We are investigating and applying a semantically enhanced query answering machine for the needs of addressing semantically meaningful data and operations within a scientific information system. We illustrate a context based
querying paradigm on the basis of a Regional Avalanche Information and Forecasting System - RAIFoS which is concerned with the collection and analysis of snow and weather related physical parameters in the Swiss Alps. The querying paradigm relies upon the issue of interactively constructing a semantically valid query rather than formulating one in a database specific query language
and for a particular implementation model. In order to achieve this goal, the query answering machine has to make inferences concerning the properties and value domains, as well as data analysis operations, which are semantically valid within particular contexts. These inferences take place when the intended query is being constructed interactively on a Web-based blackboard. A graph-based display presentation formalism is used with elements including natural language terms, measurement units, statistical quantifiers and/or specific value domains.
A meta-data database is used to organise and provide the elements of the graph each time the graph, and consequently the intended query, is expanded or further refined. Finally, the displayed graph is transformed into elements of the implementation model from which, in turn, SQL statements and/or sequences of statistical operations are created.