Cognitive Architecture for Process industries
summary
A Cross-Sectorial Data Lab implements a cloud+edge pipeline (MinIO, Spark, Python/R, model repo, realtime edge deployment) and a semantic layer (ontology/knowledge graphs) that uses GPT‑4 to extract formal task definitions from narrative problem descriptions, then auto-generate Python/R scripts by chaining operators and rewriting ontology plans into executable code. Evaluated on five cross‑domain problems, automatic extraction shows higher precision than typical human consistency (~80–85%) and good recall when examples match domain; outputs require human validation and are not yet fully autonomous.
AI-written summary · about
chapters (11)
- 0:00Cross-sector data lab architecture
- 2:27Cloud platform for data consolidation
- 5:15Predictive functions development tools
- 8:34Knowledge sharing with semantics
- 12:07LLMs for formal problem extraction
- 15:23Cognitive architecture definition
- 16:27Semantic frameworks and knowledge graphs
- 17:55Handling LLM inaccuracies
- 20:05Structured data with domain knowledge
- 21:54Automating data mining tasks
- 22:57LLMs in domain fusion potential
related talks
22mIntent-Driven Automation for Machine Data: Natural Language to Edge
48m(Unpacking) AI's Impact on System Architecture
33mEnterprise Architecture Governance in the AI Era
09mAutomating cognitive tasks to increase business value
13mBuilding High-Performance AI Literature Processing
34mData Products, From 0 to data science pipeline
22mStructuring the Unstructured : Empowering the AI Workflows
27mGenerative AI and IoT, is there a match?