Seth Earley demonstrates how knowledge graphs can transform the business at KMWorld Connect 2020
Though knowledge graphs are not a novel concept anymore, the technique is still an underrated technology that can help uncover the right data during knowledge management work. It can be confusing to some but it ultimately liberating for the enterprise.
Seth Earley CEO, Earley Information Science, discussed “Knowledge Graphs: AI Buddies?” at KMWorld Connect 2020.
To be competitive in today's data-driven world, it is more important than ever to make use of all your company's data assets with the latest technologies, according to Earley.
Decisions made in a data vacuum result in lost time and/or money. What if there was a way to ensure that everyone had access to all of the data; structured or unstructured, from any source, any silo, and it was both accurate and meaningful?
Knowledge graphs provide that connective framework for all of your data, with ontology defining meaningful relationships and context.
He explained that if businesses want executive funding and support for knowledge graphs, he recommended not to show confusing diagrams or use buzzwords to explain the concept. Instead, demonstrate capabilities and show measurable business outcomes, he said.
A knowledge graph is a representation of unstructured content categorized across multiple metadata elements. They allow for contextual navigation across an unstructured repository.
The enterprise architect faces a dilemma. There are too many tables and attributes, data experts are unavailable, impossible to understand naming, data quality is unknown and more.
“When you try to retrieve this information you have all sorts of problems from the structured and unstructured side,” Earley said. “It becomes difficult to navigate disparate systems.”
The knowledge graph is a mechanism for integration and making sense of all these data sources, he explained.
Graph data focuses on relationships between elements while knowledge graphs are representations of unstructured information categorized and classified across multiple metadata elements.
An example of graph data is movie and TV show classifications on IMDB. To map out all the connections as a knowledge graph, it can show deeper information into the relationships behind the movie and TV shows.
Knowledge graphs and graph data power AI and machine learning systems by providing reference data and knowledge about conceptual relationships between products, solutions, problems, tasks, and processes, he said.
Knowledge graphs and ontology are at the core of a unified integration framework.
Knowledge graphs express and apply enterprise data using the ontology framework. It provides consistent information architecture.
“This is aspirational,” Earley said.
There are some caveats, however, they do not fix data and human judgment is still required. Data and information governance, data quality, and metrics driven decision making frameworks are required using both conventional and emerging AI powered technologies.