An Approach for Measuring Information Content of Textual Engineering Requirements Using a Form-Neutral Representation

Abstract

This paper demonstrates a generic approach for measuring the information content of artifacts produced and used in early stages of mechanical design. Engineering design requirements are selected for information content analysis for illustration. In this method, requirements in natural language are translated to an Entity-Relation-Attribute-Value (ERAV) model composed of well-defined elements. A protocol for this translation is proposed and validated. Four different metrics, based on raw element count, count weighted by arbitrary ordinal scale, count weighted by node cardinality, and Shannon’s entropy are then applied to the ERAV model for measuring information content. The method proposed is generic enough to be applied to most design documents that use natural language as the knowledge representation formalism.