Empower Your Data Governance: Harness the Power of Large Language Models

Bojan Ciric
The Future of Data
Published in
2 min readMay 13, 2023

Large Language Models (LLMs) are currently a trending topic, offering innovative perspectives across various applications. This article will explore practical use cases on how to leverage the potential of LLMs to enhance and streamline your data governance processes.

Introduction and context

Data governance revolves around identifying vital data and ensuring its suitability to drive positive business outcomes, uphold regulatory compliance, and optimize operational efficiency. The data steward plays a pivotal role in this process, tasked with managing, maintaining, and guaranteeing the quality and reliability of an organization’s data assets.

Once critical data elements are identified, the initial steps involve establishing definitions and data quality requirements. Subsequent stages include implementing these quality rules across all relevant systems, ensuring that the data associated with each element is fit-for-purpose.

To illustrate the potential of large language models, such as GPT-4, in assisting data stewards with these tasks, let’s engage in a quick exercise. We’ll focus on the data element ‘Social Security Number’, exploring how GPT-4 can help define quality requirements for this critical data element and generate code for its quality rule implementation.

Data Element Definition

Prompt: “Please provide the definition of the social security number”

Completion:

Social Security Number definition (generated by ChatGPT4)

Data Quality Rule Requirements

Prompt: Please provide data quality requirements for SSN

Completion:

Data quality rule requirements for SSN (generated by ChatGPT4)

Data Quality Rule Implementation

Prompt: Please provide JavaScript code for data quality rule implementation for SSN

Completion:

Data Quality rule JavaScript code (generated by ChatGPT4)

Conclusion

Large Language Models (LLMs) have the potential to enhance data governance programs and aid data stewards in increasing process efficiency. The primary focus of data governance is metadata, which carries fewer security concerns compared to corporate data, although security remains a critical consideration before implementation. Moreover, while the content generated by LLMs may appear sophisticated, it should not be used ‘as is’. Instead, it requires careful validation and tailoring to ensure it aligns with the specific needs of the organization.

Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the opinions or positions of any entities author represents.

--

--

Bojan Ciric
The Future of Data

Technology Fellow at Deloitte | Data Thinker | Generative AI Hands-on | Converts data into actionable insignts