C-NARS/P: An Open-Source Tool for Classification of Narratives in Survey Data Using Pre-Trained Language Models

The code set is available at https://github.com/TEEDLab/CNARSP ; This report presents an open-source tool (C-NARS/P) created to classify narratives in survey data using pre-trained language models. Unlike conventional machine learning methods, pre-trained language models such as BERT can utilize the context of words in a text to make accurate predictions. In this report, technical details of functions and parameter setups for implementing pre-trained language models for classification are provided with screenshots of the code file to guide users in the validation and improvement of the implementation. C-NARS/P can be modified for a wide range of text classification tasks with ease. ; National Science Foundation Award FW-HTF-P 2128416. ; http://deepblue.lib.umich.edu/bitstream/2027.42/178287/4/Abramowitz&Kim&Kim_technical report_CNARSP.pdf ; 4ae71d2a-01c0-4084-84c3-c32ce960e81c ; Description of project report_CNARSP_abramowitz&kim&kim.pdf : Technical Report ; SELF