Scientific articles published in electronic format are knowledge bases, especially in Medicine. An obstacle to semantic processing of this knowledge by computers is that in spite of their digital format, articles are in text format for human reading and processing. A model is proposed for electronic publishing scientific articles both in textual format and in machine "understandable" format, in ontology format. Software agents can process the content of an article published according to the model, thus enabling semantic retrieval, consistence checking and the identification of new discoveries. The model is described and initial steps toward the development of an authoring/publishing system which implements the model proposed are related.