Navigating the Prompt Space: Improving LLM Classification of Social Science Texts Through Prompt Engineering

Erkan Gunes 1 , Christoffer Florczak 2 , and Tevfik Murat Yildirim 3 · 2026

Read Navigating the Prompt Space: Improving LLM Classification of Social Science Texts Through Prompt Engineering by Erkan Gunes 1 , Christoffer Florczak 2 , and Tevfik Murat Yildirim 3 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-25422.txt.

Description

Gunes, Florczak, and Yildirim tune prompts to make language models better coders of social science text. They vary three prompt parts: label glossaries, guiding nudges, and worked examples. A small dose of extra context gives the biggest jump in accuracy. More context after that adds little, and at times it hurts. Gains also swing by model, task, and batch size. The lesson is plain: test each coding job on its own, and do not trust blanket prompt rules.

Catalog record

CREATORErkan Gunes 1 , Christoffer Florczak 2 , and Tevfik Murat Yildirim 3
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE123480
LICENSEhttp://creativecommons.org/licenses/by-sa/4.0/
RIGHTSCC BY SA
SOURCEarXiv

Source: https://arxiv.org/abs/2603.25422

Browse the full library · About the Tower of Babel