Using LLM-Based Approaches to Enhance and Automate Topic Labeling
Trishia Khandelwal · 2025
Read Using LLM-Based Approaches to Enhance and Automate Topic Labeling by Trishia Khandelwal (2025) 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-2502-18469.txt.
Description
Trishia Khandelwal explores in 2025 how large language models can label topics found by a topic model. Such models pull themes from big text piles, yet they hand back bare word lists that people must name by hand. That naming is slow and can drift from reader to reader.
Khandelwal first ran a known topic tool, then fed varied picks of keywords and document digests to a language model for naming. Each pick stressed a different aim, such as the dominant theme or breadth across the topic. A new score gauges how well a label stands for all documents in its topic, filling a gap in how labels are judged. Better labels mean faster, steadier reads of what a big text pile is about.
Catalog record
| CREATOR | Trishia Khandelwal |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 16 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-18469