|
|
|
| Description: |
This research is part of the Alter Ego project, a joint research project of (among others) the IBM Center for Advanced Studies and the Telematica Instituut. The topic of this project is cross-domain user profiling and how a generic user profile can simplify and optimize the benefits of services and products. Alter Ego considers a user profile to be some structured description of a particular individual. The user profile can contain any user related information, like personal and professional interests and food preferences but it can also contain medical and financial data. Cross-domain user profiling is the creation, management and usage of such a user profile over different domains (e.g. entertainment, health, education, government etc.). The problem with handling profile information that originates from different domains is that the information is created, managed and provided by more than one party. Thus it is very likely the information that is to be contained in a generic user profile will have semantic disparities and will also be distributed over (physically) separated locations. Within the IBM Center for Advanced Studies there is a firm belief this kind of problem can be solved with semantic web services. All profile information provided by a certain organization will be semantically described by an ontology and made available through a semantic web service. The objective of this research is to design and prototype an architecture, based on semantic web services technology, that supports the registration, discovery, composition and querying of distributed, cross-domain profile services related to a single individual. Several questions arise from this objective: What components are necessary for such a framework? How can cross-domain profile information be described? How does an application/service query the framework for profile information? How can the framework discover relevant information to a query?
|
|
|
| Description: |
A problem with standard searching techniques (like google) is that they do not take the context of a search query into consideration. For example when doing preliminary research for the construction of a road and using standard searching techniques the user has to identify every aspect manually; these aspects could include endangered species, other construction projects in the vicinity or residential areas. When searching for these aspects manually it is easy to overlook certain aspects, which can have major consequences. Context Aware Search is our solution for this problem.
We use the concept of Context Aware Search to determine relevant aspects for a search query. Our solution takes a high level (and thus complex to answer) query, like "road construction A15 Rotterdam Maasvlakte", and determines relevant aspects to this query. For each of the aspects found the system searches for information sources relevant to the aspect. The searching for the relevant aspects and information sources is done using intelligent agent solution. This agent solution uses a marketplace analogy to create context. The concept is demonstrated by creating a proof of concept demo. This demo is developed in collaboration with Rijkswaterstaat and is targeted to improve a part of the 'Tracé/MER' process.
| |
|
