Measuring the understanding between two agents through concept similarity
Two agents previously unknown to each other cannot communicate by exchanging concepts (nodes of their own ontology): they need to use a common communication language. If they do not use a standard protocol, most likely they use a natural language. The ambiguities of it, and the different concepts th...
Gespeichert in:
Veröffentlicht in: | Expert systems with applications 2006-05, Vol.30 (4), p.577-591 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Two agents previously unknown to each other cannot communicate by exchanging concepts (nodes of their own ontology): they need to use a common communication language. If they do not use a standard protocol, most likely they use a natural language. The ambiguities of it, and the different concepts the agents possess, give rise to
imperfect understanding among them: How closely concepts in ontology O
A map
1
1
O
A and O
B are the ontologies of agents A (the talker or sender) and B (the listener), in the rest of the paper.
to which of O
B? Can we measure these mismatches?
Given a concept from ontology O
A, a method is provided to find the
most similar concept in O
B, and to measure the similarity between both concepts. The paper also gives an algorithm to gauge
du(A, B), the
degree of understanding that agent A has about the ontology of B. The procedures use word comparison, since no agent (except the Very Wise Creature, VWC) can measure
du directly. Examples are given. |
---|---|
ISSN: | 0957-4174 1873-6793 |
DOI: | 10.1016/j.eswa.2005.07.003 |