Recommender systems evaluate and filter the great amount of information available on the Web to assist people in their search processes. A fuzzy linguistic evaluation method of Web documents is presented to generate recommendations. Given an XML document type (e.g. scientific article), we consider that its components are not equally informative. This is indicated by defining linguistic importance attributes to the more meaningful elements of the XML Schema designed for Web documents. The evaluation method generates linguistic recommendations according to linguistic evaluation judgements provided by different recommenders on meaningful elements.