Details
| Original language | English |
|---|---|
| Title of host publication | Machine Learning and Knowledge Discovery in Databases |
| Subtitle of host publication | European Conference, ECML PKDD 2019, Würzburg, Germany, September 16–20, 2019, Proceedings, Part I |
| Editors | Ulf Brefeld, Elisa Fromont, Andreas Hotho, Arno Knobbe, Marloes Maathuis, Céline Robardet |
| Place of Publication | Cham |
| Publisher | Springer Verlag |
| Pages | 395-411 |
| Number of pages | 17 |
| ISBN (electronic) | 9783030461508 |
| ISBN (print) | 9783030461492 |
| Publication status | Published - 30 Apr 2020 |
| Event | 19th Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2019 - Wurzburg, Germany Duration: 16 Sept 2019 → 20 Sept 2019 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 11906 |
| ISSN (Print) | 0302-9743 |
| ISSN (electronic) | 1611-3349 |
Abstract
We propose a novel approach for learning node representations in directed graphs, which maintains separate views or embedding spaces for the two distinct node roles induced by the directionality of the edges. We argue that the previous approaches either fail to encode the edge directionality or their encodings cannot be generalized across tasks. With our simple alternating random walk strategy, we generate role specific vertex neighborhoods and train node embeddings in their corresponding source/target roles while fully exploiting the semantics of directed graphs. We also unearth the limitations of evaluations on directed graphs in previous works and propose a clear strategy for evaluating link prediction and graph reconstruction in directed graphs. We conduct extensive experiments to showcase our effectiveness on several real-world datasets on link prediction, node classification and graph reconstruction tasks. We show that the embeddings from our approach are indeed robust, generalizable and well performing across multiple kinds of tasks and graphs. We show that we consistently outperform all baselines for node classification task. In addition to providing a theoretical interpretation of our method we also show that we are considerably more robust than the other directed graph approaches.
Keywords
- cs.SI, cs.LG, stat.ML, Graph reconstruction, Link prediction, Directed graphs, Node classification, Node representations
ASJC Scopus subject areas
- Mathematics(all)
- Theoretical Computer Science
- Computer Science(all)
- General Computer Science
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2019, Würzburg, Germany, September 16–20, 2019, Proceedings, Part I. ed. / Ulf Brefeld; Elisa Fromont; Andreas Hotho; Arno Knobbe; Marloes Maathuis; Céline Robardet. Cham: Springer Verlag, 2020. p. 395-411 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11906).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Node Representation Learning for Directed Graphs
AU - Khosla, Megha
AU - Leonhardt, Jurek
AU - Nejdl, Wolfgang
AU - Anand, Avishek
N1 - Funding Information: grant agreement No. 654024). Funding Information: Acknowledgements. This work is partially funded by SoBigData (EU’s Horizon 2020
PY - 2020/4/30
Y1 - 2020/4/30
N2 - We propose a novel approach for learning node representations in directed graphs, which maintains separate views or embedding spaces for the two distinct node roles induced by the directionality of the edges. We argue that the previous approaches either fail to encode the edge directionality or their encodings cannot be generalized across tasks. With our simple alternating random walk strategy, we generate role specific vertex neighborhoods and train node embeddings in their corresponding source/target roles while fully exploiting the semantics of directed graphs. We also unearth the limitations of evaluations on directed graphs in previous works and propose a clear strategy for evaluating link prediction and graph reconstruction in directed graphs. We conduct extensive experiments to showcase our effectiveness on several real-world datasets on link prediction, node classification and graph reconstruction tasks. We show that the embeddings from our approach are indeed robust, generalizable and well performing across multiple kinds of tasks and graphs. We show that we consistently outperform all baselines for node classification task. In addition to providing a theoretical interpretation of our method we also show that we are considerably more robust than the other directed graph approaches.
AB - We propose a novel approach for learning node representations in directed graphs, which maintains separate views or embedding spaces for the two distinct node roles induced by the directionality of the edges. We argue that the previous approaches either fail to encode the edge directionality or their encodings cannot be generalized across tasks. With our simple alternating random walk strategy, we generate role specific vertex neighborhoods and train node embeddings in their corresponding source/target roles while fully exploiting the semantics of directed graphs. We also unearth the limitations of evaluations on directed graphs in previous works and propose a clear strategy for evaluating link prediction and graph reconstruction in directed graphs. We conduct extensive experiments to showcase our effectiveness on several real-world datasets on link prediction, node classification and graph reconstruction tasks. We show that the embeddings from our approach are indeed robust, generalizable and well performing across multiple kinds of tasks and graphs. We show that we consistently outperform all baselines for node classification task. In addition to providing a theoretical interpretation of our method we also show that we are considerably more robust than the other directed graph approaches.
KW - cs.SI
KW - cs.LG
KW - stat.ML
KW - Graph reconstruction
KW - Link prediction
KW - Directed graphs
KW - Node classification
KW - Node representations
UR - http://www.scopus.com/inward/record.url?scp=85084837443&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-46150-8_24
DO - 10.1007/978-3-030-46150-8_24
M3 - Conference contribution
SN - 9783030461492
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 395
EP - 411
BT - Machine Learning and Knowledge Discovery in Databases
A2 - Brefeld, Ulf
A2 - Fromont, Elisa
A2 - Hotho, Andreas
A2 - Knobbe, Arno
A2 - Maathuis, Marloes
A2 - Robardet, Céline
PB - Springer Verlag
CY - Cham
T2 - 19th Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2019
Y2 - 16 September 2019 through 20 September 2019
ER -