Combining content and link for classification using matrix factorization

By Shenghuo Zhu, Kai Yu, Yun Chi, Yihong Gong. Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 2007

297 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/5c58ad9a6c09782814a7d048bebd6ef1609c0fb4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Charu C. Aggarwal
2. Peng Cui
3. Qiang Yang
4. Shiyu Chang
5. Irwin King
6. Dacheng Tao
7. Yoshua Bengio
8. Jiliang Tang

Citations in the corpus

• Link Prediction via Matrix Factorization (2011, methodology): “For directed graphs, we can let Λ be an arbitrary asymmetric matrix, following [38,23].”
• Probabilistic Latent Document Network Embedding (2014, methodology): “Following [35], we use the following categories as four separate datasets: Data Structure (DS), Hardware and Architecture (HA), Machine Learning (ML), and Programming Language (PL).”
• A Survey of Text Classification Algorithms (2012, methodology): “It was then shown in [140] that this feature representation provides more accurate results, when used in conjunction with an SVM-classifier.”
• SoRec: social recommendation using probabilistic matrix factorization (2008, methodology): “Although recently, similar factor analysis methods have been employed in [27, 28] for document retrieval and document classification, our approach has three essential differences compared with these methods: (1) Our method can deal with missing value problem, while their methods cannot.”
• Heterogeneous Network Embedding via Deep Architectures (2015, methodology): “Unlike to traditional linear embedding models [29, 44, 50], the proposed method decomposes the feature learning process into multiple nonlinear layers of a deep architecture.”
• Collaborative Filtering beyond the User-Item Matrix (2014, methodology): “We also note that the work related to CMF was proposed slightly earlier than Singh and Gordon [2008] and applied to the task, not of recommendation but of document classification [Zhu et al. 2007].”
• Efficient community detection in large networks using content and links (2012, methodology): “Previously SA-Cluster-Inc has been shown to outperform k-SNAP [25] and PCLDC to outperform methods including PLSA-PHITS [5], LDA-Link-Word [7] and LinkContent-Factorization [31].”
• Accelerated Attributed Network Embedding (2017, methodology): “Several algorithms [20, 43] have been proposed to exploit the potential of learning a joint lowrank latent representation.”
• A Unified View of Matrix Factorization Models (2008, methodology): “[36], using a smooth variant of the hinge loss.”
• A Survey of Signed Network Mining in Social Media (2015, methodology): “Just as unsigned networks are used frequently in the context of various applications such as data classification [Zhu et al. 2007], data clustering [Long et al. 2006], information propagation [Kempe et al. 2003] and recommendation [Tang et al. 2013], signed networks can be leveraged as well.”
• Exploiting homophily effect for trust prediction (2013, methodology): “LOW-RANK MATRIX FACTORIZATION MODEL FOR TRUST PREDICTION Low-rank matrix factorization method is widely employed in various applications such as collective filtering [8, 21] and document clustering [26, 2].”
• A Survey of Multi-View Representation Learning (2016, unknown): “[172] introduce a joint matrix factorization model to combine the information of content and links for web page analysis.”
• Label Informed Attributed Network Embedding (2017, methodology): “In order to tackle the complex data structures, several efforts [28, 47] have been devoted to jointly embedding the two information sources into a unified space.”
• Generalized Latent Factor Models for Social Network Analysis (2011, unknown): “, 2004], Link-Content-Factorization (LCF) [Zhu et al., 2007], NCUT, PCL-PLSA, PHITSDC, PCL-DC and C-PLDC [Yang et al.”
• Relational learning via collective matrix factorization (2008, methodology): “[37], using a once-differentiable variant of the Hinge loss.”
• Exploring Context and Content Links in Social Media: A Latent Space Method (2012, background): “While most of these algorithms focus on learning the latent space solely with context links, some efforts have been made to incorporate content information [31].”