Understanding Design Concept Identification

In the design literature, the term design concept is often used de facto, or with only a brief definition provided. Despite the cursory definition for concept, the design process rests heavily on concepts, e.g., brainstorming and generating multiple design concepts, and subsequently identifying design concepts for concept selection, evaluation and development, etc. Concepts and concept formation are of particular interest in psychology, as concepts play a central role in human cognition. Concepts and concept identification are also of interest in other fields such as archaeology, bioinformatics and education. In this paper, we explore the process of design concept identification and address the issue of identifying design concepts in free-form text. Our exploratory experiment uses text transcripts of verbal concept generation sessions to first investigate agreeability between human concept identifiers. Next, we perform a language analysis on the transcripts to uncover language patterns that may differentiate between text segments containing concepts and text segments not containing concepts. Our results show that humans are adept at identifying and agreeing upon concepts (average agreeability >0.70), and that there are significant language differences that may distinguish concept segments from non-concept segments (i.e., non-concept segments have significantly more verbs and borderline significantly more self-references than concept segments). In general, automated concept identification may lead to better integration of early conceptual design with more detailed and computable downstream processes, resulting in a unified design workflow.