Structural dynamics of keyword networks: Liquid crystal display and plasma display panel cases

This study focuses on understanding scientific evolution by using keyword co-occurrence networks, where keywords appearing in a scientific article are linked with a weight equal to the number of co-occurrences. To characterize structural changes of the network, we examine distributions of sums of weights by node over time. In particular, a change of power-law behavior is utilized to explore scientific evolution, such as emerging scientific paradigms and advancing normal science. As an illustration of the method used, the development of Liquid Crystal Displays (LCDs) and Plasma Display Panels (PDPs) is tracked. We detect two-tiered power-law distributions in the initial stage of scientific growth in both technologies due to differences in research intensity between two groups. The groups of keywords more likely to attract researchers’ interest than others are incrementally developed until the mid-2000s to overtake those prior. Finally, we can capture a merging point of the dichotomous structure of PDPs but LCDs maintain the structural separation throughout the adjustment area. We expect that this structural investigation of keyword co-occurrence networks provides an indicator to diagnose the research evolution in that field.