ZHOU Caishen1* and YEO Wee Kiang2
1, 2 Department of Information Systems and Analytics, School of Computing
Zhou, C., & Yeo, W. K. (2024). A Language Model-enhanced Network-centric Approach to Career Skills Enhancement [Lightning Talk]. In Higher Education Conference in Singapore (HECS) 2024, 3 December, National University of Singapore. https://blog.nus.edu.sg/hecs/hecs2024-zhou-yeo
SUB-THEME
Opportunities from Generative AI
KEYWORDS
Generative AI, Personalised Learning Pathways, Large Language Models, Graph-Enhanced Dynamic Training, Competency-Based Learning.
CATEGORY
Lightning Talk
INTRODUCTION
Adult continuing education is changing rapidly to meet the needs of today’s fast-evolving job market and the unique learning preferences of adults. Our project uses Large Language Models (LLMs) and skills knowledge graphs to create personalised learning paths. This approach makes career changes and skill development easier and more effective. (Knowles, 1970; Hase and Kenyon, 2000).
BACKGROUND AND MOTIVATION
Our system employs an LLM to facilitate natural language conversations, making interactions more intuitive and user-friendly. Within this platform, the system performs detailed skills assessments. It evaluates an individual’s current skills and career goals and identifies skill gaps. This is done by analysing a skills knowledge graph specific to the user’s desired job role (Shou et al., 2023). This network-centric approach addresses the limitations of traditional linear educational pathways, which often fail to accommodate the diverse and dynamic needs of adult learners (Romero and Ventura, 2007). This approach offers personalised and efficient learning tailored to specific career goals, unlike the traditional approach, which revisits redundant content and lacks focus.
Figure 1. Comparison of linear learning path and complex interconnected skills network for personalised learning
Figure 1 shows two learning pathways. The left diagram is a linear path from A to C. The right diagram displays a complex network of skills, highlighting multiple pathways and progression routes. This illustrates the difference between traditional linear educational pathways and network-centric approaches, emphasising the system’s adaptability in mapping personalised learning paths. This adaptability is evident in the journey of Anand, a mid-level software developer proficient in Python, aiming to specialise in Artificial Intelligence (AI) and Machine Learning, particularly Generative AI. His journey contrasts two learning approaches: a network-centric approach that offers personalised, efficient learning aligned with his goals, and a traditional approach that revisits redundant content and lacks focus.
METHODOLOGY
Our approach uses a system software, “Neo4j” and Retriever-Augmented Generation (RAG) techniques to improve the learning experience. Neo4j generates skill relationships based on data from the Jobs-Skills Dashboard – SDFE23/24 by SkillsFuture Singapore (SkillsFutureSG, 2023) and thereafter identifies skill gaps using Cypher queries. These queries are generated by LLMs based on user input. RAG enhances information retrieval and response generation, ensuring coherent and optimised learning pathways (Guo and Berkhahn, 2016; Lewis et al., 2020).
SIGNIFICANCE OF THE PROJECT
The network-centric approach significantly enhances the learning experience by allowing individuals to bypass introductory courses and focus directly on new and relevant areas pertinent to the user’s career goals. This personalised learning trajectory enables learners to build on their existing skills efficiently, specialising in areas critical to their desired career paths. Our approach not only improves learning efficiency but also boosts learner engagement by offering choices that resonate with their personal interests and career aspirations (Vaswani et al., 2017; Zawacki-Richter et al., 2019). By tailoring the learning process to the specific needs and goals of each learner, we ensure that the education provided is directly relevant and immediately applicable, thereby facilitating smoother and more effective career transitions.
CONCLUSION
In conclusion, our network-centric approach, supported by Large Language Models and skills network graphs, offers personalised and efficient learning pathways for adult learners. This method overcomes the limitations of traditional learning by recognising prior knowledge and focusing on practical, job-relevant skills. It is a valuable tool for career enhancement in a fast-changing job market. (Bates, 2019; Bai and Che, 2021).
REFERENCES
Bai, J., & Che, L. (2021). Construction and application of database micro-course knowledge graph based on Neo4j. Association for Computing Machinery, 1-5, 68. https://dl.acm.org/doi/10.1145/3448734.3450798
Bates, A. W. (2019). Teaching in a digital age: Guidelines for designing teaching and learning (2nd ed.). BCcampus. https://pressbooks.bccampus.ca/teachinginadigitalagev2/
Guo, C., & Berkhahn, F. (2016). Entity embeddings of categorical variables. Cornell University. 1-9. https://doi.org/10.48550/arXiv.1604.06737
Hase, S. & Kenyon, C. (2000). From andragogy to heutagogy. Southern Cross University. 5(3), 1-10.
Knowles, M. S. (1970). The modern practice of adult education from pedagogy to andragogy. Association Press.
Lewis, P., Perez, E., Piktus, A., Petroni. F., Karpukhin. V., Goyal. N., Küttler. H., Lewis. M., Yih. W. -T., Rocktäschel, T., Riedel. S., & Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. Cornell University. https://doi.org/10.48550/arXiv.2005.11401
Romero, C., & Ventura,S. (2007). Educational data mining: A survey from 1995 to 2005. Expert Systems with Applications. 33(1), 135-146. https://doi.org/10.1016/j.eswa.2006.04.005
Shou, Z., Chen, Y., Wen, H., Liu, J., Mo, J., & Zhang, H. (2023). A knowledge concept recommendation model based on tensor decomposition and transformer reordering. Electronics, 12(7), 1593. https://doi.org/10.3390/electronics12071593
SkillsFutureSG. (2023). Jobs-Skills Dashboard – SDFE23/24 by SkillsFutureSG. https://public.tableau.com/app/profile/skillsfuturesg/viz/JobsSkillsTalentInsight-SDFE_17001475553270/Overview
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L., & Polosukhin, I. (2017). Attention is all you need. https://doi.org/10.48550/arXiv.1706.03762
Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on artificial intelligence applications in higher education – where are the educators?”. International Journal of Educational Technology in Higher Education. 16, 1-27, 39. https://doi.org/10.1186/s41239-019-0171-0