Workshop on Automated Planning and Scheduling in New Methods of Electronic, Mobile and Collaborative Work DRAFT REPORT by Lee McCluskey The Workshop on Automated P&S in New Methods of E/M/C work was organised with the primary goal of exploring the new emphasis in PLANET 2 of P&S as a technology ripe for exploitation in the modern workplace. Despite being organised rather quickly the workshop contained eight talks, with written contributions from six speakers available in the Workshop Proceedings or online at the Workshop Website. Alfredo Milani, the workshop's co-chair, introduced the day's proceedings to over 30 delegates, outlining the day's schedule and stressing that the workshop concerned topics other than the world wide web. Jose Luis Ambite of ISI, California, gave an invited talk entitled "Heracles: Building Planning and Information Assistants". He pointed out that the web is full of disparate information in many formats, and the work of his research group was aimed at the automated assembly of information from various sources to satisfy structured information needs. He gave an online demonstration of a Travel Assistant that exemplified the approach. This was a tool which brought together information such as airfares, parking fares, timetables, weather conditions etc obtained using the web. Its function was to schedule travel for an individual given meeting times and places. It performed scheduling using hierarchical networks of constraints where constraints were propagated as their values became available. An interesting part of the work was the automated induction of "wrappers" for web sites: the user clicked on examples of interest on a page (e.g. the temperature in degrees) and the tool could create a parser for that page which could help translate it into XML. This formed the wrapper for the web site that could identify and extract pieces of information when required by the travel assistant. Florian Kandler of the Ecommerce Competence Centre, Vienna, gave a talk entitled "Scheduling in a Virtual Enterprise in the Service Sector". He was investigating holiday-package virtual enterprises which had to reconcile the demands from a set of customers with the potential resources drawn from a set of service providers. He described an algorithm which interfaced with service providers' schedules to arrive at a holiday schedule which satisfied the needs of a customer. Angelo Oddi of the Italian National Research Centre described a new tool called MEXAR which supports a mixed-initiative approach to planning and scheduling. The application is to solve the "Mars-Express Memory Dumping Problem", a space application involving the building of a schedule to send stored data from a spacecraft down to Earth. During mission planning for spacecraft, the requested payload operations must be translated into a detailed mission plan. MEXAR supports the mission planner in generating the detailed plan. Maria Dolores Moreno of the University of Alcala, Spain, gave a talk which concerned the use of planning technology in Business Process Re-engineering. In particular she discussed the automated generation of process models for business. New methods of electronic business have forced organisations to restructure their business processes. They may need to generate new process models quickly in the climate of continuous adaptation to market needs. Maria described a system in which users create a model of the business domain in a high level language that is transformed into the usual input for an AI planner (currently CMU's PRODIGY). The planner is then used to generate a plan (corresponding to a sequence of activities or task dependencies in a process model) which is fed back to the user in terms s/he can understand. Steven Smith of Carnegie-Mellon University, USA, started his talk by pointing out that the web amounts to an unprecedented infrastructure for the use of collaborative planning and scheduling. He described the COMIREM system (Continuous Mixed-Initiative Resource Management) which provides the basic set of resource management functions to underpin a web-based planning and scheduling tool. In particular, the system is aimed at being used in dynamic and resource constrained domains. Steve gave a live demonstration of a Special Forces Operations application showing how a schedule would dynamically update when constraints such as resources were changed. Richard Benjamins of Intelligent Software Components, Madrid, gave an invited talk on opportunities for AI software in the world of E-commerce. He discussed intelligent web applications, and how they might overcome the shortfall of current applications. Examples were online banking agents and shopping agents, where the problem was one of extracting meaningful information from heterogeneous web sites and reasoning with it to solve (commercial) goals. Intelligent search engines, in particular, that could search the dynamic web as well as the static web would need to be developed. The `Panel Discussion' was introduced by two short talks: Bob Laithwaite of APSolve described his experience on one of the largest applications of scheduling to date: BT's Workforce Management System. As the workforce received their schedules using electronic mobile appliances, this new application seemed to capture the essence of the workshop. Next Alfredo Milani summed up with a talk encompassing the whole scope of the workshop's area. He described the opportunities for P&S within core web technologies, as well as how intelligent web-based activities might use P&S, and how the onset of the Semantic Web gave new opportunities for utilising P&S technology. Alfredo stressed that what we were about was much more than just web-oriented work, and he described opportunities for collaborative and mobile P&S technologies in new methods of work. Finally, the speakers and the delegates in general were asked for their opinions on the most desirable new developments in this area. Three conclusions follow, with the first being the most important: a) The development of a common semantics for web sites, possibly through the use of XML; b) Being able to assemble and configure existing primitive services on the web for use in P&S; c) The further development of techniques in distributed, collaborative P&S.