Context and Objectives

The Web-of-Things (WoT) envisions an ecosystem of applications that are able to interact with smart objects and services (collectively called things) which use the web (and web related technologies) as the main engine for their functionality. In such an ecosystem the web and its architecture can act like a global environment in which different types of agents (whether human or software) perform interactions and compose the functionality of web-enabled things to achieve their goals.

This is the vision of the HyperAgents community, which operates at the intersection of Multi-Agent Systems and web technologies. The HyperAgents vision holds enormous potential in terms of facilitating large-scale (internet-scale), open (provider-wise) and decentralized (privacy-preserving) application environments which are desirable both in industrial, as well as end-user settings. In social robotics, for example, it is highly important to easily understand the existing user environment (e.g. seamlessly discover and interact with available smart devices in a given location to achieve a user defined goal).

In facilitating this ability, one important issue is that of searching for and discovery of the relevant services and things. However, given the global nature of the web, as well as the dynamic availability of web-enabled things, the discoverability of things has to obey the principle of locality, that is, information about dynamic services and things is to be consumed in proximity (e.g. spatial, temporal, organizational) to where it is produced. In other words, access to the things needs to be context-dependent in order to ensure proper levels of security and privacy.

The project pursues the following set of research questions:

Q1. Expressive Context Representation: What is an appropriate means to model context information that is able to capture dynamic properties such as temporal validity or certainty?

Q2. Shared Context Definition: How is shared context modeled? How to state and validate the conditions which specify that a web resource and an agent share the same context and that the agent is authorized to access the resource in a context-aware manner?

Q3. Context-Aware Search and Discovery: How can existing procedures and tools for indexing and search of web-enabled things be augmented to facilitate context-aware access authorization?

Q4. Agent tools for context-aware discovery of web resources: Development of support APIs for different agent programming frameworks that offer support for handling notifications of being in a shared context and for searching and discovering of relevant web resources in that context.

To tackle the questions, the main objective of CASHMERE is the development of a framework for context-aware search and discovery of web-enabled things (services) in hypermedia-driven environments. The requirement is to facilitate the detection and understanding of shared context to enable corresponding discovery of and access to relevant web resources to agents operating in the web environment. The context is shared in between the web resources (producers of information / providers of capabilities) and the agents that require their functionality. This type of context-aware authorization is highly relevant in dynamic situations.