In this paper, we introduce a system, written in Haskell, for filtering information from XML data. Essentially,
the system implements a simple declarative language which allows one to extract relevant data as well as to exclude useless and misleading contents from an XML document by matching patterns against XML documents.
The matching mechanism employes a cost-based pattern transformation algorithm which searches for patterns in an approximate way (i.e. modulo renaming, insertion, and deletion of XML items) and ranks the results w.r.t. their cost. In order to improve efficiency, the implementation uses sophisticated indexing techniques and exploits laziness to automatically avoid the construction of unnecessary data structures. We analyzed both the expressiveness of our filtering language and the performance of the system using the well known XMark benchmark suite.
