From the course: Introduction to Probabilistic Knowledge Graphs: AI-Driven Inference and Real-World Applications

Unlock this course with a free trial

Join today to access over 25,500 courses taught by industry experts.

Limitations of unstructured data

Limitations of unstructured data

From the course: Introduction to Probabilistic Knowledge Graphs: AI-Driven Inference and Real-World Applications

Start my 1-month free trial Buy for my team

Limitations of unstructured data

“

In earlier videos, we showed how unstructured data like text, audio, images, with a focus on text, like how they can be processed and mapped into a graph schema, then annotated with probabilities in a probabilistic knowledge graph. Today, we'll turn to the important question. What are the limits of doing that? And why does having a solid ontology or schema matter when working with unstructured sources? First, even though we rely on a fixed schema or ontology for mapping extracted facts, unstructured sources often do not align neatly with that schema. The text or media might express relationships in ways the schema did not anticipate or omit information that your ontology expects. That means the extraction pipeline may fail to map facts or mismap them because the schema lacks the matching predicate or class. This mismatch introduces noise, ambiguity, and gaps. Second, ontologies themselves are static relative to the fluid world of unstructured content. When you extract from text using…

Contents