Reproduce “Multi-Granularity Prompts for Topic Shift Detection in Dialogue” 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

Jiangyi Lin, Yaxin Fan, Xiaomin Chu, Peifeng Li, Qiaoming Zhu

The authors detect dialogue topic shifts when the current response is available by combining a shift classifier with generated clues at label, topic, and turn levels. On the Chinese CNTD context–response evaluation, their model achieves 84.7 Macro-F1, compared with 82.4 for BERT+BiLSTM, with the comparison marked significant at p<0.01. This result applies to response-known detection, not prediction before a reply arrives.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.