An important use case of document transformations is the consumption and construction of web service messages. Such messages often contain a variety of information items with subtle semantics, best understood by subject matter experts. These experts play a key role in the correct implementation of transformations. How can their contribution be made as efficient as possible?

Creating a SNAT tree does usually not require genuine coding skills. Most entries are simple path expressions identifying source items by an item name or a path which is a sequence of item names. Nevertheless, as a rule also a few more complex expressions are required, which would be difficult to supply for a person who is not a software developer.

The question arises if SNAT trees might themselves be generated from simpler input, which can be provided by subject matter experts without an IT background. The core of their expertise is a thorough understanding of item semantics on both sides, source and target. The expert recognizes the alignments between source items and target items, and in this section we explore possibilities of leveraging this competence in an optimized way. The idea is to obtain SNAT documents from an identification of alignments and a minimum of additional information.