Analysis of content topics, user engagement and library factors in public library social media based on text mining

Purpose: The purpose of this paper is to explore topics of Facebook posts created by public libraries using the bi-term topic model, and examine the relationships between types of topics and user engagement. The authors further investigated the effects of three library factors, namely, staff size, budget and urbanization degrees, on Facebook content and user engagement based on multilevel generalized linear modeling. Design/methodology/approach: This study suggested a novel method, a combination of the bi-term topic modeling and MGLM, to enhance the understanding of social media in the context of public libraries. Findings: The findings revealed that posts related to community events, awards and photos were likely to receive more likes and shares, whereas posts about summer reading programs received relatively more comments. In addition, the authors found that a larger staff size and the inclusion of multimedia had positive impacts on user engagement. Originality/value: This study analyzed the content of public library-generated social media based on text mining. Then, the authors examined the effects of contextual library-level factors on social media practice in public libraries. Based on empirical findings, the study suggested a range of practical implications for effective use of social media in public libraries.