Stand for Something or Fall for Everything: Predict Misinformation Spread with Stance-Aware Graph Neural Networks

2023 | International Conference on Information Systems | Citations: 0

Authors: Chen, Zihan; Sun, Jingyi; Liu, Rong Emily; Mai, Feng

Abstract: Although pervasive spread of misinformation on social media platforms has becom ... Expand

Abstract: Although pervasive spread of misinformation on social media platforms has become a pressing challenge, existing platform interventions have shown limited success in curbing its dissemination. In this study, we propose a stance-aware graph neural network (stance-aware GNN) that leverages users' stances to proactively predict misinformation spread. As different user stances can form unique echo chambers, we customize four information passing paths in stance-aware GNN, while the trainable attention weights provide explainability by highlighting each structure's importance. Evaluated on a realworld dataset, stance-aware GNN outperforms benchmarks by 32.65% and exceeds advanced GNNs without user stance by over 4.69%. Furthermore, the attention weights indicate that users' opposition stances have a higher impact on their neighbors' behaviors than supportive ones, which function as social correction to halt misinformation propagation. Overall, our study provides an effective predictive model for platforms to combat misinformation, and highlights the impact of user stances in the misinformation propagation. Collapse

Topics: misinformation Twitter filter bubble social media social influence

Methods: machine learning RoBERTa experiment deep learning graph neural network

Live streaming recommendations based on dynamic representation learning

2023 | Decision Support Systems | Citations: 0

Authors: Gao, Ge; Liu, Hongyan; Zhao, Kang

Abstract: As an emerging form of social media, live streaming services (e.g., Twitch and C ... Expand

Abstract: As an emerging form of social media, live streaming services (e.g., Twitch and Clubhouse) allow users to interact with hosts and peers in real time while enjoying shows or participating in discussions. These platforms are also dynamic, with shows or discussions changing quickly inside a room and users frequently switching between rooms. To improve user engagement and experience on such platforms, we design a new recommendation model named Dynamic Representations for Live Streaming Rooms (DRIVER) to provide room recommendations. Guided by the Integrated Framework for Consumer Path Modeling and the social affordance theory, DRIVER infers dynamic representations of live streaming rooms by leveraging users’ behavior paths in entering, staying in, and leaving rooms. One contribution of our model is a new and efficient dynamic learning framework to model instantaneous and ever-changing inter-room relationships by considering individual users’ behavior paths after leaving a room. Also supported by social affordance theory, another methodological novelty of our model is to capture dynamic characteristics of a room by incorporating features of the current audience inside the room. Experiments on real-world datasets from two different types of live streaming platforms demonstrate that DRIVER outperforms state-of-the-art representation learning methods and sequential recommender systems. The proposed method also has implications for recommender system design in other contexts, in which items are characterized by users’ dynamic behavior paths and ongoing social interactions. Collapse

Topics: live streaming platform recommender system user experience knowledge sharing systems design

Methods: experiment sensitivity analysis ablation analysis descriptive statistic deep learning

Theories: theory of affordance

Fine-grained classification of drug trafficking based on Instagram hashtags

2023 | Decision Support Systems | Citations: 0

Authors: Hu, Chuanbo; Liu, Bin; Ye, Yanfang; Li, Xin

Abstract: Social networks have become important platforms for the marketing and sale of il ... Expand

Abstract: Social networks have become important platforms for the marketing and sale of illicit drugs. Hashtags make it easier for users to engage in drug trafficking, further increasing the risk of drug abuse. However, there are significant challenges in the detection and management of drug trafficking activities. In addition, the rapid legalization of some drugs has required a fine-grained classification of drugs to distinguish them from those that are illegal. Motivated by these observations, in this paper, our aim is to develop a methodology using the latest advances in AI technology to classify hashtags from posts advertising illicit drugs for sale on social networks. We present a semi-supervised deep learning approach to classify hashtags from posts advertising illicit drugs. An elegant combination of Bidirectional Encoder Representations from Transformers (BERT) with Graph Convolu­ tional Network (GCN) allows us to analyze the characteristics (e.g., shipping region and platform self-regulation) of illegal drug trafficking. Our BERT+GCN model achieved the best performance with more than 75% accuracy compared to the other three baseline models. Then, fine-grained hashtags identified are applied to explore the characteristics of drug trafficking. Finally, we report our results for further exploration of shipping regions and self-regulation of drug trafficking on the platform in two analysis scenarios. Our developed approach has shown its effectiveness in detecting hashtags for different types of drugs from illegal drug sellers. Based on hashtag classification, we also provide two case studies that indicate that (1) there are differences in self-regulation for different types of drugs on social media, (2) there are regional differences in the demand for different types of drugs. Collapse

