%0 Journal Article %1 goyal_benchmarks_2019 %A Goyal, Palash %A Huang, Di %A Goswami, Ankita %A Chhetri, Sujit Rokka %A Canedo, Arquimedes %A Ferrara, Emilio %D 2019 %J arXiv:1908.06543 cs %K Benchmark Node_Embeddings Survey %T Benchmarks for Graph Embedding Evaluation %U http://arxiv.org/abs/1908.06543 %X Graph embedding is the task of representing nodes of a graph in a low-dimensional space. In recent years, graph embedding applications for link prediction, node classification, and graph visualization have gained significant traction in academia and industry. The primary difference among the many recently proposed graph embedding methods is the way they preserve the inherent properties of the graphs. However, in practice, comparing these methods is very challenging. The majority of methods report performance boosts on few selected real graphs. Therefore, it is difficult to generalize these performance improvements to other types of graphs because they may be attributed to specific properties of the selected networks. Given a graph, it is currently impossible to conclusively determine which methods perform best, and to quantify the advantages of one approach over another. In this work, we introduce a principled framework to compare graph embedding methods. Our goal is three fold: (i) provide a unifying framework for comparing the performance of various graph embedding methods, (ii) establish a benchmark with real-world graphs that exhibit different structural properties, and (iii) provide users with a tool to identify the best graph embedding method for their data.

@article{goyal_benchmarks_2019, abstract = {Graph embedding is the task of representing nodes of a graph in a low-dimensional space. In recent years, graph embedding applications for link prediction, node classification, and graph visualization have gained significant traction in academia and industry. The primary difference among the many recently proposed graph embedding methods is the way they preserve the inherent properties of the graphs. However, in practice, comparing these methods is very challenging. The majority of methods report performance boosts on few selected real graphs. Therefore, it is difficult to generalize these performance improvements to other types of graphs because they may be attributed to specific properties of the selected networks. Given a graph, it is currently impossible to conclusively determine which methods perform best, and to quantify the advantages of one approach over another. In this work, we introduce a principled framework to compare graph embedding methods. Our goal is three fold: (i) provide a unifying framework for comparing the performance of various graph embedding methods, (ii) establish a benchmark with real-world graphs that exhibit different structural properties, and (iii) provide users with a tool to identify the best graph embedding method for their data.}, added-at = {2020-02-21T16:09:44.000+0100}, author = {Goyal, Palash and Huang, Di and Goswami, Ankita and Chhetri, Sujit Rokka and Canedo, Arquimedes and Ferrara, Emilio}, biburl = {https://www.bibsonomy.org/bibtex/21a58b24263cf4589470043d95c0ad433/tschumacher}, file = {Goyal et al - Benchmarks for Graph Embedding Evaluation.pdf:C\:\\Users\\Admin\\Documents\\Research\\_Paperbase\\Graph Embeddings\\Goyal et al - Benchmarks for Graph Embedding Evaluation.pdf:application/pdf}, interhash = {001dcb2aaf1ed0664d790470a091f84e}, intrahash = {1a58b24263cf4589470043d95c0ad433}, journal = {arXiv:1908.06543 [cs]}, keywords = {Benchmark Node_Embeddings Survey}, language = {en}, month = aug, note = {arXiv: 1908.06543}, timestamp = {2020-02-21T16:09:44.000+0100}, title = {Benchmarks for {Graph} {Embedding} {Evaluation}}, url = {http://arxiv.org/abs/1908.06543}, urldate = {2019-12-10}, year = 2019 }