LOGIN
>> Home
>> Topics
>> Students
>> Partners
>> Statistics


Information for topics

Topic Id:
ID topic: 468
Partner Email: L.J.M.Rothkrantz@tudelft.nl
Project Title: Agent technology for personalised unified messaging
Abstract: Now that telecommunications and Internet are merging to one global communication network, new possibilities and their related problems arise. Unified messaging as used in practise nowadays only provides for the exchange of data. It is much more complicated if one wishes to receive the information rather than the data. More and more people start to complain about the large amounts of e-mail they receive every day. User profiling has often been applied to e-mail as text, but with unified messaging one uses various media. The combination of unified messaging and user profiling with agent technology is a last element of this study. To handle this problem, an architecture has been developed. This architecture is based on multi- and mobile agent technology. Intelligent user profiling is directly incorporated in the design. Due to the separation of profiling in two phases, it can be achieved across different media. This is accomplished by using meta-data to describe a message, rather than using its content directly. The usage of agents supports a high level of personalisation of the system. The proposed architecture has been implemented in a prototype version. This prototype was used for an evaluation of the concept. This evaluation shows that agents are great facilitators for personalised unified messaging. Overcoming a communication overload is handled by an intelligent agent that is capable of learning an user profile. This learning problem is clearly defined, and a possible solution is found in the k-nearest-neighbour algorithm. Further benefits of the usage of agents are shown as well. The system is dynamically extensible and highly adaptable, with robustness in a distributed environment.
Advisor: Leon Rothkrantz
Link:
Degree: Master
 Keywords:
Computer Software
Algorithms & problem solving
Artificial intelligence & Neural networks
Data modeling