Overview
With Jamal Al-Khadhuri, David Passmore, Rose Baker, and John Turner, I presented a text-mining study of the four refereed journals of the Academy of Human Resource Development (ADHR, HRDI, HRDQ, and HRDR). The aim was to demonstrate a topic modeling procedure on the whole body of peer-reviewed articles published from 1990 to 2015.
From 2,824 items we removed 919 editorials, book reviews, forum pieces, invited reactions, and front and back matter, leaving 1,905 peer-reviewed articles. Titles, keywords, and abstracts were entered together with publication year and journal name as covariates, pre-processed, and modeled in R with the stm package.
Key points
- Candidate solutions from 10 to 100 topics were simulated and compared on an estimation sample and a hold-out sample.
- A twenty-topic solution was examined first; semantic coherence and exclusivity pointed to a final five-topic solution.
- The five topics were labeled definition of HRD, globalization of HRD, analytics and methods, workplace performance, and HRD role reflection.
- Topic prevalence was compared across publication years and across the four journals.
- We argued that unsupervised and supervised topic modeling can help researchers keep up with a fast-growing literature.
Event
2017 Academy of Human Resource Development International Research Conference in the Americas, San Antonio, Texas, March 2017.
Citation
Chae, C., Al-Khadhuri, J., Passmore, D., Baker, R. M., & Turner, J. (2017, March). Structural model of topics in Academy of Human Resource Development journals, 1990-2015. Paper presented at the 2017 Academy of Human Resource Development International Research Conference in the Americas, San Antonio, Texas. doi:10.6084/m9.figshare.4725577.v1