creative Storytelling

Zhang,

Liu,

Ziska

et al. 2024

Proceedings of the CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

show abstract

“…When doing the task, prompt students to use the GAI to get more explanations and examples of concepts they do not understand. In addition, a feedback mechanism could be set during the interaction with GAI [see Herft (2023) and Jia et al (2021) for more details to support and improve teaching assessment and feedback practices]. Students can know immediately whether their answers are correct.…”

Section: Practical Implicationsmentioning

confidence: 99%

The relationship between student interaction with generative artificial intelligence and learning achievement: serial mediating roles of self-efficacy and cognitive engagement

Liang,

Wang,

Luo

et al. 2023

Front. Psychol.

View full text Add to dashboard Cite

Generative artificial intelligence (GAI) shocked the world with its unprecedented ability and raised significant tensions in the education field. Educators inevitably transition to an educational future that embraces GAI rather than shuns it. Understanding the mechanism between students interacting with GAI tools and their achievement is important for educators and schools, but relevant empirical evidence is relatively lacking. Due to the characteristics of personalization and real-time interactivity of GAI tools, we propose that the students–GAI interaction would affect their learning achievement through serial mediators of self-efficacy and cognitive engagement. Based on questionnaire surveys that include 389 participants as the objective, this study finds that: (1) in total, there is a significantly positive relationship between student–GAI interaction and learning achievement. (2) This positive relationship is mediated by self-efficacy, with a significant mediation effect value of 0.015. (3) Cognitive engagement also acts as a mediator in the mechanism between the student–GAI interaction and learning achievement, evidenced by a significant and relatively strong mediating effect value of 0.046. (4) Self-efficacy and cognitive engagement in series mediate this positive association, with a serial mediating effect value of 0.011, which is relatively small in comparison but also shows significance. In addition, the propensity score matching (PSM) method is applied to alleviate self-selection bias, reinforcing the validity of the results. The findings offer empirical evidence for the incorporation of GAI in teaching and learning.

show abstract

“…LLMs have also been used to foster curiosity and question-asking skills among undergraduate students explaining coding concepts in computing education (MacNeil et al, 2022). Additionally, peer learning has been augmented through LLMs, as evidenced by Jia et al (2021), who employ Bidirectional Encoder Representations from Transformers (BERT) to evaluate peer assessments, promoting a collaborative learning atmosphere. Pereira et al Int J Educ Technol High Educ (2024) 21:25 This body of work collectively illustrates the expanding role of LLMs in enhancing educational practices through automation, personalized learning experiences, and support for both instructors and students.…”

mentioning

confidence: 99%

Constructive alignment in a graduate-level project management course: an innovative framework using large language models

Pereira,

Nsair,

Pereira

et al. 2024

Int J Educ Technol High Educ

View full text Add to dashboard Cite

Constructive alignment is a learning design approach that emphasizes the direct alignment of the intended learning outcomes, instructional strategies, learning activities, and assessment methods to ensure students are engaged in a meaningful learning experience. This pedagogical approach provides clarity and coherence, aiding students in understanding the connection of their learning activities and assessments with the overall course objectives. This paper explores the use of constructive alignment principles in designing a graduate-level Introduction to Project Management course by leveraging Large Language Models (LLMs), specifically ChatGPT. We introduce an innovative framework that embodies an iterative process to define the course learning outcomes, learning activities and assessments, and lecture content. We show that the implemented framework in ChatGPT was adept at autonomously establishing the course's learning outcomes, delineating assessments with their respective weights, mapping learning outcomes to each assessment method, and formulating a plan for learning activities and the course's schedule. While the framework can significantly reduce the time instructors spend on initial course planning, the results demonstrate that ChatGPT often lacks the specificity and contextual awareness necessary for effective implementation in diverse classroom settings. Therefore, the role of the instructor remains crucial in customizing and finalizing the course structure. The implications of this research are vast, providing insights for educators and curriculum designers looking to infuse LLMs systems into course development without compromising effective pedagogical practices.

show abstract

ALL-IN-ONE: Multi-Task Learning BERT models for Evaluating Peer Assessments

Cited by 7 publications

References 27 publications

Mathemyths: Leveraging Large Language Models to Teach Mathematical Language through Child-AI Co-Creative Storytelling

Mathemyths: Leveraging Large Language Models to Teach Mathematical Language through Child-AI Co-Creative Storytelling

The relationship between student interaction with generative artificial intelligence and learning achievement: serial mediating roles of self-efficacy and cognitive engagement

Constructive alignment in a graduate-level project management course: an innovative framework using large language models

Contact Info

Product

Resources

About