Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Logic Induced High-Order Reasoning Network for Event-Event Relation Extraction

Peixin Huang, Xiang Zhao, Minghao Hu, Zhen Tan, Weidong Xiao

TL;DR

LogicERE tackles event-event relation extraction by replacing heuristic cross-sentence edges with a logic constraint induced graph (LCG) that encodes coreference, symmetry, and conjunction constraints for high-order reasoning. It initializes event and event-pair embeddings and processes them with a Relational Graph Transformer, incorporating edge-type biases and a joint logic learning objective that converts constraints into differentiable losses $L_{sym}$ and $L_{conj}$, combined with the primary loss $L_1$ as $L = L_1 + \gamma_{sym} L_{sym} + \gamma_{conj} L_{conj}$. The framework unifies TRE and SRE through two node types (events and event-pairs) and three edge types (C_{ee}, C_{pp}, C_{ep}), enabling global coherence in event evolution graphs. Empirical results on MATRES, TCR, HiEve, and MAVEN-ERE demonstrate state-of-the-art TRE performance and significant gains for SRE under joint learning, validating the effectiveness of logic-guided high-order reasoning for cross-task event relation understanding. Overall, LogicERE offers a scalable, knowledge-free approach to coherent document-level event evolution with strong practical implications for information extraction and event forecasting.

Abstract

To understand a document with multiple events, event-event relation extraction (ERE) emerges as a crucial task, aiming to discern how natural events temporally or structurally associate with each other. To achieve this goal, our work addresses the problems of temporal event relation extraction (TRE) and subevent relation extraction (SRE). The latest methods for such problems have commonly built document-level event graphs for global reasoning across sentences. However, the edges between events are usually derived from external tools heuristically, which are not always reliable and may introduce noise. Moreover, they are not capable of preserving logical constraints among event relations, e.g., coreference constraint, symmetry constraint and conjunction constraint. These constraints guarantee coherence between different relation types,enabling the generation of a uniffed event evolution graph. In this work, we propose a novel method named LogicERE, which performs high-order event relation reasoning through modeling logic constraints. Speciffcally, different from conventional event graphs, we design a logic constraint induced graph (LCG) without any external tools. LCG involves event nodes where the interactions among them can model the coreference constraint, and event pairs nodes where the interactions among them can retain the symmetry constraint and conjunction constraint. Then we perform high-order reasoning on LCG with relational graph transformer to obtain enhanced event and event pair embeddings. Finally, we further incorporate logic constraint information via a joint logic learning module. Extensive experiments demonstrate the effectiveness of the proposed method with state-of-the-art performance on benchmark datasets.

Logic Induced High-Order Reasoning Network for Event-Event Relation Extraction

TL;DR

LogicERE tackles event-event relation extraction by replacing heuristic cross-sentence edges with a logic constraint induced graph (LCG) that encodes coreference, symmetry, and conjunction constraints for high-order reasoning. It initializes event and event-pair embeddings and processes them with a Relational Graph Transformer, incorporating edge-type biases and a joint logic learning objective that converts constraints into differentiable losses and , combined with the primary loss as . The framework unifies TRE and SRE through two node types (events and event-pairs) and three edge types (C_{ee}, C_{pp}, C_{ep}), enabling global coherence in event evolution graphs. Empirical results on MATRES, TCR, HiEve, and MAVEN-ERE demonstrate state-of-the-art TRE performance and significant gains for SRE under joint learning, validating the effectiveness of logic-guided high-order reasoning for cross-task event relation understanding. Overall, LogicERE offers a scalable, knowledge-free approach to coherent document-level event evolution with strong practical implications for information extraction and event forecasting.

Abstract

To understand a document with multiple events, event-event relation extraction (ERE) emerges as a crucial task, aiming to discern how natural events temporally or structurally associate with each other. To achieve this goal, our work addresses the problems of temporal event relation extraction (TRE) and subevent relation extraction (SRE). The latest methods for such problems have commonly built document-level event graphs for global reasoning across sentences. However, the edges between events are usually derived from external tools heuristically, which are not always reliable and may introduce noise. Moreover, they are not capable of preserving logical constraints among event relations, e.g., coreference constraint, symmetry constraint and conjunction constraint. These constraints guarantee coherence between different relation types,enabling the generation of a uniffed event evolution graph. In this work, we propose a novel method named LogicERE, which performs high-order event relation reasoning through modeling logic constraints. Speciffcally, different from conventional event graphs, we design a logic constraint induced graph (LCG) without any external tools. LCG involves event nodes where the interactions among them can model the coreference constraint, and event pairs nodes where the interactions among them can retain the symmetry constraint and conjunction constraint. Then we perform high-order reasoning on LCG with relational graph transformer to obtain enhanced event and event pair embeddings. Finally, we further incorporate logic constraint information via a joint logic learning module. Extensive experiments demonstrate the effectiveness of the proposed method with state-of-the-art performance on benchmark datasets.

Paper Structure

This paper contains 18 sections, 21 equations, 2 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Temporal relation extraction
  4. Subevent relation extraction
  5. Model
  6. Sequence Encoder
  7. Logic Constraint Induced Graph
  8. High-Order Reasoning Network on LCG
  9. Joint Logic Learning
  10. Experiments
  11. Datasets and Metrics
  12. Parameter Settings
  13. Baselines
  14. Comparison
  15. Ablation Study
  16. ...and 3 more sections

Figures (2)

  • Figure 1: An example of an event evolution graph described in the document.
  • Figure 2: The induction table for conjunctive constraints on temporal and subevent relations. The abbreviations PC, CP , CR, NR, BF, AF, EQ and VG denote PARENT-CHILD, CHILD-PARENT, COREF, NOREL, BEFORE, AFTER, EQUAL and VAGUE, respectively. Subevent relations are in black, and temporal relations are in blue. "$\backslash$" denotes no constraints.