The work presented in this paper addresses a project of the Computer Centre CIRTIL who supported it. This company wants to save and capitalize its knowledge and its know-how about the production activities, especially concerning the technical incidents relating to software applications encountered during the exploitation time. Indeed using a well accessing documents base, actors will be able to better solve problems. Our purpose is to focus on ontology-based framework for indexing documents. The domain ontology OntoCIRTIL has a structure which supports a semantic model based on semantic links and inference mechanisms. In this paper, we present a new model called S3 which permits to model knowledge in upstream and index documents (or formalized knowledge) in downstream. To illustrate partial results, this model is then applied to OntoCIRTIL.