Semantic content analysis is an activity that can greatly sup- p ort a broad range of user mo delling applications. Several automatic to ols are available, however such systems usually provide little tuning p ossibilities and do not supp ort integration with dierent systems. Per- sonalization applications, on the other hand, are b ecoming increasingly multi-lingual and cross-domain. In this pap er we present a novel frame- work for Knowledge Extraction, whose main goal is to supp ort the de- velopment of new strategies and technologies and to ease the integration of the existing ones.