File Download
There are no files associated with this item.
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1007/s00778-023-00805-0
- Scopus: eid_2-s2.0-85166263046
- WOS: WOS:001040174300001
- Find via
Supplementary
- Citations:
- Appears in Collections:
Article: Accelerating directed densest subgraph queries with software and hardware approaches
Title | Accelerating directed densest subgraph queries with software and hardware approaches |
---|---|
Authors | |
Keywords | Convex programming Densest subgraph discovery Directed graph GPU |
Issue Date | 31-Jul-2023 |
Publisher | Springer |
Citation | The VLDB Journal, 2023, v. 33, p. 207 How to Cite? |
Abstract | Given a directed graph G, the directed densest subgraph (DDS) problem refers to finding a subgraph from G, whose density is the highest among all subgraphs of G. The DDS problem is fundamental to a wide range of applications, such as fake follower detection and community mining. Theoretically, the DDS problem closely connects to other essential graph problems, such as network flow and bipartite matching. However, existing DDS solutions suffer from efficiency and scalability issues. In this paper, we develop a convex-programming-based solution by transforming the DDS problem into a set of linear programs. Based on the duality of linear programs, we develop efficient exact and approximation algorithms. Particularly, our approximation algorithm can support flexible parameterized approximation guarantees. We further investigate using GPU to speed up the solution of convex programs in parallel and achieve hundreds of times speedup compared to the original Frank–Wolfe computation. We have performed an extensive empirical evaluation of our approaches on eight real large datasets. The results show that our proposed algorithms are up to five orders of magnitude faster than the state of the art. |
Persistent Identifier | http://hdl.handle.net/10722/338628 |
ISSN | 2023 Impact Factor: 2.8 2023 SCImago Journal Rankings: 1.853 |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Ma, C | - |
dc.contributor.author | Fang, Y | - |
dc.contributor.author | Cheng, R | - |
dc.contributor.author | Lakshmanan, LVS | - |
dc.contributor.author | Han, X | - |
dc.contributor.author | Li, X | - |
dc.date.accessioned | 2024-03-11T10:30:18Z | - |
dc.date.available | 2024-03-11T10:30:18Z | - |
dc.date.issued | 2023-07-31 | - |
dc.identifier.citation | The VLDB Journal, 2023, v. 33, p. 207 | - |
dc.identifier.issn | 1066-8888 | - |
dc.identifier.uri | http://hdl.handle.net/10722/338628 | - |
dc.description.abstract | Given a directed graph G, the directed densest subgraph (DDS) problem refers to finding a subgraph from G, whose density is the highest among all subgraphs of G. The DDS problem is fundamental to a wide range of applications, such as fake follower detection and community mining. Theoretically, the DDS problem closely connects to other essential graph problems, such as network flow and bipartite matching. However, existing DDS solutions suffer from efficiency and scalability issues. In this paper, we develop a convex-programming-based solution by transforming the DDS problem into a set of linear programs. Based on the duality of linear programs, we develop efficient exact and approximation algorithms. Particularly, our approximation algorithm can support flexible parameterized approximation guarantees. We further investigate using GPU to speed up the solution of convex programs in parallel and achieve hundreds of times speedup compared to the original Frank–Wolfe computation. We have performed an extensive empirical evaluation of our approaches on eight real large datasets. The results show that our proposed algorithms are up to five orders of magnitude faster than the state of the art. | - |
dc.language | eng | - |
dc.publisher | Springer | - |
dc.relation.ispartof | The VLDB Journal | - |
dc.rights | This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. | - |
dc.subject | Convex programming | - |
dc.subject | Densest subgraph discovery | - |
dc.subject | Directed graph | - |
dc.subject | GPU | - |
dc.title | Accelerating directed densest subgraph queries with software and hardware approaches | - |
dc.type | Article | - |
dc.identifier.doi | 10.1007/s00778-023-00805-0 | - |
dc.identifier.scopus | eid_2-s2.0-85166263046 | - |
dc.identifier.volume | 33 | - |
dc.identifier.epage | 207 | - |
dc.identifier.eissn | 0949-877X | - |
dc.identifier.isi | WOS:001040174300001 | - |
dc.identifier.issnl | 1066-8888 | - |