Grid search-based LDA optimization for agricultural topic discovery and intertopic distance map visualization
Abstract
Agricultural research has generated a substantial volume of scientific publications, yet this corpus remains underutilized for systematically identifying major research topics and thematic structures. This study aims to identify dominant topics in agricultural publications from 2012 to 2023 using latent Dirichlet allocation (LDA) with systematic hyperparameter tuning and intertopic distance map visualization. An exploratory quantitative approach was applied to 75,031 publications retrieved from Scopus, ProQuest, and Web of Science (WoS). The analysis involved text preprocessing, phrase modeling, dictionary and corpus construction, and LDA topic modeling. An exhaustive grid search over a discretized parameter space identified a 17 topic model with alpha=0.1, beta=0.70, and a coherence score of 0.5075, followed by interactive visualization using pyLDAvis. The results revealed six major thematic groups: agricultural practices and food systems; data analysis applications; plant production and growth factors; environmental and land-use changes; effects and sample analysis; and plant genetics under environmental pressures. The findings demonstrate that LDA combined with systematic parameter tuning can reveal latent thematic structures and clarify relationships among agricultural research topics. The study provides a data-driven basis for exploratory mapping of research directions, while temporal trend evolution and direct policy implications require further longitudinal analysis.
Keywords
Agricultural research topics; Big scholarly data; Grid search; Intertopic distance map; Latent Dirichlet allocation
Full Text:
PDFDOI: https://doi.org/10.11591/eei.v15i6.14194
Refbacks
- There are currently no refbacks.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Bulletin of Electrical Engineering and Informatics (BEEI)
ISSN: 2089-3191
,
e-ISSN: 2302-9285
This journal is published by the
Institute of Advanced Engineering and Science (IAES)
.