University of Macedonia Economic and Social Sciences
Decision making in non-deterministic environments
Abstract
dc:descriptionThis dissertation tackles the problem of automated web service composition, taking into account the non-deterministic nature of web services. Web services, a major ingredient of the Semantic Web, aim to solve interoperability problems between heterogeneous systems, with transparency over the underlying technologies used to implement them and the platforms they are based on. Importantly, due to the semantic markup of a web service, it is possible to automate the tasks of web service discovery, selection and composition.Since no single web service may provide the desired functionality, it is necessary to perform the task of web service composition in order to achieve it. However, web services exist and operate in an ever-changing and expanding environment; for that reason, searching for the appropriate web services for each goal is hard. What makes web service composition difficult and time-consuming is the additional burden of monitoring whether a web service taking part in an existing solution is still active and has the same usage and interface. Moreover, the possible uncertainty in the web service’s execution results must also be taken into account in the web service composition process.In the last decades, a plethora of methods for planning in classical domains have been proposed providing promising results. In parallel, and frequently inspired by such deterministic methods, algorithms and tools for planning in non-deterministic settings, either in fully, partially, or non-observable domains have been developed. Such methods can be efficiently utilized for the purposes of automated web service composition, using existing tools, following a translation of the original semantic web service composition domain to an AI planning one.The dissertation has three main parts. First, it critically reviews existing non-deterministic AI planning approaches, as well as deterministic and non-deterministic web service composition ones. Second, it reviews the evaluation process in web service composition approaches and proposes a new set of reproducible use case scenarios. Third, and most importantly, it proposes a non-deterministic AI planning algorithm aimed for automated web service composition. The aforementioned have been implemented into a web-based application, MADSWAN, that comprises a semantic web service registry, an editor of semantic description files, as well as manual and automated web service composition modules. The system adheres to the reusable nature of web services by utilizing existing open source projects as sub-element and makes use of an anytime probabilistic planner, MAPPPA2, tailor-made for web service composition problems so as to tackle the inherent non-determinism in the domain.
Degree
thesis:*- Grantor dc:publisher
- University of Macedonia Economic and Social Sciences
- Year dc:date
- 2015
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Markou, Georgios
- Μάρκου, Γεώργιος
Subjects
dc:subject × 20- Υπηρεσίες Ιστού
- Σύνθεση
- Αβεβαιότητα
- Πλάνα πολλαπλών ενδεχομένων
- Κριτική ανασκόπηση
- Σχεδιασμός ενεργειών
- Web services
- Composition
- Non - determinism
- Contingent plans
- Critical review
- Planning
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Επιστήμες Μηχανικού και Τεχνολογία
- Άλλες Επιστήμες Μηχανικού και Τεχνολογίες
- Natural Sciences
- Computer and Information Sciences
- Engineering and Technology
- Other Engineering and Technologies
Rights
dc:rights- Statement dc:rights
-
- BY_NC_SA
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/41773