Team Design Cognition And Computation Based On Language Data Analysis

Innovation through design thinking and creative skills are considered core attributes for successful enterprises however the design process comprises many complex cognitive activities. Understanding how to capture and analyze the design cognition process has important meaning in the understanding of the design process and may enhance design capability. Conducting research on design cognition to parse the complex design process and explore the abstract thinking pattern has vital academic and application value.Deep mining on the protocol record data is conducted to analyze team design cognition from the perspective of information processing. The analysis is based on the multi-dimensional model which is constructed from the view of design process development, low-level information storage and semantic concept evolution. By selecting linkography as the representation form of the design process over time, dividing the concept nodes and judging the linking relationships, a structured expression of design inference process is realized. The deterministic information theory is also introduced to study the complexity characteristics; A matrix is used to store the result of language data analysis in order to support the quantitative computation, which in result achieves transformation from unstructured language data to operable structured data; meanwhile, the semantic network is also applied to explore the concept evolution process which can identify creative concepts and search concept evolution path.Effectiveness of the multi-dimensional model is validated through experiment. Analysis based on the model includes data visualization in the form of linkography and network, and position computation and context substitution of important concepts. The idea is to combine both qualitative and quantitative analysis to study design cognition. An auxiliary prototype system is also developed to improve the research efficiency on design cognition and computation and provide sufficient quantitative information by means of packaging bottom layer calculations.