Deciding Agent Orientation on Ontology Mappings

Abstract

Abstract.Effective communication in open environments relies on the ability of agents to reach a mutual understanding of the exchanged message by reconciling the vocabulary (ontology) used. Various approaches have considered how mutually acceptable mappings between corresponding concepts in the agents' own ontologies may be determined dynamically through argumentation-based negotiation (such as Meaning-based Argumentation, MbA). In this paper we present a novel approach to the dynamic determination of mutually acceptable mappings, that allows agents to express a private acceptability threshold over the types of mappings they prefer. We empirically compare this approach with the Meaning-based Argumentation and demonstrate that the proposed approach produces larger agreed alignments thus better enabling agent communication. Furthermore, we compare and evaluate the fitness for purpose of the generated alignments, and we empirically demonstrate that the proposed approach has comparable performance to the MbA approach.

Generating Arguments for Ontology Matching

“…[4], [5]). First, it makes possible through the condition function that the fact of accepting/rejecting a given correspondence influences positively or negatively the acceptance/rejection of others correspondences.…”

“…Approaches relying on an argument-based negotiation such as [4], [5] were proposed to allow agents to reach a consensus about the correspondences that must be part of the alignment enabling them to communicate and mutually understand each other. However, on those approaches the correspondences are provided by a single ontology alignment service that is common to all agents.…”

Improved Relaxation-based Ontology Matching Negotiation

“…Meaning-based Argumentation (MbA) [15], further improved in [16] into a more flexible approach (FDO), adopts the Value Argumentation Framework (VAF) [2]. Agents can express their matching preferences according to the classification of the matching algorithms: Arguments are represented as 3-uples, where is a match, is one of , depending on the agent's belief that the correspondence does or does not hold, and is the grounds justifying .…”

Argumentation for Reconciling Agent Ontologies

“…The use of confidence profiles has since been explored to specify correspondence-type specific thresholds, resulting in the agreement over a greater diversity of agreed correspondences, and consequently more inclusive alignments[9].…”