Authors: Yu Shi, Xinwei He, Naijing Zhang, Carl Yang, Jiawei Han (Submitted on 28 Nov 2018) Abstract: As one type of complex networks widely-seen in real-world application, heterogeneous information networks (HINs) often encapsulate higher-order interactions that crucially reflect the complex nature among nodes and edges in real-world data. Beijing Xinwei Yachen Network Information Service Co., Ltd. (Xinwei Yachen), founded in February 2012 with a registered capital of 200 million RMB, is a company devoted to the construction and operation of Xinwei domestic wireless government affairs networks.
Abstract: Heterogeneous information networks (HINs) with rich semantics are ubiquitousin real-world applications. For a given HIN, many reasonable clustering resultswith distinct semantic meaning can simultaneously exist. User-guided clusteringis hence of great practical value for HINs where users provide labels to asmall portion of nodes. To cater to a broad spectrum of user guidance evidencedby different expected clustering results, carefully exploiting the signalsresiding in the data is potentially useful. Meanwhile, as one type of complexnetworks, HINs often encapsulate higher-order interactions that reflect theinterlocked nature among nodes and edges. Network motifs, sometimes referred toas meta-graphs, have been used as tools to capture such higher-orderinteractions and reveal the many different semantics. We therefore approach theproblem of user-guided clustering in HINs with network motifs. In this process,we identify the utility and importance of directly modeling higher-orderinteractions without collapsing them to pairwise interactions. To achieve this,we comprehensively transcribe the higher-order interaction signals to a seriesof tensors via motifs and propose the MoCHIN model based on joint non-negativetensor factorization. This approach applies to arbitrarily many, arbitraryforms of HIN motifs. An inference algorithm with speed-up methods is alsoproposed to tackle the challenge that tensor size grows exponentially as thenumber of nodes in a motif increases. We validate the effectiveness of theproposed method on two real-world datasets and three tasks, and MoCHINoutperforms all baselines in three evaluation tasks under three differentmetrics. Additional experiments demonstrated the utility of motifs and thebenefit of directly modeling higher-order information especially when userguidance is limited.
Xinwei Networks Network & Wireless Cards Drivers
Submission history
From: Yu Shi [view email][v1] Wed, 28 Nov 2018 00:16:03 UTC (614 KB)
[v2] Thu, 27 Jun 2019 02:51:29 UTC (1,509 KB)
[v3]Sun, 22 Sep 2019 22:39:09 UTC (1,048 KB)
Xinwei Networks Network & Wireless Cards Drivers
Full-text links:Download:
Current browse context:References & Citations
DBLP - CS Bibliography
Xinwei He
Naijing Zhang
Carl Yang
Jiawei Han
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs and how to get involved.