A standard TMF modeling for Arabic patents

International audience Patent applications are similarly structured worldwide. They consist of a cover page, a speci cation, claims, drawings (if necessary) and an abstract. In addition to their content (text, numbers and citations), all patent publications contain a relatively rich set of well-de ned metadata. In the Arabic world, there is no North African or Arabian Intellectual Property O ce and therefore no uniform collections of Arabic patents. In Tunisia, for example, there is no digital collection of patent documents and therefore no XML collections. In this context, we aim to create a TMF standardized model for scienti c patents and develop a generator of XML patent collections having a uniform and easy to use structure. To test our approach, we will use a collection of XML scienti c patent documents in three languages (Arabic, French, and English).