Popularity Prediction on Social Platforms with Coupled Graph Neural Networks

Influence measurement in social networks is vital to various real-world applications, such as online marketing and political campaigns. In this paper, we investigate the problem of measuring time-sensitive and topic-specific influence based on streaming texts and dynamic social networks. A user's influence can change rapidly in response to a new event and vary on different topics. For example, the political influence of Douglas Jones increased dramatically after winning the Alabama special election, and then rapidly decreased after the election week. During the same period, however, Douglas Jones' influence on sports remained low. Most existing approaches can only model the influence based on static social network structures and topic distributions. Furthermore, as popular social networking services embody many features to connect their users, multi-typed interactions make it hard to learn the roles that different interactions play when propagating information. To address these challenges, we propose a Time-sensitive and Topic-specific Influence Measurement (TTIM) method, to jointly model the streaming texts and dynamic social networks. We simulate the influence propagation process with a self-attention mechanism to learn the contributions of different interactions and track the influence dynamics with a matrix-adaptive long short-term memory. To the best of our knowledge, this is the first attempt to measure time-sensitive and topic-specific influence. Furthermore, the TTIM model can be easily adapted to supporting online learning which consumes constant training time on newly arrived data for each timestamp. We comprehensively evaluate the proposed TTIM model on five datasets from Twitter and Reddit. The experimental results demonstrate promising performance compared to the state-of-the-art social influence analysis models and the potential of TTIM in visualizing influence dynamics and topic distribution.

show abstract

“…• CoupledGNN [27] applies two coupled graph neural networks to iteratively model and predicts the network-aware popularity.…”

Section: ) Baselinesmentioning

confidence: 99%

Measuring Time-Sensitive and Topic-Specific Influence in Social Networks With LSTM and Self-Attention

et al. 2020

View full text Add to dashboard Cite

show abstract

“…(1) Compared with TAP [12], TwitterRank [15] and TS-SRW that directly identify influencers in rare topics from the massive social messages, DID outperforms by iteratively learning an accurate subject filter. (2) As combination models, Filter+ReF [13], Filter+RR-LT [17] and Fil-ter+CoupledGNN [3] train the influencer detection model with fixed social message filters. In contrast, our proposed DID can adaptively update the subject filter for better influencer detection performance.…”

Section: Experiments 31 Experimental Setupmentioning

confidence: 99%

“…To take full advantage of multiple interactions in social networks, we integrate them with trainable weight functions. The influence propagation is modeled by GNNs with a loss function considering neighborhood and topic [3,13,17], while several studies aim to identify topic-specific influencers [12,14,15]. Although some approaches [1,2,7] also integrate topic discovery and social influence analysis, the influencers are restricted to popular topics in social media only.…”

Section: Introductionmentioning

confidence: 99%

On-demand Influencer Discovery on Social Media

Zheng

Zhang

Young

et al. 2020

Proceedings of the 29th ACM International Conference on Information &Amp; Knowledge Management

View full text Add to dashboard Cite

Identifying influencers on social media, such as Twitter, has played a central role in many applications, including online marketing and political campaigns. Compared with social media celebrities, domain-specific influencers are less expensive to hire and more engaged in spreading messages such as new treatment or timely prevention for HIV. However, most of the existing topic modeling based approaches fail to identify influencers who are dedicated to the rare yet important topics such as HIV and suicide. To alleviate this limitation, we investigate an on-Demand Influencer Discovery (DID) framework that is able to identify influencers on any subject depicted by a few user-specified keywords, regardless of its popularity on social media. The DID model employs an iterative learning process that integrates the language attention network as a subject filter and the influence convolution network built on user interactions. Comprehensive evaluations on Twitter datasets show that the DID model can reliably identify influencers even on rare subjects such as HIV and suicide, outperforming existing topic-specific influencer detection models.

show abstract

“…Consequently, the elements of % refer to buckets or lines, each containing a list of entries. Each % [8] contains the address of the i-th bucket. Since entries have a similar size, i.e., a user 83 and a set of positions with a number of elements between 1 and f 1, we can store in a block with size ⌫ between ⌫/(|83 |+(f 1).|?…”

Section: =3mentioning

confidence: 99%

“…These dierent characteristics make these platforms a tool particularly used to communicate information to a large number of people. In these networks, the most popular and inuential users have quickly been the center of attention for many applications, since they will accelerate the spread of information to the greatest number of users [8]. For instance, for online advertising campaigns on social networks or on the Web, advertisers seek to place their advertisements among the users who have the most visibility in order to reach a maximum of people [2,5,9].…”

Section: Introductionmentioning

confidence: 99%

A pattern-based approach for an early detection of popular Twitter accounts

Debure

Brunessaux

Constantin

et al. 2020

Proceedings of the 24th Symposium on International Database Engineering &Amp; Applications

View full text Add to dashboard Cite

Social networks (SN) are omnipresent in our lives today. Not all users have the same behaviour on these networks. If some have a low activity, rarely posting messages and following few users, some others at the other extreme have a signicant activity, with many followers and regularly posts. The important role of these popular SN users makes them the target of many applications for example for content monitoring or advertising. It is therefore relevant to be able to predict as soon as possible which SN users will become popular. In this work, we propose a technique for early detection of such users based on the identication of characteristic patterns. We present an index, 2 ", which allows a scaling up of our approach to large social networks. We also describe our rst experiments that conrm the validity of our approach. CCS CONCEPTS • Information systems ! Social networks.

show abstract

Popularity Prediction on Social Platforms with Coupled Graph Neural Networks

Cited by 117 publications

References 29 publications

Measuring Time-Sensitive and Topic-Specific Influence in Social Networks With LSTM and Self-Attention

Measuring Time-Sensitive and Topic-Specific Influence in Social Networks With LSTM and Self-Attention

On-demand Influencer Discovery on Social Media

A pattern-based approach for an early detection of popular Twitter accounts

Contact Info

Product

Resources

About