Qualitative data analysis involves coding to identify themes and patterns, but it can be time-consuming, especially with large datasets. The study examines how LLMs can assist in coding tasks of varying complexity, highlighting challenges for both human coders and LLMs. Factors influencing coding complexity are discussed, along with the usefulness and limitations of incorporating LLMs in qualitative research.
Large Language Models (LLMs), such as GPT-4, show promise in speeding up the coding process but require careful evaluation. The study compares human coders' performance with LLMs in different coding tasks, revealing insights into agreement levels and challenges faced by both parties. Methodological considerations, model choices, and ethical implications of using LLMs in qualitative research are also addressed.
Til et andet sprog
fra kildeindhold
arxiv.org
Vigtigste indsigter udtrukket fra
by Elisabeth Ki... kl. arxiv.org 03-12-2024
https://arxiv.org/pdf/2403.06607.pdfDybere Forespørgsler