Temporal Knowledge Graphs : Uncovering Hidden Patterns | GPH3 w/ Paco Nathan, Tom Smoker & Chia Yang

Extracting Insights From Unstructured Text Temporal knowledge graphs provide new ways to extract insights from unstructured text data. By discovering relationships and dynamically identifying patterns, these graphs make it possible to track trends, identify key actors, and map shifts in topics or events. Extracting entities, modeling how relationships change over time, and structuring the evolving data all require overcoming substantial technical hurdles. While temporal graphs offer significant theoretical promise, practical implementation reveals unexpected complexities—from managing incomplete time data to integrating models which keep pace with evolving information. In this podcast, we’re joined by both Chia Jeng Yang and Tom Smoker from WhyHow.AI, as they share real-world lessons from constructing temporal knowledge graphs using congressional hearing transcripts. We’ll look at the practical techniques they used to transform messy, text-based data into meaningful, actionable knowledge, and discuss strategies for handling evolving datasets in research and analytics. Resources referenced in this session: github.com/whyhow-ai/knowledge-table github.com/whyhow-ai/knowledge-graph-studio github.com/whyhow-ai/knowledge-graph-studiogithub.com/whyhow-ai/knowledge-table synthetichealth.github.io/synthea/ hl7.org/fhir/patient.schema.json.html Case Studies referenced in this session: medium.com/enterprise-rag/case-study-turning-congressional-hearing-transcripts-into-temporal-knowledge-graphs-0d78075181c7 medium.com/enterprise-rag/case-study-turning-doctor-transcripts-into-temporal-medical-record-knowledge-graphs-cf624d4927eb medium.com/enterprise-rag