Notes taken from ‘A Generative Inquiry Dialogue System’ (2007), by Elizabeth Black and Anthony Hunter
“… We focus on inquiry dialogues that allow two agents to share knowledge in order to construct an argument for a specific claim…”
1, Introduction
Dialogue games are now a common approach to defining communicative agent behaviour, especially when this behaviour is argumentation-based… Dialogue games are normally made up of a set of communicative acts called moves, a set of rules that state which moves it is legal to make at any point in a dialogue (the protocol), a set of rules that define the effect of making a move, and a set of rules that determine when a dialogue terminates. Most of the work so far has looked at modelling different types of dialogue in the Walton and Krabbe typology… here we provide a generative system.
... A key contribution of this work is that we not only provide a protocol for modelling inquiry dialogues but we also provide a specific strategy to be followed, making this system sufficient to also generate inquiry dialogues… and this allows us to consider soundness and completeness properties of our system.
2, Motivation
Our work has been motivated by the medical domain. Argumentation allows us to deal with the incomplete, inconsistent and uncertain knowledge that is characteristic of medical knowledge. There are often many different healthcare professionals involved in the care of a patient, each of whom has a particular type of specialised knowledge and who must cooperate in order to provide the best possible care for the patient…
Inquiry dialogues are a type of knowledge that would be of particular use in the healthcare domain, where it is often the case that people have distinct types of knowledge and so need to interact with others in order to have all the information necessary to make a decision…
… We compare the outcome of our dialogues with the outcome that would be arrived at by a single agent that has at its beliefs the union of both the agents participating in the dialogues beliefs. This is, in some sense, the ideal situation, where there are no constraints on the sharing of beliefs.
3, Knowledge Representation and Arguments
We adapt Garcia and Simari’s Defeasible Logic Programming (DeLP)… for representing each agent’s beliefs…
The presentation in this section differs slightly from that in (Garcia and Simari’s DeLP)… as (they) assume a set of strict rules, which we assume to be empty, and they assume facts to be non-defeasible. We assume that all knowledge is defeasible due to the nature of medical knowledge, which is constantly expanding…
3.1, A defeasible rule is denoted 'alpha1 ^ … ^ alphaN -> alpha0' where alphai is a literal… A defeasible fact is denoted 'alpha' where alpha is a literal. A belief is either a defeasible rule of a defeasible fact.
3.2, A belief base associated with an agent x is a finite set…
3.3… A defeasible derivation of (a literal) ‘alpha’ from (a set of beliefs)… is a finite sequence alpha1, alpha2, …, alphaN of literals such that alphaN is alpha and each literal… is in the sequence because…
3.4, An argument constructed from a set of, possibly inconsistent, beliefs… is a minimally consistent set from which the claim can be defeasibly derived…
No comments:
Post a Comment