Mining the evolutionary process of knowledge through multiple relationships between keywords

Knowledge evolution offers a road map for understanding knowledge creation, knowledge transfer, and performance in everyday work. Understanding the knowledge evolution of a research field is crucial for researchers, policymakers, and stakeholders. Further, paper keywords are considered efficient knowledge components to depict the knowledge structure of a research field by examining relationships between keywords. However, multiple relationships between keywords provided by papers are rarely used to explore knowledge evolution. Three relationships were applied: a direct co-occurrence relationship, indirect relationship by keyword pair citation, and same author trace, providing temporal and sequential knowledge evolution. The direct co-occurrence relationship is constructed by keyword co-occurrence pair and acts as the temporal structure of knowledge pairs. The indirect relationship is constructed by a keyword pair-based citation relationship, meaning the citation relationship between keyword co-occurrence pairs, acting as the sequential structure of knowledge pairs. Additionally, the same author trace represents an indirect relationship that a keyword pair provided by the same author in a different paper. Thus, knowledge evolution could be mined quantitatively from a different perspective. Therefore, we present an empirical study of the informetrics field with five evolution stages: knowledge generation, growth, obsolescence, transfer, and intergrowth. The results indicate that knowledge evolution is not a continuous trend but alternating growth and obsolescence. During evolution, knowledge pairs stimulate each other’s growth, and some knowledge pairs transfer to others, demonstrating a small step toward knowledge change. According to the indirect keyword relationship paired with the same author trace, creators and followers of knowledge evolution are different.