Toward a University Knowledge Warehouse

What bridges a university's siloed systems to turn fragmented data into actionable intelligence and value without buying yet another IT system? A Knowledge Warehouse!

Higher education data is routinely trapped in departmental silos, dotted across a myriad of systems right across the enterprise. Achieving business value from these siloed systems and data is a perennial issue for all institutions.

For decades, organisations have sought to overcome these challenges, investing in vendor-driven platforms and solutions, consultation, and digital transformation programmes. Yet, for all this investment, challenges persist for organisations with the wonderful "spaghetti junction" of legacy, custom, and cloud infrastructure. We could just jump onto the AI bandwagon to help assist!? However, blindly deploying AI risks masking the structural flaws in the enterprise system and associated datasets. Instead, at Dublin City University (DCU), we are on a journey towards developing what we call a Knowledge Warehouse, an approach involving the development of ontologies for the domains concerning third-level institutions.

As we build towards the Knowledge Warehouse, this ontology-based approach serves as an IT agnostic semantic map, allowing the institution to suffuse raw data across our various systems databases with important institutional context. This tried-and-tested method - common in other industries (e.g. financial services) but novel in education - requires human capital over heavy IT investment, leveraging existing heterogeneous systems. Systems may change, but the ontology persists for master data management.

The Pilot Domain - External Engagement:

Our strategic pilot focuses on External Engagement - a sector spanning industry partnerships, alumni, government, and community outreach historically fragmented by native silos. This targeted pilot achieves three strategic goals:

1. Foundation for Scalability: Establishing methodologies, governance, and architectures to systematically map remaining university domains.

2. Industry Standardisation: Designed as an asset to be shared, adopted, and further developed by the sector.

3. Implementation Blueprint: Delivers an actionable playbook guiding peer institutions through building and deploying an enterprise ontology.

Strategic Asset - Institutional Knowledge and Sector Leadership:

The primary benefit of this architecture is a permanent, system-agnostic knowledge asset. The ontology serves as our IT agnostic framework for scalable Master Data Management - ensuring that while platforms change, our knowledge remains persistent and under our control. As an open-standard asset, it positions the university to guide the wider education sector.

Foregrounding and Capitalising on AI Advances:

Beyond the warehouse, this approach places us in a strong position to more fully capture the potential of Large Language Models (LLMs). Generative AI represents one of the most transformative technical advancements since modern computing began. Towards overcoming natural limitations of GenAI, most notably their propensity for hallucinations, a semantically rich ontology provides the mathematically sound context they require. This pivots AI away from speculative computation or guesswork with our institutional data to a reliable partner where we can confidently use Natural Language Querying (NLQ). As result, our stakeholders can confidently interrogate the knowledge warehouse using conversational language in GenAI, receiving precise, real-time insights to drive strategic decisions.