The author’s abstract states, "...many disciplines regularly face the need to analyze large volumes of numerical data..... Nevertheless, use of experience and specialized knowledge are required in numerical processing, either to understand output or to adjust input parameters." The objective of the work reported on is to apply expert systems to both interpret symbolic features of processed measurement data, "and to adjust input parameters according to automatic expert diagnosis."

The LITHO project utilized this approach. The project was concerned with interpreting physical property measurements, which are then represented by plotter curves called logs. The approach was to merge three analysis methods: expert systems (production rules) for the symbolic reasoning; pattern recognition, which scans the logs for various features; and a clustering technique or decision process. The function of the latter is to refine the results and detect inconsistencies.

This paper addresses an important problem and provides an example of an approach to its solution. The expert system knowledge base is stored in about 500 production rules which include conference factors. The results are reported as "satisfying," but problems remained. In particular, there is a mismatch between the expert system and the cluttering/refinement stage. This problem is discusses to some extent in the paper: the suggested direction of improvement is to integrate the cluster (also termed the :Idecision module) into the expert system itself.