What's merit in this context? What NLP/ML algorithms would you use? I can't think of any that are suited to or good enough for any definition of merit.
Using K-nearest neighbor (among others) against training documents to classify content, in areas you designate as interest. If they meet a predetermined relevance score and sufficient length range - they would then be considered for recommendation to the user.
What would this service have to offer for you to actually pay for it? Of course you could say you wouldn't pay for it, simply curious...