How to Get Started with Knowledge Graphs with Andy Fitzgerald
Jun 21, 2022
auto_awesome
In this discussion, Andy Fitzgerald, an expert in information and content strategy, unpacks the intriguing world of knowledge graphs. He explains their role in enhancing user experiences through effective data integration. The conversation delves into the humorously depicted pitfalls of poor data connections and the critical need for a cohesive knowledge model. Andy also explores the significance of standards in knowledge representation, showing how starting small with knowledge graphs can lead to substantial business benefits.
Knowledge graphs allow organizations to integrate diverse data sources for improved user experiences through personalized and relevant recommendations.
Choosing between buying or building a knowledge graph solution highlights the importance of customization to reflect unique organizational knowledge and goals.
Deep dives
Understanding Knowledge Graphs
Knowledge graphs integrate data and knowledge from various sources to create machine-readable facts, enabling organizations to achieve specific business goals. These goals can include personalization, advanced analytics, and improved recommendation systems. A clear example demonstrates how organizations often misinterpret data, leading to irrelevant recommendations for users. By employing knowledge graphs, businesses can enhance the relevance of data connections and improve user experience.
Building vs. Buying Knowledge Graphs
Organizations often face the decision of whether to buy a knowledge graph solution or build one in-house, with many approaches requiring a combination of both. Off-the-shelf solutions may not adequately represent unique organizational knowledge, prompting the need for customization. A notable example is schema.org, which provides a generic vocabulary that lacks depth for differentiating an organization’s offerings. Thus, building a tailored knowledge model becomes crucial for gaining competitive advantage.
Creating a Knowledge Model
The development of a knowledge model is essential for a machine-readable representation of subject domains, enabling accurate data handling and recommendations. Practical examples, such as a pet store or veterinary clinic, illustrate how domain-specific models can predict customer needs based on their inquiries. Utilizing established knowledge models like SNOMED CT can significantly ease the process in well-defined sectors such as healthcare. The technical side involves standards like SCOS and OWL, which help organizations structure their data efficiently, ensuring it remains actionable.
Starting Small with Knowledge Graphs
When faced with extensive, unstructured data, organizations should focus on small increments to build an effective knowledge graph system. Unlike relational databases, knowledge graphs benefit from the open world assumption, making them adaptable to evolving data without accumulating technical debt. Initiating a project with a specific business need can provide immediate value, allowing organizations to explore scalable solutions over time. This incremental approach, supported by industry standards, facilitates smoother transitions between different systems, promoting long-term sustainability.
On this podcast, APQC’s Mercy Harper and Lauren Trees talk with information and content strategy consultant Andy Fitzgerald about what knowledge graphs are and how organizations can get started with this powerful approach for improving user experiences. For another approachable discussion of knowledge graphs, Andy recommends the Content Strategy Insights podcast episode “Aaron Bradley: Knowledge Graph Strategy for Content.”