SRR [Koba02] doc title here

Posted in expert finder,Summary, Response and Relevance by haan on Saturday, December 10, 2005





Response (baddies in paper, first impressions)



This paper describes a domain model of expert finder systems and gives good guidance to examples of expert finder systems.



Posted in expert finder,Information Retrieval 2005/2006S1,literature by haan on Saturday, December 10, 2005

[Koba02] Coming up…



Posted in expert finder,Information Retrieval 2005/2006S1 by haan on Friday, December 9, 2005

Just about all things on earth work better with some form of a purpose, probably no exceptions here, so

this project, and the resulting prototype system is to allow people with some need for knowlegde to find the most adapt expert that can help them.

As this is about a broad as a purpose can come without losing specificity, we also define 

Expert systems are not at all new. Several systems exist [link to overview here]. 

In order to determine a user’s expertise, expert finder systems are known to have utillized email traffic, newsgroup postings, user web-sites, user’s scientific publications and even records of user’s internet surfing history as source for expertise indicators [Koba02].

[Koba02] notes several distinctive characteristics. For starters, expert finder system may be centralized (i.e. single server searching publically available information) or distributed (i.e. having an agent on every expert’s personal computer, gaining access to private sources too). Another distinction is in the need of human intervention. Systems may be fully automated or may need people to actively update an expertise database. This project is only interested in full-automatic, central systems, as this poses the lowest intrusion for possible experts. Related to this, the system will rely on user websites and scientific publications as its only source of expertise indicators.

For now, we’re focusing on finding the best way to model expertise, and let other important subjects such as the interpretation of user queries and the presentation of results to users be.

Research goal
Since both the expert finder system and your standard issue text-based information retrieval system rely on the same sources, how can an expert finder system achieve advantage over, let’s say, Google?

In particular, we’re interested in how the structure of hyperlinks inherent to websites of both persons and research groups may help us better link expertise to experts.