Building an Active Content Warehouse

Abstract

Non-quantitative content represents a large part of the information available nowadays, such as Web pages, e-mails, metadata about photos, etc. In order to manage this new type of information, we introduce the concept of content warehousing, the management of loosely structured data. The construction and maintenance of a content warehouse is an intricate task, involving many aspects such as feeding, cleaning and enriching semi-structured data. In this chapter, we introduce the Acware (for active content warehouse) specification language, whose goal is to help all sorts of users to organize content in a simple manner. The problem we are faced with is the following: The data are semi-structured, and the operations to be executed on this data may be of any sort. Therefore, we base our approach on XML to represent the data, and Web Services, as genericcomponents that can be tailored to specific applicative needs. In particular, we discuss the specification of mappings between the warehouse data and the parameters/results of services that are used to acquire and enrich the content. From the implementation point of view, an Acware specification of a content warehouse is compiled into a set of Active XML documents, i.e., XML documents with embedded Web service calls. These Active XML documents are then used to build and maintain the warehouse using the Active XML runtime environment. We illustrate the approach with a particular application drawn from microbiology and developed in the context of the French RNTL e.dot project.