580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How do Agent Communication Languages (ACLs) balance formal semantics and agent autonomy to enable effective multi-agent interaction?
This research area focuses on the development of well-defined semantics for ACLs that reconcile the need for clear meaning and interoperability with the inherent autonomy and heterogeneity of agents. The challenge is to establish communication protocols that allow agents to coordinate, negotiate, and collaborate meaningfully in open, dynamic environments without assuming internal mental states are fully accessible or shared. This theme is crucial for enabling scalable, interoperable multi-agent systems in practice.
Key finding: Introduces an observation-based empirical semantics that models agent communication meanings as probabilistic expectations of future actions (Social Interaction Structures) learned from observed behaviors rather than... Read more
Key finding: Critiques prevailing ACL approaches centered on mentalistic semantics (beliefs and intentions), arguing they unrealistically assume agents can read each other's minds. Advocates for a semantics emphasizing social agency,... Read more
Key finding: Proposes the use of UML to formally model the abstract syntax of ACLs and their content languages, facilitating automatic mapping to executable agent platform components. This supports diverse ACL experimentation and... Read more
Key finding: Analyzes ACL components emphasizing the balance between the necessity of explicit, precise communication mechanisms and the complexity arising from agent autonomy. Discusses how ACLs provide the intentional and social layers... Read more
Key finding: Provides a comprehensive survey of various ACLs and their semantic models, noting the lack of universally adopted standards and the challenges related to communication interoperability in heterogeneous MAS. Highlights the... Read more
2. What domain-specific and practical communication languages and protocols support efficient agent interaction in specialized multi-agent systems?
This research theme investigates the design and specification of tailored ACLs and communication languages that address domain-specific requirements and system constraints, such as embedded intelligent buildings, ecological simulations, semantic web integration, or multi-agent protocol enactment. The goal is to provide languages that are computationally efficient, flexible, and capable of expressing necessary multi-agent interactions, while often incorporating hierarchical or declarative approaches.
Key finding: Identifies limitations of general ACLs like KQML and FIPA for resource-constrained embedded intelligent building agents, and proposes DIBAL, a lightweight, hierarchical communication language optimized for embedded-agent... Read more
Key finding: Develops ECOLANG as a communication language enabling interaction between ecological simulation applications (EcoDynamo) and external agents representing users or stakeholders. Focuses on supporting complex ecological... Read more
Key finding: Presents SEA_L, a domain-specific language supporting multi-agent system development in Semantic Web contexts, integrating semantic web technologies with MAS abstractions. Proposes metamodel-based textual syntax and... Read more
Key finding: Extends the Blindingly Simple Protocol Language (BSPL) with novel constructs like dynamic role bindings, multicast communication, and subroles to model operational constraints and complex interaction patterns in decentralized... Read more
3. What are the methodological and practical challenges in developing and deploying conversational agents, and how do communication languages influence their design and implementation?
This theme covers the intersection of natural language processing, conversational agent (CA) development tools, and underlying communication protocols, focusing on practitioner challenges, development processes, and technology selections in building effective conversational systems. It highlights gaps between academic research and industry practices, the role of ACLs or dialog management languages, and emerging trends affecting CA lifecycle and capabilities.
Key finding: Analyzes leading conversational agent development platforms, revealing the predominance of AI and NLP capabilities framed around intent recognition and entity extraction. Provides framework for selecting appropriate tools... Read more
Key finding: Through practitioner interviews, identifies major challenges in conversational agent design, including gaps in socio-technical tooling, lack of common conversational archetypes, and divergence in design processes from... Read more