Reasoning Engine (AJAN)

• Domain-free multi-agent system web service for behavior modeling and execution of agents, based on SPARQL-BTs
• AJAN is fully integrated into the MOSIM infrastructure and is used for dynamic execution of available MMUs based on redefined tasks of the High-Level Task Editor
• Has a web editor for intuitive modeling of so-called MOSIM-breakdowns
• It can be easily extended (with the AJAN-Plug-In System) with additional
• AI methods like, Reinforcement Learning, Action Planning a.o.

Input(s):

• Behavior Models (SPARQL-BTs)
• High-Level Tasks
• MOSIM Framework endpoints, like Scene-Access

Main feature(s):

• Execution of MMU instructions -- based on High-Level Tasks -- based on internal behavior models
• Domain-free, can be used separately from MOSIM

Output(s):

• MMU Instructions
• MOSIM Service calls to get dynamic paths, to get avatar transformation a.o.

Provider:

• AJAN-service: https://github.com/aantakli/AJAN-service
• AJAN-editor: https://github.com/aantakli/AJAN-editor
• AJAN-MOSIM-Unity: https://github.com/aantakli/AJAN-MOSIM-unity

• MOSIM context: MOSIM Framework, MMUs, High-Level Task Editor
• General: HTTP + RDF

• MOSIM users in general who want to run avatar tasks or model avatar behavior in detail or test MMU behavior
• Expert users who are skilled in modeling agents and thus in implementing autonomous system behavior
• Novices who have no prior knowledge of RDF and SPARQL but still want to model avatar behavior based on expert template

• AJAN-service: LGPL-2.1 License
• AJAN-editor: MIT License
• AJAN-MOSIM-Unity: MIT License