Reproduce “Dynamic Embedded Topic Models: properties and recommendations based on diverse corpora” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Elisabeth Fittschen, Bella Xia, Leib Celnik, Paul Dilley, Tom Lippincott
This research provides a systematic evaluation of the Dynamic Embedded Topic Model (DETM) across diverse text corpora, offering practical guidance for its application. The study found that DETM performs better with larger topic counts and is robust to time window choices and vocabulary sizes, making it particularly useful for analyzing conceptual evolution in historical texts.
Key insights
- DETM performs better with a larger number of topics (up to 160 tested), suggesting its embedding-based approach effectively utilizes extensive topic inventories.
- The model demonstrates robustness to various time window granularities, performing well with both simple and more granular temporal divisions.
- DETM maintains stable performance across different vocabulary sizes, indicating its embedding-based approach effectively handles large vocabularies, though memory can be a practical limit.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.