Enhancing LLMs in Predictive Political QA with Semi-Structured Data

The paper introduces PSL, a dual‑view framework that enhances large language models for predictive political question answering by leveraging semi‑structured political records. PSL extracts stance signals from actor records in a semantic view and learns structure‑aware actor representations from an interaction graph in a vector view. Experiments on three real‑world datasets show that PSL consistently outperforms baseline methods, with ablation studies confirming the complementary benefits of stance and structure signals.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• llms
• nlp