Information quality is multidimensional and can be modeled as an ordered set of weighted criteria. For a collection of XML documents, our approach consists firstly in harvesting and generating information quality indicators and enriching the meta-description of XML documents. Quality metadata are then exploited within the query processing for metadata-driven information retrieval and filtering in order to propose quality-adaptive recommendation strategies (with a quality view as a result) of the queried XML documents. This quality view depends on the user's profile and on his specific information quality requirements. The paper describes the architecture of a quality-based re-commender system for XML documents.