Paper Info
Title
Inferring Commitment Semantics In Multi-Agent Interactions
Abstract
Commitments are a useful abstraction to specify the social semantics of multi-agent communication languages. To use them in open and heterogeneous systems, it is necessary to develop solutions to the problem of interoperability, an effort that has already provided methods to, for example, align commitments between interlocutors. In this paper we consider the problem of commitment semantics inference, which can be summarized as follows: how can an agent that arrives to a community with an established language discover its social semantics, only by observing interactions? We introduce a method based on simple learning techniques that tackles this problem. We show that the basic commitment semantics is not possible to infer, and discuss different ways of enriching it that make inference feasible. We show experimentally how our technique performs for each of these extensions.
Year
DOI
Venue
2018
10.5555/3237383.3237867
PROCEEDINGS OF THE 17TH INTERNATIONAL CONFERENCE ON AUTONOMOUS AGENTS AND MULTIAGENT SYSTEMS (AAMAS' 18)
Keywords
Field
DocType
Commitments, Semantic Inference, Multi-Agent Communication
Programming language,Abstraction,Interoperability,Inference,Computer science,Artificial intelligence,Social semantics,Semantics,Machine learning
Conference
Citations 
PageRank 
References 
0
0.34
0
Authors
2
Name
Order
Citations
PageRank
Paula Chocron1153.80
W. Marco Schorlemmer2111385.18