Building Multiagent Systems: An Ontology-Driven Approach with SPADE
summary
An ontology-to-code pipeline translates a domain ontology into runnable agent templates for SPADE, generating agent classes, roles and behaviors (cyclic, periodic, oneshot, timer, finite-state-machine behavior) and wiring them to an XMPP chat server. Examples include a working coffee-machine, traffic-light FSM, weather update and a recipe-world with three factory providers and three order consumers. The repo supplies Docker Compose for an XMPP server, Conda env, self-signed TLS guidance and demo scripts that run locally and succeed.
AI-written summary · about
chapters (9)
- 0:00Multi-agent systems & SPADE
- 9:47Defining agents
- 20:13Multi-agent systems & communication
- 30:55SPADE agent implementation
- 45:55Agent behaviors in SPADE
- 55:15Ontologies & their role
- 1:08:30Ontology-driven framework
- 1:16:51Ontology to SPADE agents
- 1:23:23Future directions & summary
related talks
28mThe Put That Sh*t On Agent: Building Multi-Agent AI Systems That Actually Talk to Each Other*
56mDesign Patterns for Multi Agent Systems
41mPatterns & Practices for building Multi-Agent Systems
01h 40mOrchestrating a Role-Playing Game Universe with Multi-Container AI Agents
31mReal-World Multi-Agent Development | Ship Faster with AI Agents
32mEvent-Driven & Multi-Agent Architectures for Gen AI Workflows
30mLive Demo – Develop and Deploy a Trustworthy Multi-Agent System
54mMicro-Mind Services: Building Complex Behavioral Systems