Agents on the Web

Motivation

The World Wide Web has emerged as the application layer of choice for most distributed systems envisioned on the Internet. The evolution towards Hypermedia APIs, Linked Data, and the Web of Things is turning hypermedia into a homogeneous information fabric that interconnects heterogeneous Resources - devices, physical objects, information resources, abstract concepts, etc. The latest standards allow automated software clients to navigate, query, observe, and act on this hypermedia fabric in a reliable manner. Use cases that have long been envisioned for artificial agents on the Web - as published in the original Semantic Web vision in 2001 and going back to the early days of the Internet - are now closer to their practical implementation and deployment. Nevertheless, the engineering of such modern Web-based systems poses research challenges that have yet to be addressed.

Today’s Web-based systems are inherently complex, heterogeneous, and increasingly dynamic (e.g., in particular in the context of the Web of Things). Moreover, the World Wide Web is inherently decentralized, designed to span not only geographical boundaries but also organizational boundaries. In such settings, traditional paradigms for engineering Web-based systems become impractical. Many of the research questions underlying these challenges - such as how to balance goal-directed and reactive behavior, or how to design and govern interactions in decentralized systems - have been investigated in the scientific literature on autonomous agents and multi-agent systems. Thus, this research needs to take a broader perspective that can only be achieved through a concerted effort of the research communities around Web Architecture and the Web of Things, Semantic Web and Linked Data, and Autonomous Agents and Multi-Agent Systems.

This Dagstuhl Seminar aims to consolidate and further investigate the research opportunities identified in the Dagstuhl Seminar 21072 (Autonomous Agents on the Web) , and to continue the transfer of knowledge and results across the involved research communities. We believe this seminar can break new ground in all these areas of research - and can help pave the way for a new generation of Web-based autonomous systems composed of people and intelligent agents interacting and collaborating through the Web. Concretely, the seminar will pursue the following objectives:

• to identify and align the various research threads in the targeted communities that are relevant for advancing the research on Web-based autonomous systems;
• to work towards a shared conceptualization and shared theoretical underpinnings for Web-based autonomous systems;
• to identify representative use cases in different domains that would help demonstrate the potential impact of this joint research effort on society and economy;
• to evaluate the state of technologies available for prototyping and deploying Web-based autonomous systems.