The increasing usage of personal computers and Internet in organizations made it possible to create, edit and share various documents between different employees of the organization. However, document management becomes very troublesome, especially if several employees can contribute changes to a singe document: it is very hard to locate latest document version, or determine which changes to the document were made by which employee. Document management systems are aimed to solve these problems. However these systems often are highly specialized and very costly to implement, or they are general-purpose and hard to customize and apply to organization business domain. Inabilities to customize often arise from strict and inflexible metadata model, used in document management system. The aim of this work is propose abstract document metadata definition and processing model, based on XML data definition language and concept of XML data processing pipeline. The proposed model is general-purpose and highly flexible at the same time, enabling to apply model to any business domain and customize it to reflect any features specific to this domain.