Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to build knowledge graphs from unstructured data through this 25-minute conference talk that introduces KG-ETL pipelines as a solution to traditional ETL limitations in the LLM era. Discover three competing architectures for constructing knowledge graphs from raw data: LLM-based, traditional NLP-based, and hybrid vector search-based approaches. Explore how to use LLM prompts as a new transformation layer in your pipeline and understand how hybrid retrieval workflows leverage vector stores for entity resolution beyond semantic search. Compare cost, latency, scalability, and observability trade-offs across different approaches while examining a novel vector-store technique for high-precision entity resolution using FastText embeddings. Master a decision framework to select the optimal KG-ETL pipeline based on your specific data type and business requirements, moving beyond traditional structured table processing to capture the context and implicit relationships that modern AI systems require.
