ABSTRACT
In the Age of Big Data, graph embedding has received increasing attention for its ability to accommodate the explosion in data volume and diversity, which challenge the foundation of modern recommender systems. Respectively, graph facilitates fusing complex systems of interactions into a unified structure and distributed embedding enables efficient retrieval of entities, as in the case of approximate nearest neighbor (ANN) search. When combined, graph embedding captures relational information beyond entity interaction and towards a problem's underlying structure, as epitomized by struct2vec [20] and PinSage [26]. This session will start by brushing up on the basics about graphs and embedding methods and discussing their merits. We then quickly dive into using the mathematical formulation of graph embedding to derive the modular framework: Sampler-Mapper-Optimizer for Recommendation, or SMORe. We demonstrate existing models used for recommendation, such as MF and BPR, can all be assembled using three basic components: sampler, mapper, and optimizer. The tutorial is accompanied by a hands-on session, where we show how graph embedding can model complex systems through the multi-task learning and the cross-platform data sparsity alleviation tasks.
- O. Barkan and N. Koenigstein. {n.d.}. Item2vec: neural item embedding for collaborative filtering (MLSP 2016).Google Scholar
- Z. Cao, T. Qin, T.-Y. Liu, M.-F. Tsai, and H. Li. {n.d.}. Learning to rank: from pairwise approach to listwise approach (ICML 2007). Google ScholarDigital Library
- C.-Y. Chao, Y.-F. Chu, H.-W. Yang, C.-J. Wang, and M.-F. Tsai. {n.d.}. Text Embedding for Sub-Entity Ranking from User Reviews (CIKM 2017). Google ScholarDigital Library
- C.-M. Chen, P.-C. Chien, Y.-C. Lin, M.-F. Tsai, and Y.-H. Yang. {n.d.}. Exploiting Latent Social Listening Representations for Music Recommendations (RecSys 2015).Google Scholar
- C.-M. Chen, M.-F. Tsai, Y.-C. Lin, and Y.-H. Yang. {n.d.}. Query-based Music Recommendations via Preference Embedding (RecSys 2016). Google ScholarDigital Library
- C.-M. Chen, C.-J. Wang, M.-F. Tsai, and Y.-H. Yang. {n.d.}. Collaborative Similarity Embedding for Recommender Systems (WWW 2019). Google ScholarDigital Library
- C.-M. Chen, C.-Y. Yang, C.-C. Hsia, Y. Chen, and M.-F. Tsai. {n.d.}. Music Playlist Recommendation via Preference Embedding (RecSys 2016). Google ScholarDigital Library
- C.-M. Chen, Y.-H. Yang, Y. Chen, and M.-F. Tsai. arXiv 2017. Vertex-Context Sampling for Weighted Network Embedding.Google Scholar
- Y. Dong, N. V. Chawla, and A. Swami. {n.d.}. metapath2vec: Scalable representation learning for heterogeneous networks (SIGKDD 2017). Google ScholarDigital Library
- R. He, W.-C. Kang, and J. McAuley. {n.d.}. Translation-based Recommendation (RecSys 2017). Google ScholarDigital Library
- C.-C. Hsia, K.-H. Lai, Y. Chen, C.-J. Wang, and M.-F. Tsai. {n.d.}. Representation Learning for Image-based Music Recommendation (LBRS of RecSys 2018).Google Scholar
- Y. Koren, R. Bell, and C. Volinsky. 2009. Matrix Factorization Techniques for Recommender Systems. Computer (2009). Google ScholarDigital Library
- K.-H. Lai, C.-M. Chen, M.-F. Tsai, and C.-J. Wang. {n.d.}. NavWalker: Information Augmented Network Embedding (WI 2018).Google Scholar
- K.-H. Lai, T.-H. Wang, H.-Y. Chi, Y. Chen, M.-F. Tsai, and C.-J. Wang. {n.d.}. Superhighway: Bypass Data Sparsity in Cross-Domain CF (LBRS of RecSys 2018).Google Scholar
- T. Mikolov, I. Sutskever, K. Chen, G. S. Corrado, and J. Dean. {n.d.}. Distributed representations of words and phrases and their compositionality (NIPS 2013). Google ScholarDigital Library
- X. Ning and G. Karypis. {n.d.}. Slim: Sparse linear methods for top-n recommender systems (ICDM 2011). Google ScholarDigital Library
- B. Perozzi, R. Al-Rfou, and S. Skiena. {n.d.}. DeepWalk: Online Learning of Social Representations (SIGKDD 2014). Google ScholarDigital Library
- T. Qin, X.-D. Zhang, M.-F. Tsai, D-S. Wang, T.-Y. Liu, and H. Li. 2008. Query-level loss functions for information retrieval. Information Processing & Management (2008). Google ScholarDigital Library
- B. Recht, C. Re, S. Wright, and F. Niu. {n.d.}. Hogwild: A lock-free approach to parallelizing stochastic gradient descent (NIPS 2011). Google ScholarDigital Library
- L. F. R. Ribeiro, P. H. P. Saverese, and D. R. Figueiredo. {n.d.}. struc2vec: Learning node representations from structural identity (SIGKDD 2017). Google ScholarDigital Library
- J. Tang, M. Qu, M. Wang, M. Zhang, J. Yan, and Q. Mei. {n.d.}. LINE: Large-scale information network embedding (WWW 2015). Google ScholarDigital Library
- M.-F. Tsai, T.-Y. Liu, T. Qin, H.-H. Chen, and W.-Y. Ma. {n.d.}. FRank: a ranking method with fidelity loss (SIGIR 2007). Google ScholarDigital Library
- C.-H. Wang, K.-C. Fan, C.-J. Wang, and M.-F. Tsai. {n.d.}. UGSD: User Generated Sentiment Dictionaries from Online Customer Reviews (AAAI 2019).Google Scholar
- C.-J. Wang, T.-H. Wang, H.-W. Yang, B.-S. Chang, and M.-F. Tsai. {n.d.}. ICE: Item concept embedding via textual information (SIGIR 2017). Google ScholarDigital Library
- J.-H. Yang, C.-M. Chen, C.-J. Wang, and M.-F. Tsai. {n.d.}. HOP-rec: High-order Proximity for Implicit Recommendation (RecSys 2018). Google ScholarDigital Library
- R. Ying, R. He, K. Chen, P. Eksombatchai, W. L. Hamilton, and J. Leskovec. {n.d.}. Graph convolutional neural networks for web-scale recommender systems (SIGKDD 2018). Google ScholarDigital Library
Index Terms
- SMORe: modularize graph embedding for recommendation
Recommendations
The nonorientable genus of joins of complete graphs with large edgeless graphs
We show that for n=4 and n>=6, K"n has a nonorientable embedding in which all the facial walks are hamilton cycles. Moreover, when n is odd there is such an embedding that is 2-face-colorable. Using these results we consider the join of an edgeless ...
Embedding into Bipartite Graphs
The conjecture of Bollobás and Komlós, recently proved by Böttcher, Schacht, and Taraz [Math. Ann., 343 (2009), pp. 175-205], implies that for any $\gamma>0$, every balanced bipartite graph on $2n$ vertices with bounded degree and sublinear bandwidth ...
Families of Dot-Product Snarks on Orientable Surfaces of Low Genus
We introduce a generalized dot product and provide some embedding conditions under which the genus of a graph does not rise under a dot product with the Petersen graph. Using these conditions, we disprove a conjecture of Tinsley and Watkins on the genus ...
Comments