Task Memories
Web Service Discovery and Recommendation with Task Memories
Improving Service Selection Through Management and Sharing of Task Memories
Web services promise to provide on-demand interactions among users and services. One of the challanges in realising the promise is to deliver useful user experience by choosing services tailored to their time, place and device capabilities. A large body of research exists in the general area of Web services discovery, selection and composition. In recent years, We have seen important outcomes from standardization efforts and semantic Web technologies towards the area. These approaches typically rely on UDDI-like service registry, keyword search on service description, or meta-data search such as service capabilities and non-functional properties (e.g., quality of service properties)
WS-Advisor is an service discovery, selection and composition framework, which complements the existing work in the research areas. It does so by incorporating the management and sharing user experience in the form of “task memories”. The design of the framework is based on the observations that most of the activities in service composition are repetitive in nature. The main objective of WS-Advisor is to model and capture the context associated with past activities in order to effectively repeat or adapt them in the future as needed. If the service composition system can remember how the users reacted to the chosen services in the past, it can use the knowledge (e.g. the information about the contexts in which a certain combination services were considered most appropriate by users) to assist the future service selection process by making relevant recommendations automatically.
The same idea can apply for handling exceptions. If the system learns how certain exception were dealt with in the past, the similar- ity between current and historical exceptions can be analysed to reuse, refine and generalise previous exception handling policies when possible. In this project, we introduce and formalise a notion of task memories, modelling and management of task memories and sharing of task memories in a social network. In particular, we propose a social network of task memories as a platform for mass sharing of the knowledge acquired during service selection process and rec- ommendations. We show how the task memories can support various ser- vice selection and composition strategies, namely local, global and goal-driven service selection.
References:
- Task Memories & Task Forums: A Foundation for Sharing Service-based Personal Processes, Bova R, Paik H Benatallah B, Zeng L and Benbernou S, 5th International Conference on Service Oriented Computing (ICSOC 2007) September 17-20, Vienna, Austria, pp.365-376 (PDF)
- WS-Advisor: A Task Memory for Service, Bova R, Paik H, Hassas S, Benbernou S and Benatallah B, 16th International Conference on Computer Communications and Networks (ICCCN 2007), August 13-16, Hawaii, USA, pp.535-540 (PDF)
- On Embedding Task Memory in Services Composition Frameworks, Bova R, Paik H, Hassas S, Benbernou S and Benatallah B, 7th International Conference on Web Engineering (ICWE 2007), July 16-20, Como, Italy, Springer, pp.1-16 (PDF)