A linguistic knowledge analysis approach for industrial engineers

Industrial Engineers use various kinds of knowledge analysis approaches to understand, analyze, and represent an environment or a task of interest. The suitability of a given approach is determined by the requirements of the task and the alternative approaches presented or known to the engineer. This research proposes to introduce Industrial Engineers to a knowledge analysis and representation approach of sentence structure that is based on modern generative linguistics. The significance of the proposed approach stems from several reasons. First, generative linguistics provides a way to directly and comprehensively encode large segments of the human language into a concise and finite formalism. This direct encoding of the human language enriches and deepens the analyst's semantic sensitivity. Second, as Industrial Engineers attempt to probe into and employ various knowledge analysis approaches that possess human features (e.g. Object-Oriented and Expert Systems), it seems imperative to also look into generative linguistics; a discipline that extensively examines the features of one of the human mind faculties, namely the human language. Finally, the proposed knowledge analysis approach opens the possibility for developing a powerful representation that is suitable for both human and computer processing. Numerical Control part programming, one of the selected domains of application, will be used to illustrate that possibility.;The challenge and approach of this research was to take the work of the linguist and cognitive scientist, Ray Jackendoff, augment it with insights from the syntactic subfield of generative linguistics, and represent it in a way that is understandable to the Industrial Engineer. This tailoring and packaging process was then followed with an implementation phase. Three applications were selected to test the proposed approach. The first application demonstrated how the proposed linguistic semantic model can be used in a principled manner to derive and incorporate other models such as IDEFO and Conceptual Graphs. In the second application, the linguistic approach was shown to offer a single semantic pattern that governs all basic Numerical Control part programming sentences and allow a homogeneous classification of the vocabulary in that domain. Finally, the linguistic analysis was used to analyze unconstrained questionnaire responses in order to support a real-life consultation task.