Many ubiquitous computing applications have been developed to support the capture of human experiences towards providing documents for later review. In most of the cases, those documents are generated after the end of each captured session, leading to the duplication of data. Moreover, as those applications were developed as ad hoc solutions, the mechanisms implemented both for capturing and for presenting the captured content are usually hard to be reused and extended. We tackled this problem creating a component-based infrastructure that uses document processing techniques as one of its main building blocks. In this paper we present the architecture of our infrastructure from the document-processing point of view, showing that our approach makes it easier for exploiting tools that allows the presentation of the captured content in different formats on the web.