Soon after its introduction, relevance feedback turned
out to be of much value in Information Retrieval. The 
method yields a new query that should match the users 
information need with better results than the initial 
query does. In order to reformulate the initial query, 
a (Rocchio) learning algorithm is frequently used. 
Another approach, which is relatively new and origi-
nates from artificial intelligence research, is the 
use of a genetic algorithm. 

In this paper, we explain both algorithms and we 
compare relevance feedback using a Rocchio algorithm 
with relevance feedback using a genetic algorithm. We 
report the effectiveness of the two approaches to 
relevance feedback in the context of information 
filtering.