Your browser does not support JavaScript!

Home    Collections    Type of Work    Post-graduate theses  

Post-graduate theses

Search command : Author="Στεφανίδης"  And Author="Κωνσταντίνος"

Current Record: 4650 of 6472

Back to Results Previous page
Next page
Add to Basket
[Add to Basket]
Identifier uch.csd.msc//2000kalopsikakis
Title Επαναλαμβανόμενη Διαπραγμάτευση μέσω Ηλεκτρονικού Αντιπροσώπου για Αποδοτική Επιλογή Συμβολαίου Τηλεπικοινωνιακής Υπηρεσίας
Alternative Title Repeated Agent-Baed Negotiation for Efficient Contract Selection of Telecommunication Service
Creator Kalopsikakis, Dimitrios E
Abstract Telecommunications markets are becoming increasingly competitive due to both technological and regulatory changes. Moreover, the wide diversity and the level of customization of services have already made it possible for users to buy affordable services matching well their preferences. However, the optimal service selection, may prove a very complicated task in case there are several ``degrees of freedom'' involved. On the other hand, this task is appropriate for a software agent that will negotiate (and/or re-negotiate) with service retailers on behalf of the user. In the present Master's Thesis, we analyze an efficient heuristic algorithm for selecting the contract for an application-level telecommunications service (e.g. News-on-Demand) based on user preferences reflected in an approximation of the userutility function. Our work emphasizes on the application of this algorithm to cases that negotiation between user and service retailer is repeated. First, we analyse the case of re-negotiation or re-selection of service contract. The main purpose is to find some computationally simple conditions based on which the agent can either decide for the new outcome of the selection or decide to re-execute the selection algorithm. This analysis showed that, under realistic assumptions, there can be established simple conditions under which the re-execution of the selection algorithm can be avoided in various interesting cases. In the sequel, the selection algorithm is enriched with a learning process aiming at adaptation of the agent to varying conditions, such as unstable (with respect to their preferences) users or randomly varying charging. The main idea is to offer the user the ability to express a preference that is possibly different from the agent's recommendation. This information is exploited by means of adjustments performed to the internal utility model that is maintained by the agent. As a result, the approximation of user's actual preferences improves. We then present an assessment of this learning process by means of a simulation environment. The experimental results show that the user after a number of successive selections is indeed approximated well by the agent, at least, with respect to his service contract selections. Finally, we study the applicability of the above ideas with certain technologies of the underlying network. In particular, we examine the cases of IPv4, IPv6, Diffserv, ATM and Frame Relay, and we argue that the proposed approach is indeed applicable to all of them.
Issue date 2000-03-01
Date available 2000-04-03
Collection   School/Department--School of Sciences and Engineering--Department of Computer Science--Post-graduate theses
  Type of Work--Post-graduate theses
Permanent Link https://elocus.lib.uoc.gr//dlib/d/d/3/metadata-dlib-2000kalopsikakis.tkl Bookmark and Share
Views 402

Digital Documents
No preview available

Download document
View document
Views : 5

No preview available

View document