Abstract
Fluent, effortless and diverse e-business transactions depend on the ability of automated agents to interact. The difficulties of tailoring representation and information to be consistent and therefore interoperable needs to fall not on human users but on these automated agents. In this chapter, the authors present our system, ORS (Ontology Repair System), which is designed to be a tool for automated agents, acting on behalf of people or systems, which need to interact, to enable them to understand one another, despite the fact that they are not centrally or consistently designed.
TopFocus Of Approach
In a system such as the Semantic Web, where there is no centralised control, we cannot have a complete global overview of the agents and data in the system. Agents may join and leave the system freely and they will all have their own ontologies and data that may be large and complex and may be confidential. We cannot hope for a complete description of the relations between every agent in the system. Our approach is therefore not to consider how such a system can be controlled but how an individual agent can successful make its way in such a system, interacting with the agents that it needs to interact with, even if these agents are not using the same ontological terms or representations, and even if it is not known in advance of the interaction which agents these will be.
Key Terms in this Chapter
Semantic Mismatch: Mismatch between two ontological terms where the structure is the same but the meanings of the words within the terms is not.
ORS: Ontology repair system which equips a planning agent to proceed successfully with problematic interaction by identifying and repairing any ontological problems which may have led to the difficulties.
Planning Agent (PA): An agent which responds to goals through forming plans to achieve them and then enacting these plans through interaction within a domain.
Ontology: Formal representation of domain knowledge, containing a class hierarchy and possibly relations and functions between these classes.
Service Providing Agent (SPA): An agent which is able to provide a service to a planning agent, providing the correct conditions are met.
Surprising Questions: When two agents are communicating with a particular goal in mind, a question that is posed by one agent which is not thought to be pertinent to the situation by the other agent is considered by that agent to be a surprising question.
Structured Ontological Term: An ontological term, such as a relation or function, the meaning of which is determined not only by the meanings of the words in the term but also by their structure.
Structural Mismatch: Mismatch between two ontological terms where the meanings of the words within the terms is the same but their structure is not (two ontological objects may be mismatched both semantically and structurally).