Derek P. Long

Goal creation in motivated agents

Goal creation is an important consideration for an agent that is required to behave autonomously in a real-world domain. This paper describes an agent that is directed, not by a conjunction of top level goals, but by a set of motives. The agent is motivated to create and prioritise different goals at different times as a part of an on-going activity under changing circumstances.

'Slick Systems' and 'Happy Hackers': experience with group projects at UCL

The authors describe their experience of group projects in the practical teaching of software engineering over a period of eight years. Their initial projects tended to be too challenging, and few groups managed to produce complete pieces of work. They have deliberately simplified tasks slightly, resulting in less frustration and better projects, so that students reap more benefits. Students learn about division of work, co-operation with others and scheduling of time. As students are required to provide assessments of other projects and of the contributions of members of their own project group, they are also encouraged to develop critical faculties. The staff effort involved in this method of teaching compares quite reasonably with traditional lectures.

A model for learning by source control

The information we receive is often changing, inconsistent and incomplete, thus bound to generate contradictions. Clearly we must recover reasonably from inconsistencies if to make sense of the world. We introduce a model to cope with this situation. This paper continues the work presented in (6,7). We propose that an adaptive reasoning system should use a model of its sources, recognise patterns in their behaviour and adjust that model on the basis of evidence and general principles. The relation with TMS is then discussed.

Persuasion as a Form of Inter-Agent Negotiation

Agents in a multi-agent environment must often cooperate to achieve their objectives. In this paper an agent, B, cooperates with another agent, A, if B adopts a goal that furthers As objectives in the environment. If agents are independent and motivated by their own interests, cooperation cannot be relied upon and it may be necessary for A to persuade B to adopt a cooperative goal. This paper is concerned with the organisation and construction of persuasive argument, and examines how a rational agent comes to hold a belief, and thus, how new beliefs might be engendered and existing beliefs altered, through the process of argumentation. Argument represents an opportunity for an agent to convince a possibly sceptical or resistant audience of the veracity of its own beliefs. This ability is a vital component of rich communication, facilitating explanation, instruction, cooperation and conflict resolution. An architecture is described in which a hierarchical planner is used to develop discourse plans which can be realised in natural language using the LOLITA system. Planning is concerned with the intentional, contextual and pragmatic aspects of discourse structure as well as with the logical form of the argument and its stylistic organisation. In this paper attention is restricted to the planning of persuasive discourse, or monologue.

A Note on Chapman`s Modal Truth Criterion

The formal statement of the Modal Truth Criterion presented by Chapman [1] is shown not to be necessary, in contradiction of Chapmans original claim that the criterion is both necessary and sufficient. This is achieved through the presentation of a counter-example which is a partial plan in which a goal is necessarily true, and yet fails to satisfy Chapmans criterion. A corollary of this result is that the validity checking problem for partial plans, claimed by Chapman to be polynomially solvable, is now open.

Inheritance in hierarchical relational structures

A brief survey is conducted of the inheritance principle - the conveyance of properties between components within a hierarchical relational structure. The standard form of inheritance is considered, using the subset (is-a) relation and highlighted as an example of downward inheritance. Downward inheritance is extended to specialisation of actions, and cases are presented in which the rule fails.An alternative and less well-known form of inheritance is introduced - upward inheritance. Several examples in which upward inheritance is valid and others in which it is not valid are given, in a treatment highlighting the analogy with downward inheritance. The validity of the rule, in those case in which it operates, is underlined, to distinguish it from induction.A brief account is given of the search for the underlying reasons for the validity of inheritance rules and these are then given. The solution turns out to be due to a hidden or implicit quantifier within the relations that are used. The semantical nature of the problem and of its solution are stressed, emphasising the impossibility of a purely syntactic analysis and solution to the problem. Various points of interest arising from the analysis are listed and discussed.