Online Genetic Algorithms

This paper present a technique based on genetic algorithms for generating online adaptive services. Online adaptive systems provide flexible services to a mass of clients/users for maximising some system goals, they dynamically adapt the form and the content of the issued services while the population of clients evolve over time. The idea of online genetic algorithms (online GAs) is to use the online clients response behaviour as a fitness function in order to produce the next generation of services. The principle implemented in online GAs, “the application environment is the fitness”, allow modelling highly evolutionary domains where both services providers and clients change and evolve over time. The flexibility and the adaptive behaviour of this approach seems to be very relevant and promising for applications characterised by highly dynamical features such as in the web domain (online newspapers, e- markets, websites and advertising engines). Nevertheless the proposed technique has a more general aim for application environments characterised by a massive number of anonymous clients/users which require personalised services, such as in the case of many new IT applications.

A Two Layered Model for Evolving Web Resources

In this paper the key features of a two-layered model for describing the semantic of dynamical web resources are introduced. In the current Semantic Web proposal [Berners-Lee et al., 2001] web resources are classified into static ontologies which describes the semantic network of their inter-relationships [Kalianpur, 2001][Handschuh & Staab, 2002] and complex constraints described by logical quantified formula [Boley et al., 2001][McGuinnes & van Harmelen, 2004][McGuinnes et al., 2004], the basic idea is that software agents can use techniques of automatic reasoning in order to relate resources and to support sophisticated web application. On the other hand, web resources are also characterized by their dynamical aspects, which are not adequately addressed by current web models. Resources on the web are dynamical since, in the minimal case, they can appear or disappear from the web and their content is upgraded. In addition, resources can traverse different states, which characterized the resource life-cycle, each resource state corresponding to different possible uses of the resource. Finally most resources are timed, i.e. they information they provide make sense only if contextualised with respect to time, and their validity and accuracy is greatly bounded by time. Temporal projection and deduction based on dynamical and time constraints of the resources can be made and exploited by software agents [Hendler, 2001] in order to make previsions about the availability and the state of a resource, for deciding when consulting the resource itself or in order to deliberately induce a resource state change for reaching some agent goal, such as in the automated planning framework [Fikes & Nilsson, 1971][Bacchus & Kabanza,1998].

Planning Technologies for the Web Environment: Perspectives and Research Issues

This work will explore and motivate perspectives and research issues related with the applications of automated planning technologies in order to support innovative web applications. The target for the technology transfer, i.e. the web, and, in a broader sense, the new Information Technologies (IT) is one of the most changing, evolving and hottest areas of current computer science. Nevertheless many sub-area in this field could have potential benefits from Planning and Scheduling (P&S) technologies, and, in some cases, technology transfer has already started. This paper will consider and explore a set of topics, guidelines and objectives in order to implement the technology transfer a new challenges, requirements and research issues for planning which emerge from the web and IT industry. Sample scenarios will be depicted to clarify the potential applications and limits of current planning technology. Finally we will point out some new P&S research challenge issues which are required to meet more advanced applicative goals.

Managing Interval Resources in Automated Planning

In this paper RDPPLan, a model for planning with quantitative resources specified as numerical intervals, is presented. Nearly all existing models of planning with resources require to specify exact values for updating resources modified by actions execution. In other words these models cannot deal with more realistic situations in which the resources quantities are not completely known but are bounded by intervals. The RDPPlan model allow to manage domains more tailored to real world, where preconditions and effects over quantitative resources can be specified by intervals of values, in addition mixed logical/quantitative and pure numerical goals can be posed. RDPPlan is based on non directional search over a planning graph, like DPPlan, from which it derives, it uses propagation rules which have been appropriately extended to the management of resource intervals. The propagation rules extended with resources must verify invariant properties over the planning graph which have been proven by the authors and guarantee the correctness of the approach. An implementation of the RDPPlan model is described with search strategies specifically developed for interval resources.

Timed Transition Automata as Numerical Planning Domain

A general technique for transforming a timed finite state automaton into an equivalent automated planning domain based on a numerical parameter model is introduced. Timed transition automata have many applications in control systems and agents models; they are used to describe sequential processes, where actions are labelling by automaton transitions subject to temporal constraints. The language of timed words accepted by a timed automaton, the possible sequences of system or agent behaviour, can be described in term of an appropriate planning domain encapsulating the timed actions patterns and constraints. The time words recognition problem is then posed as a planning problem where the goal is to reach a final state by a sequence of actions, which corresponds to the timed symbols labeling the automaton transitions. The transformation is proved to be correct and complete and it is space/time linear on the automaton size. Experimental results shows that the performance of the planning domain obtained by transformation is scalable for real world applications. A major advantage of the planning based approach, beside of the solving the parsing problem, is to represent in a single automated reasoning framework problems of plan recognitions, plan synthesis and plan optimisation.