.png){kind=icon}
.png){kind=icon}
Return
Anew approach for competency frameworks mapping using large language models
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2024.125648.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Competency frameworks are essential for organizations to align their workforce with strategic goals and for individuals to assess and develop their skills. However, the absence of a universal or unified competency framework presents a challenge, as each framework is subject to distinct guidelines and standards. Moreover, within a single framework, continuous updates to align with evolving standards can lead to equivalent competencies being expressed indifferent ways. Despite the fundamental similarity of the competencies, this divergence across frameworks can impede interoperability and complicate the aggregation of data from multiple frameworks. This paper addresses this issue by proposing an approach that leverages large language models (LLMs) for mapping competency frameworks to enhance interoperability among frameworks. We investigated various pre-trained LLMs to encode competency names from each framework. Subsequently, we employed cosine similarity to measure semantic similarity scores, which facilitated the identification of equivalent or closely related competencies across different frameworks. We evaluated our approach using three competency frameworks for project management, each aligned with different editions of the Project Management Body of Knowledge (PMBOK) standards. The experimental results demonstrate the effectiveness of the proposed approach in ameliorating frameworks interoperability.
Keywords:
Competency framework
Framework competency mapping
Large language models
Natural language processing
Project management
PMBOK
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W