A Structure-aware Semantic-Topic Interactive Framework for Generative Joint Entity and Relation Extraction

Jiao Luo

doi:10.1016/j.patrec.2026.09.025

A Structure-aware Semantic-Topic Interactive Framework for Generative Joint Entity and Relation Extraction

2026-09-20

0

PRE

AI

DOI：10.1016/j.patrec.2026.09.025

Original

Original request for help

Save

Abstract

En 中文

• A novel SSTIF with synergistic interactions between topic- and token-level semantics. • A bi-level topic control improves decoding via topic priors at attention and vocabulary levels. • Role encoding models triple structure and mitigates cross-triple dependencies.

Keywords:

Joint Entity and Relation Extraction

Generative Information Extraction

Topic Modeling

Topic-Conditioned Language Modeling

A Structure-aware Semantic-Topic Interactive Framework for Generative Joint Entity and Relation Extraction

Abstract

Journal

Organization

Cited Papers