Our Seminars in 2013

CaféDSL Seminars for 2013

Maintaining Goal Models Part Two

Abstract: Goal models need to be maintained as stakeholder motivation and domain assumptions evolve. An intrinsic problem of goal model maintenance is managing inconsistencies (e.g. when new stakeholder goals are contradictory with their existing motivtional representation). The problem of managing inconsistencies has a long history in requirements engineering. However, gaps have remained in leveraging the structure of a goal model and in e?cient procedures for computing optimal change options. In this talk I will go through some of the results that address the above problems. In particular, I provide an improved formalization of a goal model that has computational advantages in the maintenance process and a family of hierarchy aware maintenance operator classes and their efficient realization via A* search.

A new perspective of optimisation

Cognition, Computation and Information: A Survey

Abstract: Which notion of computation (if any) is essential for explaining cognition? In cognitive science five answers have been given to this question. (1) The classicist answer: symbolic (digital) computation is required for explaining cognition; (2) The broad digital computation list answer: digital computation broadly construed is required for explaining cognition; (3) The connectionist answer: sub-symbolic computation is required for explaining cognition; (4) The computational neuroscientist answer: neural computation (that, strictly, is neither digital nor analogue) is required for explaining cognition; (5) The extreme dynamicist answer: computation is not required for explaining cognition. We will not adjudicate on the correct answer to this problem, but rather point out that, to a large extent, the answer depends on what we take computation to be. In the second part of the talk we will argue that whether we take cognition to be computational or not, it seems clear that cognitive behaviour is underpinned by information processing. This raises a different, but related, question: ‘What types of information are required for cognition?’. We shall sketch a tentative answer to this question too.

BDI Agent Compliance

Abstract: Real-time traffic services such as TomTom HD Traffic service, Google Maps have been widely used in recent years. Based on the real-time traffic information, these services provide drivers a number of alternative routes for considering in order to avoid traffic jams, incident or road work. Re-routing strategies which are common to drivers that they would prefer to switch simultaneously from their current routes to alternative routes with the least estimated travel times. Unfortunately, German mathematician Braess points out in his discovered paradox that if most drivers choose a route that is promising for them then the overall performance which is the total travel time of all drivers will be reduced. Moreover, these traffic services which are the centralised traffic systems have a number of limitations such as processing time, system failure, low reactivity to the changes of environment and privacy issue. In this talk, we propose a model of decentralised traffic management system using novel approach that combines Distributed Constraint Optimisation Problems and Auction techniques. We show that by giving on-demand “collaborative” optimal routes to drivers, the traffic system is prevented from becoming a case of Braess’s paradox. Actually, by this proposed model, drivers would choose the routes that seem unlikely to affect the performance of other drivers as well as overall system. Mali The paper is prepared for submitting to Optimisation in Multi-Agent System (OptMAS) 2013 workshop in conjunction with AAMAS 2013.

Distributed Constraint Optimization: A Survey