Keyword standardization and restructuring: the impact on analysing network-based science maps in innovation management research

Content analysis with the use of keywords experiences a take-off period in science mapping. Within the family of co-word analyses, analyzing the keyword content is commonly preceded by computer-assisted preprocessing, which may leave substantial noise and bias in the structure of the network set up. Despite these flaws, only few articles have attempted to go beyond conventional keyword standardization steps, although leveraging expert knowledge holds the promise to reduce the tradeoff between interpretability and representativeness in scaled bibliometric studies. We propose systematic manual preprocessing, an algorithmic keyword standardization and restructuring (KSR) procedure, and the paper is a validation study of the method. The innovation management (IM) disciplinary area is used to demonstrate the extent to which the quality and interpretability of bibliometric networks change and improve if in-depth keyword standardization and restructuring is implemented. For the demonstration, two networks of more than 5000 articles were set up and analyzed using identical steps, keyword preprocessing being the only difference. The impact of the KSR procedure on the clusterings is considerable and interpretation is greatly affected. Recommendations have been compiled for researchers, who would like to build keyword-based science maps to analyze content.