Broadway is a browsing advisor reusing past navigation from a group of users. Broadway is able to follow the user during a browsing session to infer his goal, and then it advise him of potentially relevant documents to visit next.
The computation of advice is based on Case-Based Reasoning (CBR) problem solving paradigm. Our approach is based on the following hypothesis:
if two users went through a similar sequence of documents, they might have similar browsing intent, so that we can advise one user of the documents evaluated as relevant by the other one
Our goal is to follow each user in its navigation through the WWW and the activity of this kind of browsing assistant is highly coupled with the streams of requests and responses. Thus, we have designed Broadway as a stream transducer (Proxy) to watch all navigation moves of a group of users. Broadway also manages an internal information space that can be accessed by users through a graphical interface mainly to get the advised pages and submit their evaluation. Based on this external architecture, Broadway integrates a case-based reasoner that computes advised pages asynchronously from the user's navigation.
In addition, Broadway is implemented in Java using our
software library CBR*Tools that supports our framework for the
management of cases with time-extended situations and the HTTP
server Jigsaw from the W3C.
|
|
|
