@inproceedings{li-etal-2026-shape,
title = "{SHAPE}: Structural Hypergraph Augmentation with Personalized Event Memory for Temporal Knowledge Graph Reasoning",
author = "Li, Qing and
Lv, Jialin and
Song, Jingjing and
Fei, Duan and
Guo, Zhongwen",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1412/",
pages = "32251--32265",
ISBN = "979-8-89176-459-0",
abstract = "Temporal knowledge graph (TKG) reasoning aims to predict future events from historical facts. A key challenge for graph-based TKG models is to simultaneously capture higher-order structural dependencies and identify query-relevant temporal evidence, yet existing methods are largely limited to pairwise interactions or query-agnostic history aggregation. To address this, we propose SHAPE (Structural Hypergraph Augmentation with Personalized Event Memory), a plug-and-play enhancement framework for graph-based TKG models that strengthens the backbone at both structural and temporal levels without modifying its architecture. Specifically, Dual Hypergraph Pre-Enhancement(DHPE) constructs learnable entity and relation prototype hypergraphs with cross-view information exchange to inject higher-order structural priors before temporal encoding, while Query-Denoised Event Memory(QDEM) retrieves query-relevant recent events and converts them into subject enhancement and candidate-level scoring bias via structure-aware expert gating. Extensive experiments on three benchmarks with two heterogeneous graph backbones verify the effectiveness of SHAPE."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-etal-2026-shape">
<titleInfo>
<title>SHAPE: Structural Hypergraph Augmentation with Personalized Event Memory for Temporal Knowledge Graph Reasoning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Qing</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jialin</namePart>
<namePart type="family">Lv</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingjing</namePart>
<namePart type="family">Song</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Duan</namePart>
<namePart type="family">Fei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhongwen</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Temporal knowledge graph (TKG) reasoning aims to predict future events from historical facts. A key challenge for graph-based TKG models is to simultaneously capture higher-order structural dependencies and identify query-relevant temporal evidence, yet existing methods are largely limited to pairwise interactions or query-agnostic history aggregation. To address this, we propose SHAPE (Structural Hypergraph Augmentation with Personalized Event Memory), a plug-and-play enhancement framework for graph-based TKG models that strengthens the backbone at both structural and temporal levels without modifying its architecture. Specifically, Dual Hypergraph Pre-Enhancement(DHPE) constructs learnable entity and relation prototype hypergraphs with cross-view information exchange to inject higher-order structural priors before temporal encoding, while Query-Denoised Event Memory(QDEM) retrieves query-relevant recent events and converts them into subject enhancement and candidate-level scoring bias via structure-aware expert gating. Extensive experiments on three benchmarks with two heterogeneous graph backbones verify the effectiveness of SHAPE.</abstract>
<identifier type="citekey">li-etal-2026-shape</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1412/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>32251</start>
<end>32265</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SHAPE: Structural Hypergraph Augmentation with Personalized Event Memory for Temporal Knowledge Graph Reasoning
%A Li, Qing
%A Lv, Jialin
%A Song, Jingjing
%A Fei, Duan
%A Guo, Zhongwen
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F li-etal-2026-shape
%X Temporal knowledge graph (TKG) reasoning aims to predict future events from historical facts. A key challenge for graph-based TKG models is to simultaneously capture higher-order structural dependencies and identify query-relevant temporal evidence, yet existing methods are largely limited to pairwise interactions or query-agnostic history aggregation. To address this, we propose SHAPE (Structural Hypergraph Augmentation with Personalized Event Memory), a plug-and-play enhancement framework for graph-based TKG models that strengthens the backbone at both structural and temporal levels without modifying its architecture. Specifically, Dual Hypergraph Pre-Enhancement(DHPE) constructs learnable entity and relation prototype hypergraphs with cross-view information exchange to inject higher-order structural priors before temporal encoding, while Query-Denoised Event Memory(QDEM) retrieves query-relevant recent events and converts them into subject enhancement and candidate-level scoring bias via structure-aware expert gating. Extensive experiments on three benchmarks with two heterogeneous graph backbones verify the effectiveness of SHAPE.
%U https://aclanthology.org/2026.emnlp-main.1412/
%P 32251-32265
Markdown (Informal)
[SHAPE: Structural Hypergraph Augmentation with Personalized Event Memory for Temporal Knowledge Graph Reasoning](https://aclanthology.org/2026.emnlp-main.1412/) (Li et al., EMNLP 2026)
ACL