LLM Driven Approach for Capability Modelling with Context-enriched Prompt Engineering

Authors

Iromie Samarasekara, University of New South WalesFollow
Madhushi Bandara, University of Technology SydneyFollow
Fethi Rabhi, University of New South WalesFollow
Boualem Benatallah, Insight SFI Research Centre on Data Analytics, Dublin City University, IrelandFollow
Rouzbeh Meymandpour, Capsifi Australia Pty LtdFollow

Abstract

Business/IT alignment in information systems is widely discussed among researchers and industry practitioners alike. IT departments often struggle to deliver business value, often due to the lack of a coherent representation of the business operating model that can be referred to in building business applications. Whilst business capability maps offer a promising solution for the above concern, it is a tedious exercise to develop them manually. We explore the possibility of leveraging Large Language Models (LLMs) in generating industry specific capability maps, using innovative prompting that effectively incorporates concept definitions into LLM prompts. Our preliminary evaluation results indicate the capacity of the proposed LLM driven approach and the prompt engineering process in producing an effective prompting technique that improves LLM output quality. The proposed method for automated capability modelling is an original contribution to the business architecture discipline and contributes to the steadily growing literature on LLM prompt engineering.

Recommended Citation

Samarasekara, Iromie; Bandara, Madhushi; Rabhi, Fethi; Benatallah, Boualem; and Meymandpour, Rouzbeh, "LLM Driven Approach for Capability Modelling with Context-enriched Prompt Engineering" (2024). ACIS 2024 Proceedings. 38.
https://aisel.aisnet.org/acis2024/38