Quantitative content analysis is a cornerstone of communication research, yet it often involves time-consuming and labor-intensive manual labeling, or requires technical expertise for automation. This article introduces LabelGenius, a Python library that reduces activation energy for researchers beginning computational labeling projects by providing validated workflows and systematic decision frameworks. Leveraging GPT and CLIP architectures, LabelGenius supports zero-shot, few-shot, and fine-tuned classification of text, images, and text-image pair content - minimizing the need for manual effort. Validated through two case studies on (1) text-image pair news labeling and (2) multiple-theme classification of immigration news headlines, LabelGenius shows high accuracy and scalability, reducing technical barriers and enhancing labeling efficiency.
Huang et al. (Mon,) studied this question.