Adaptive Empathy Learning Support in Peer Review Scenarios

Title of the Paper

Adaptive Empathy Learning Support in Peer Review Scenarios

Bibliographic Information

• Subject Areas: Educational Technology, Natural Language Processing (NLP), Learning Support Systems
• Keywords: Educational Applications, Writing Support Systems, Automated Feedback, Empathy Learning, Self-Regulated Learning, Machine Learning, Human-Computer Interaction

Research Background and Issues

• Identified Problems or Challenges:

  1. College students' empathy skills have declined by more than 30% from 1979 to 2009.
  2. Traditional empathy training in education (e.g., role-playing) is often limited to a small number of students and cannot meet the needs of large-scale classrooms or online learning environments.
  3. Sustained personalized feedback is critical for students in developing empathy skills, but it is difficult to achieve due to limited educational resources.
  4. The potential of text-based empathy detection and intelligent feedback for cultivating students' empathy skills has not been fully explored.

• Significance of the Research: Empathy is an essential core skill for social interaction, professional communication, and future work environments. Cultivating empathy skills is crucial in modern educational frameworks, as it helps students better adapt to complex team environments, enhances learning outcomes, and advances the possibilities of human-AI collaboration.

• Motivation and Related Work: Through an analysis of existing literature, the authors found that while technologies such as virtual reality and intelligent tutoring systems can assist in empathy training, there is a lack of an empirically supported design framework for developing adaptive learning tools that provide personalized intelligent feedback. This research aims to explore a solution leveraging natural language processing and machine learning technologies.

Solution

• Proposed Method and Solution: Developed an adaptive learning support system called ELEA, which uses NLP technology to assess students' cognitive and emotional empathy levels in their written texts and provides personalized feedback based on these assessments.

• Innovative Features:

  1. ELEA evaluates the empathy structure of texts based on cognitive and emotional dimensions, grounded in theoretical frameworks like the "Toronto Empathy Scale."
  2. Utilizes the advanced BERT model to classify students' text paragraphs.
  3. Compared to simple dictionary-based recommendations (e.g., reference tools), ELEA provides more granular adaptive feedback.
  4. Designed an iterative testing mechanism to ensure user needs are met.

• Implementation Steps:

  1. Needs Analysis and UI Design: Combined theory-driven and user-centered design approaches, conducted literature reviews, and interviewed students to extract design requirements.
  2. Algorithm Design: Trained a BERT model using 500 annotated student texts labeled with cognitive and emotional empathy levels to predict whether paragraphs were non-empathetic, neutral, or empathetic.
  3. Tool Development: Built a responsive web application, designed interactive interfaces, and provided instant feedback and theoretical explanations.
  4. Experimental Evaluation: Randomly assigned 119 students into groups using either ELEA or a reference tool for peer review tasks in a business model course, measuring and comparing the effectiveness of empathy skill development.

Research Findings

• Specific Results:

  1. Texts written by students using ELEA demonstrated significantly higher levels of emotional empathy (average score increased from 2.20 with the reference tool to 2.75).
  2. Students reported that ELEA significantly improved their empathy skills (average rating of 5.03, significantly higher than the reference tool's 3.93).
  3. Technology acceptance evaluations (e.g., ease of use, intention to use, enjoyment) showed significant advantages.

• Comparison with Existing Solutions: ELEA's adaptive feedback approach significantly outperforms traditional dictionary-based feedback tools. In particular, it demonstrated a remarkable improvement in emotional empathy skills, with superior technology acceptance and user experience ratings compared to the reference tool.

• Experimental or Evaluation Results:

  • Emotional Empathy Assessment: Students in the ELEA group showed significant improvement in emotional empathy levels (p=0.0027, effect size Cohen’s d = 0.5699).
  • Cognitive Empathy Assessment: No significant difference was found between the two groups (p=0.9299).
  • Technology Acceptance Metrics: Students using ELEA reported higher perceived ease of use (average 5.72), intention to use (average 5.14), and enjoyment (average 5.31) compared to the reference tool group.

• Limitations and Future Directions:

  1. Limitations:

     • ELEA's applicability is currently limited to German-speaking students in peer review scenarios; adaptations are needed for different languages, cultures, or educational contexts.
     • The study only validated short-term effects and did not evaluate long-term learning benefits.
     • The evaluation method relies primarily on expert annotations, highlighting the need for further development of more granular skill assessment frameworks.

  2. Future Directions:

     • Develop system versions adapted to other languages and cultures.
     • Explore the impact on different user groups (e.g., individuals on the autism spectrum).
     • Integrate ELEA with conversational agent technologies to enhance social presence and improve learning outcomes.
     • Conduct long-term experiments to verify the sustained impact on students' empathy skills.