Event Ontology Expansion via LLM-Based Conceptualization

2026-06-18

AI Summary

Existing event ontology expansion methods rely on instance-level semantics, struggling to capture the hierarchical conceptual structure of event types, which leads to unstable clustering and unreliable ontology extension. This work proposes ConceptE, a novel framework that introduces, for the first time, a large language model–driven conceptualization mechanism to generate unified concept names and natural language descriptions for event triggers, integrating trigger information to construct concept-enhanced representations. By elevating event representations from the instance level to the conceptual level, ConceptE enables ontology-consistent type naming, stable clustering, and hierarchical expansion. Evaluated on ACE, ERE, and MAVEN datasets, ConceptE significantly outperforms existing approaches, achieving up to a 12.37% improvement in BCubed-F1 for event clustering and a 6.48% gain in Taxo_F1 for hierarchical ontology extension.

Abstract

Event ontology expansion aims to discover emerging event types from data and extend them to appropriate positions in the existing event ontology.. Existing methods typically cluster contextualized trigger representations and attach induced clusters to the ontology based on instance-level similarity. However, ontology expansion requires concept-level semantics that characterize event types, whereas contextualized trigger representations often conflate these semantics with surface contextual variation, leading to unstable clustering and unreliable hierarchy expansion. To address this issue, we propose ConceptE, a conceptualization-enhanced framework for event ontology expansion. ConceptE first derives concept-level semantics by prompting an LLM with the sentence and event trigger, producing a concise concept name and a natural-language description. It then jointly encodes these semantics with trigger information to build concept-enhanced representations aligned with ontology-level reasoning. This representation design supports more coherent event clustering, more reliable hierarchy expansion, and ontology-consistent type naming. Experiments on ACE, ERE, and MAVEN demonstrate that ConceptE consistently outperforms state-of-the-art approaches across all subtasks of event ontology expansion. In particular, it achieves improvements of up to 12.37\% in BCubed-F1 for event clustering and 6.48\% in Taxo\_F1 for hierarchy expansion, demonstrating the effectiveness of the proposed ConceptE method.

Problem

• event ontology expansion
• concept-level semantics
• contextualized trigger representations
• hierarchy expansion
• event clustering

Innovation

• event ontology expansion
• LLM-based conceptualization
• concept-level semantics
• hierarchy expansion
• event clustering