Large-Scale Embedding Learning in Heterogeneous Event Data

Gui, Huan; Liu, Jialu; Tao, Fangbo; Jiang, Meng; Norick, Brandon; Han, Jiawei

doi:10.1109/icdm.2016.0111

Cited by 75 publications

(74 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This observation is in line with previous studies [7], and can be explained by the heterogeneity of node types in HINs. The nodes of different types in HINs have different properties, such as degrees distribution.…”

Section: Methodssupporting

confidence: 93%

“…Chang et al propose to embed HIN with additional node features via deep architectures [2], which does not suit for typical HINs consisting of only typed nodes and edges. Gui et al devise an HIN embedding algorithm to model a special type of HINs with hyper-edges, which does not apply to general HINs [7]. More recently, an HIN embedding algorithm is proposed, which transcribes semantics in HINs by meta-paths [4].…”

Section: Related Workmentioning

confidence: 99%

“…To design the embedding algorithm for one aspect, we extend the skip-gram model [9] in an approach inspired by existing network embedding studies [7, 21, 22]. We note that ASPEM is a flexible framework that can be directly integrated with other homogeneous network embedding methods [6, 12, 14, 24], other than the adopted skip-gram–based approach, while still enjoying the benefits of modeling aspects in HINs.…”

Section: The Aspem Frameworkmentioning

confidence: 99%

“…We hence propose to study the problem of learning embedding in HINs with an emphasis on leveraging the rich and intrinsic type information. There are multiple attempts in studying HIN embedding or tackling specific application tasks using HIN embedding [2, 3, 7, 21]. Though these studies formulate the problem differently with respective optimization objectives, they share a similar underlining philosophy: using a unified objective function to embed all the nodes into one low-dimensional space.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

AspEm: Embedding Learning by Aspects in Heterogeneous Information Networks

Shi¹,

Gui²,

Zhu³

et al. 2018

Proceedings of the 2018 SIAM International Conference on Data Mining

Self Cite

View full text Add to dashboard Cite

Heterogeneous information networks (HINs) are ubiquitous in real-world applications. Due to the heterogeneity in HINs, the typed edges may not fully align with each other. In order to capture the semantic subtlety, we propose the concept of aspects with each aspect being a unit representing one underlying semantic facet. Meanwhile, network embedding has emerged as a powerful method for learning network representation, where the learned embedding can be used as features in various downstream applications. Therefore, we are motivated to propose a novel embedding learning framework—ASPEM—to preserve the semantic information in HINs based on multiple aspects. Instead of preserving information of the network in one semantic space, ASPEM encapsulates information regarding each aspect individually. In order to select aspects for embedding purpose, we further devise a solution for ASPEM based on dataset-wide statistics. To corroborate the efficacy of ASPEM, we conducted experiments on two real-words datasets with two types of applications—classification and link prediction. Experiment results demonstrate that ASPEM can outperform baseline network embedding learning methods by considering multiple aspects, where the aspects can be selected from the given HIN in an unsupervised manner.

show abstract

Section: Methodssupporting

confidence: 93%

Section: Related Workmentioning

confidence: 99%

Section: The Aspem Frameworkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

AspEm: Embedding Learning by Aspects in Heterogeneous Information Networks

Shi¹,

Gui²,

Zhu³

et al. 2018

Proceedings of the 2018 SIAM International Conference on Data Mining

Self Cite

View full text Add to dashboard Cite

show abstract

“…The proximity of the interaction between the two words can be modeled as the conditional probability of predicting the observed target word given the context word [7], where the conditional probability is estimated via softmax function, with low-dimensional vectors of words as parameters. This model has also been generalized to network data, such as [8], [11], [23]. …”

Section: Introductionmentioning

confidence: 99%

Embedding Learning with Events in Heterogeneous Information Networks

Gui

Liu

Tao

et al. 2017

IEEE Trans. Knowl. Data Eng.

Self Cite

View full text Add to dashboard Cite

In real-world applications, objects of multiple types are interconnected, forming Heterogeneous Information Networks. In such heterogeneous information networks, we make the key observation that many interactions happen due to some event and the objects in each event form a complete semantic unit. By taking advantage of such a property, we propose a generic framework called HyperEdge-Based Embedding (Hebe) to learn object embeddings with events in heterogeneous information networks, where a hyperedge encompasses the objects participating in one event. The Hebe framework models the proximity among objects in each event with two methods: (1) predicting a target object given other participating objects in the event, and (2) predicting if the event can be observed given all the participating objects. Since each hyperedge encapsulates more information of a given event, Hebe is robust to data sparseness and noise. In addition, Hebe is scalable when the data size spirals. Extensive experiments on large-scale real-world datasets show the efficacy and robustness of the proposed framework.

show abstract