GraphGAN: Graph Representation Learning With Generative Adversarial Nets

Wang, Hongwei; Wang, Jia; Wang, Jialin; Zhao, Miao; Zhang, Weinan; Zhang, Fuzheng; Xie, Xing; Guo, Minyi

doi:10.1609/aaai.v32i1.11872

Cited by 390 publications

(71 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For each user node

V_{u_{i}}

, the generator tries to generate projects that

V_{u_{i}}

might like. Similar to work, 8,46 we use the following softmax function to achieve our generator

G

G (V_{p_{j}} ∣ V_{u_{i}}; θ_{G}) = \frac{exp ()(, g_{V_{p_{j}}}^{T} g_{V_{u_{i}}})}{\sum_{V_{p_{k}}} exp ()(, g_{V_{p_{k}}}^{T} g_{V_{u_{i}}})}

where

g_{V_{u_{i}}}

g_{V_{p_{j}}} \in θ_{G}

are the d‐dimensional vectors of

V_{u_{i}}

and

V_{p_{j}}

for generator

G

.…”

Section: The Proposed Scratchganmentioning

confidence: 99%

“…SDNE 11 utilizes first‐order proximity and second‐order proximity to preserve the local structure and global structure of the network, respectively. GraphGAN 8 brings in the generative adversarial nets to learn node representations. It builds a generator to simulate neighbor nodes over all nodes and trains a discriminator to identify true neighbor nodes from fake nodes generated by the generator.…”

Section: Related Workmentioning

confidence: 99%

“…For the former, in view of the essence that preferences are generated from users and passing to projects, we propose a new generating strategy for the generator. Instead of generating connected nodes for every node like GraphGAN, 8 our generator only tries to generate connected (liked) project nodes for user nodes, that is, our generator tries to learn the preference distribution over user nodes. For the latter, our discriminator tries to distinguish “fake” project nodes generated by the generator from “true‐connected” project nodes.…”

Section: Introductionmentioning

confidence: 99%

“…In the last few years, NRL has been a hot topic in network study. There has emerged a lot of excellent NRL models, such as DeepWalk, 4 LINE, 5 Node2Vec, 6 struc2vec, 7 GraphGAN, 8 ANE, 9 HARP, 10 SDNE, 11 and so on. However, there is little work focusing on representation learning for Scratch network.…”

Section: Introductionmentioning

confidence: 99%

“…For each user node V u i , the generator tries to generate projects that V u i might like. Similar to work, 8,46 we use the following softmax function to achieve our generator G.…”

mentioning

confidence: 99%

See 4 more Smart Citations

ScratchGAN: Network representation learning for scratch with preference‐based generative adversarial nets

Wang

Sun

et al. 2021

Int J of Intelligent Sys

View full text Add to dashboard Cite

With the rapid increase of users, Scratch, as a popular online social and programming platform, has accumulated massive project resources and complex relations across its social and programming learning network. However, it is challenging to utilize the network information for providing Scratch users with personalized services, despite there are already some useful network representation learning models. In this paper, a network representation learning model with preference-based generative adversarial nets for Scratch (ScratchGAN) is proposed to resolve this problem. In ScratchGAN, we first design a node-vector initialization approach to preserve structure information and side information of Scratch network. Then, considering to learn the fine-grained user preference information of network, we propose a novel Scratch adversarial learning model which includes a Scratch generative adversarial net and a user preference difference constraint component. The former aims to capture user preferences through a new generating strategy based on the delivery nature of preference. The latter attempts to embed users' detailed preference differences according to their interaction behaviors. ScratchGAN can mine user preferences while preserving network structure information and side information. Extensive experiments on the Scratch network

show abstract

“…For each user node

V_{u_{i}}

, the generator tries to generate projects that

V_{u_{i}}

might like. Similar to work, 8,46 we use the following softmax function to achieve our generator

G

G (V_{p_{j}} ∣ V_{u_{i}}; θ_{G}) = \frac{exp ()(, g_{V_{p_{j}}}^{T} g_{V_{u_{i}}})}{\sum_{V_{p_{k}}} exp ()(, g_{V_{p_{k}}}^{T} g_{V_{u_{i}}})}

where

g_{V_{u_{i}}}

g_{V_{p_{j}}} \in θ_{G}

are the d‐dimensional vectors of

V_{u_{i}}

and

V_{p_{j}}

for generator

G

.…”

Section: The Proposed Scratchganmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…For each user node V u i , the generator tries to generate projects that V u i might like. Similar to work, 8,46 we use the following softmax function to achieve our generator G.…”

mentioning

confidence: 99%

See 3 more Smart Citations