Predicting Sustainable Development Goals for University Courses Using Large Language Models and Conventional Foundation Models

Konsep Inti

Predicting the most relevant United Nations Sustainable Development Goals (SDGs) for university courses using large language models (LLMs) and fine-tuned smaller foundation models.

Abstrak

The researchers present a novel approach to predicting the United Nations Sustainable Development Goals (SDGs) relevant to university courses. They use the large language model PaLM 2 to generate training data by providing course descriptions as input and obtaining the top SDG goals predicted by the LLM. This generated data is then used to fine-tune several smaller foundation models, including BERT, mBERT, RoBERTa, XLM-RoBERTa, and BART, for the multi-label SDG prediction task.
The dataset consists of 2,125 English course descriptions from Metropolia University of Applied Sciences, spanning the years 2021-2023. After preprocessing and cleaning the data, the researchers use the LLM to generate the SDG labels for each course, which are then encoded into a binary format for the training of the smaller models.
The performance of the models is evaluated using precision, recall, and F1-score. The results show that the BART model achieves the highest F1-score of 0.786, outperforming the other fine-tuned models. The researchers also analyze the performance of the models across individual SDGs, identifying areas where certain models excel and highlighting the need for addressing data imbalances to improve generalization.
This research contributes to the integration of SDGs in higher education by providing a practical methodology for universities to assess the alignment of their course offerings with the UN's sustainable development goals. The findings open up avenues for further research and implementation of similar approaches to foster sustainable practices in academic institutions worldwide.

Statistik

The dataset consists of 2,125 English course descriptions from Metropolia University of Applied Sciences, spanning the years 2021-2023.
The dataset was split 70:15:15 into training, validation, and testing subsets.

Kutipan

"The best performing model in our experiments was BART with an F1-score of 0.786."
"The nuanced performance variations across the models underscore the significance of model selection tailored to specific NLP tasks' requirements."

Wawasan Utama Disaring Dari

Predicting Sustainable Development Goals Using Course Descriptions -- from LLMs to Conventional Foundation Models

by Lev ... pada arxiv.org 04-24-2024

https://arxiv.org/pdf/2402.16420.pdf

Pertanyaan yang Lebih Dalam

How can the proposed methodology be extended to incorporate additional data sources, such as student feedback or industry-academia collaborations, to further enhance the prediction of SDGs for university courses?

To enhance the prediction of Sustainable Development Goals (SDGs) for university courses, the proposed methodology can be extended by incorporating additional data sources such as student feedback and industry-academia collaborations.

Student Feedback Integration:

Sentiment Analysis: By incorporating sentiment analysis of student feedback related to specific courses, the model can better understand the impact and relevance of SDGs in those courses. Positive sentiments towards certain SDGs can indicate successful integration and understanding, while negative sentiments can highlight areas for improvement.
Topic Modeling: Utilizing topic modeling techniques on student feedback can help identify recurring themes related to SDGs, providing valuable insights into the effectiveness of SDG implementation in different courses.

Industry-Academia Collaborations:

Industry Reports and Case Studies: Integrating data from industry reports and case studies can offer real-world perspectives on the practical application of SDGs. This data can enrich the model's understanding of how SDGs are implemented in professional settings, influencing the relevance of certain goals in academic courses.
Collaborative Projects: Engaging in collaborative projects with industry partners can provide firsthand data on the SDGs that are most pertinent in current industry practices. This data can be used to validate and update the model's predictions based on real-time industry trends.

By incorporating these additional data sources, the model can gain a more comprehensive understanding of the alignment between university courses and SDGs, leading to more accurate predictions and informed decision-making in curriculum development.

How can the potential challenges and ethical considerations in deploying such SDG prediction models at scale across multiple educational institutions, particularly in terms of data privacy and algorithmic bias, be addressed?

Deploying SDG prediction models at scale across multiple educational institutions comes with various challenges and ethical considerations that need to be addressed to ensure responsible and effective implementation.

Data Privacy:

Anonymization: Implement strict anonymization protocols to protect sensitive student and course data. Ensure that personally identifiable information is removed or encrypted before processing.
Data Encryption: Utilize encryption techniques to secure data both in transit and at rest, preventing unauthorized access.
Compliance: Adhere to data privacy regulations such as GDPR to safeguard student information and maintain transparency in data handling practices.

Algorithmic Bias:

Bias Detection: Regularly audit the model for biases by analyzing its predictions across different demographic groups. Address any biases identified through retraining or adjusting the model's parameters.
Diverse Training Data: Ensure the model is trained on diverse and representative datasets to mitigate biases that may arise from skewed or limited data.
Ethical Review: Conduct ethical reviews of the model's predictions and decisions to assess potential biases and ensure fairness in outcomes.

Ethical Considerations:

Informed Consent: Obtain informed consent from students and faculty before using their data for model training or prediction purposes.
Transparency: Maintain transparency in the model's decision-making process and clearly communicate how SDG predictions are generated to build trust with stakeholders.
Accountability: Establish clear accountability mechanisms to address any ethical concerns or issues that may arise during model deployment.

By proactively addressing these challenges and ethical considerations, educational institutions can deploy SDG prediction models responsibly and ethically, fostering trust and promoting the effective integration of SDGs in academic curricula.

How can the insights gained from this research be leveraged to design interdisciplinary curricula and learning experiences that holistically address the interconnected nature of the UN Sustainable Development Goals?

The insights gained from the research on predicting Sustainable Development Goals (SDGs) for university courses can be leveraged to design interdisciplinary curricula and learning experiences that holistically address the interconnected nature of the UN SDGs in the following ways:

Interdisciplinary Course Design:

Cross-Curricular Integration: Identify common themes and goals across different disciplines and design courses that integrate multiple SDGs to promote interdisciplinary learning.
Collaborative Projects: Encourage collaborative projects that bring together students from various disciplines to work on real-world challenges related to SDGs, fostering a holistic understanding of sustainability issues.

Experiential Learning:

Field Trips and Internships: Incorporate field trips and internships that expose students to sustainable practices in different industries, allowing them to apply SDGs in real-world contexts.
Service-Learning Programs: Engage students in service-learning programs that involve community projects aligned with specific SDGs, providing hands-on experience in addressing societal challenges.

Technology Integration:

Data Analytics and Visualization: Integrate data analytics and visualization tools into courses to analyze and communicate progress towards SDGs, promoting data-driven decision-making.
Simulation and Modeling: Use simulation and modeling tools to help students understand the complex interconnections between different SDGs and explore potential solutions to global sustainability challenges.

Global Perspective:

International Collaborations: Facilitate international collaborations with universities and organizations from different countries to provide students with a global perspective on SDGs and promote cultural exchange.
Global Challenges Courses: Offer courses that focus on global challenges such as climate change, poverty, and inequality, emphasizing the interconnected nature of SDGs and the need for collective action.

By incorporating these strategies, educational institutions can design interdisciplinary curricula and learning experiences that not only address individual SDGs but also foster a comprehensive understanding of the interconnectedness and interdependence of the UN Sustainable Development Goals.