A Two-Stage Framework for Directed Hypergraph Link Prediction
Hypergraphs, as a special type of graph, can be leveraged to better model relationships among multiple entities. In this article, we focus on the task of hyperlink prediction in directed hypergraphs, which finds a wide spectrum of applications in knowledge graphs, chem-informatics, bio-informatics,...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-07-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/10/14/2372 |
| _version_ | 1797417219988127744 |
|---|---|
| author | Guanchen Xiao Jinzhi Liao Zhen Tan Xiaonan Zhang Xiang Zhao |
| author_facet | Guanchen Xiao Jinzhi Liao Zhen Tan Xiaonan Zhang Xiang Zhao |
| author_sort | Guanchen Xiao |
| collection | DOAJ |
| description | Hypergraphs, as a special type of graph, can be leveraged to better model relationships among multiple entities. In this article, we focus on the task of hyperlink prediction in directed hypergraphs, which finds a wide spectrum of applications in knowledge graphs, chem-informatics, bio-informatics, etc. Existing methods handling the task overlook the order constraints of the hyperlink’s direction and fail to exploit features of all entities covered by a hyperlink. To make up for the deficiency, we present a performant pipelined model, i.e., a two-stage framework for directed hyperlink prediction method (TF-DHP), which equally considers the entity’s contribution to the form of hyperlinks, and emphasizes not only the fixed order between two parts but also the randomness inside each part. The TF-DHP incorporates two tailored modules: a Tucker decomposition-based module for hyperlink prediction, and a BiLSTM-based module for direction inference. Extensive experiments on benchmarks—WikiPeople, JF17K, and ReVerb15K—demonstrate the effectiveness and universality of our TF-DHP model, leading to state-of-the-art performance. |
| first_indexed | 2024-03-09T06:15:39Z |
| format | Article |
| id | doaj.art-6c4cfd4ec0df40d18be2abbde66cb1ca |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-09T06:15:39Z |
| publishDate | 2022-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-6c4cfd4ec0df40d18be2abbde66cb1ca2023-12-03T11:53:15ZengMDPI AGMathematics2227-73902022-07-011014237210.3390/math10142372A Two-Stage Framework for Directed Hypergraph Link PredictionGuanchen Xiao0Jinzhi Liao1Zhen Tan2Xiaonan Zhang3Xiang Zhao4Science and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha 410073, ChinaScience and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha 410073, ChinaScience and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha 410073, ChinaHarbin Flight Academy, Harbin 150000, ChinaScience and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha 410073, ChinaHypergraphs, as a special type of graph, can be leveraged to better model relationships among multiple entities. In this article, we focus on the task of hyperlink prediction in directed hypergraphs, which finds a wide spectrum of applications in knowledge graphs, chem-informatics, bio-informatics, etc. Existing methods handling the task overlook the order constraints of the hyperlink’s direction and fail to exploit features of all entities covered by a hyperlink. To make up for the deficiency, we present a performant pipelined model, i.e., a two-stage framework for directed hyperlink prediction method (TF-DHP), which equally considers the entity’s contribution to the form of hyperlinks, and emphasizes not only the fixed order between two parts but also the randomness inside each part. The TF-DHP incorporates two tailored modules: a Tucker decomposition-based module for hyperlink prediction, and a BiLSTM-based module for direction inference. Extensive experiments on benchmarks—WikiPeople, JF17K, and ReVerb15K—demonstrate the effectiveness and universality of our TF-DHP model, leading to state-of-the-art performance.https://www.mdpi.com/2227-7390/10/14/2372hyperlink predictionhypergraphTucker decomposition |
| spellingShingle | Guanchen Xiao Jinzhi Liao Zhen Tan Xiaonan Zhang Xiang Zhao A Two-Stage Framework for Directed Hypergraph Link Prediction Mathematics hyperlink prediction hypergraph Tucker decomposition |
| title | A Two-Stage Framework for Directed Hypergraph Link Prediction |
| title_full | A Two-Stage Framework for Directed Hypergraph Link Prediction |
| title_fullStr | A Two-Stage Framework for Directed Hypergraph Link Prediction |
| title_full_unstemmed | A Two-Stage Framework for Directed Hypergraph Link Prediction |
| title_short | A Two-Stage Framework for Directed Hypergraph Link Prediction |
| title_sort | two stage framework for directed hypergraph link prediction |
| topic | hyperlink prediction hypergraph Tucker decomposition |
| url | https://www.mdpi.com/2227-7390/10/14/2372 |
| work_keys_str_mv | AT guanchenxiao atwostageframeworkfordirectedhypergraphlinkprediction AT jinzhiliao atwostageframeworkfordirectedhypergraphlinkprediction AT zhentan atwostageframeworkfordirectedhypergraphlinkprediction AT xiaonanzhang atwostageframeworkfordirectedhypergraphlinkprediction AT xiangzhao atwostageframeworkfordirectedhypergraphlinkprediction AT guanchenxiao twostageframeworkfordirectedhypergraphlinkprediction AT jinzhiliao twostageframeworkfordirectedhypergraphlinkprediction AT zhentan twostageframeworkfordirectedhypergraphlinkprediction AT xiaonanzhang twostageframeworkfordirectedhypergraphlinkprediction AT xiangzhao twostageframeworkfordirectedhypergraphlinkprediction |