Machine learning algorithms can be used to build recommendation tools which allow to predict document relevance for a particular topic considering previous relevance judgements given for other documents for the same topic. Although this research domain has been often studied, tools for monitoring information on the web are very rare. Here, we present the web application called Re-Watch allowing topic definition, information source selection related to the topic and relevance score adaptation to user relevance feebacks. The application provides also, for each topic, an auto-evaluation of the filter quality and an adapted query to the Google search engine. The development of the application is still ongoing however it is currently available on the web at the following url : http://www.specific search.com.