Topics: Instagram social media Google+ artificial intelligence advertising management

Methods: BERT case study machine learning deep learning semi-supervised learning

A Graph Learning Model of Network Resources for Early Stage Startup Success Prediction

2023 | International Conference on Information Systems | Citations: 0

Authors: Liu, Mucan; Hu, Manting; Junming, Liu

Abstract: With high profitability accompanied by high risks, startups are driving industr ... Expand

Abstract: With high profitability accompanied by high risks, startups are driving industry evolution and innovation. However, due to high information asymmetry, startup success prediction remains challenging. From the perspective of network resources, we propose a variant heterogeneous graph attention network (ResourceNet) to model how a focal startup can access and leverage network resources from inter-organizational networks connected by investment, co-portfolio, and VC syndication relationships, for future success. We follow the design science paradigm to develop the node and link-aware attention mechanisms in graph network representations that jointly explore the impact of different mechanisms explaining the value of network resources, i.e., reach, richness, and receptivity. This project provides contributions to the startup success prediction studies by demonstrating the value of network resources in a topological interorganizational network, and also important managerial implications for startup companies (to seek network resources for future success) and VC (to pick the winners). Collapse

Methods: machine learning design science graph neural network artificial neural network simulation

Emotion-regulatory chatbots for enhancing consumer servicing: An interpersonal emotion management approach

2023 | Information & Management | Citations: 0

Authors: Luo, Bei; Lau, Raymond Y.K.; Li, Chunping

Abstract: To reduce service staff’s burden in customer servicing and improve the performan ... Expand

Abstract: To reduce service staff’s burden in customer servicing and improve the performance of automatic service re­ covery, we propose an emotion-regulatory chatbot for service recovery applications drawing on interpersonal emotion management (IEM) theory. In addition, we develop a model to examine the underlying mechanisms through which perceived IEM strategies influence consumers’ emotions and behavioral intentions. Our experi­ mental results verify the effectiveness of IEM strategies. We find that appraisals and consumers’ post-recovery emotions sequentially mediate the relationship between perceived emotion regulatory strategies and positive word-of-mouth (PWOM). Collapse

Topics: behavioral intention robotic electronic commerce customer service word of mouth

Methods: chatbot experiment experimental group structural equation modeling parametric test

Theories: control theory trust theory affect control theory

Leveraging Interactions for Stationary and Dynamic Financial Distress Prediction: A Spatio-Temporal Financial Graph Attention Network

2023 | International Conference on Information Systems | Citations: 0

Authors: Qi, Qi; Xing, Frank

Abstract: Financial distress prediction is a prominent research topic in information syst ... Expand

Abstract: Financial distress prediction is a prominent research topic in information systems, with two primary modelling categories: stationary and dynamic modelling. Recent stationary modelling works have leveraged company interactions to improve prediction performance, considering the heterogeneity of interactions while ignoring the dynamicity. However, few dynamic modelling works utilized interactions. To address the inconsistency and limitation of stationary and dynamic modelling works in leveraging interactions, we propose the Spatio-Temporal Financial Graph Attention Network with Metalearning (STFGAN-Meta). STFGAN-Meta leverages interactions' spatial heterogeneity via the Spatial Aggregation module and temporal dynamicity via the Temporal Aggregation module. STFGAN-Meta introduces the Meta-learning Optimization module to unify stationary and dynamic modelling. Our experimental evaluation demonstrates that leveraging dynamicity and heterogeneity of interactions outperforms leveraging dynamicity or heterogeneity alone. Meta-learning succeeds in providing a generalized approach between stationary and dynamic modelling. STFGAN-Meta can be a promising risk assessment and decision-making tool in the financial industry. Collapse

Topics: decision making database system accounting Python

Methods: meta learning experiment machine learning metamodel design science

CATCHM: A novel network-based credit card fraud detection method using node representation learning

2023 | Decision Support Systems | Citations: 0

Authors: Van Belle, Rafaël; Baesens, Bart; De Weerdt, Jochen

Abstract: Advanced fraud detection systems leverage the digital traces from (credit-card) ... Expand

Abstract: Advanced fraud detection systems leverage the digital traces from (credit-card) transactions to detect fraudulent activity in future transactions. Recent research in fraud detection has focused primarily on data analytics combined with manual feature engineering, which is tedious, expensive and requires considerable domain expertise. Furthermore, transactions are often examined in isolation, disregarding the interconnection that exists between them. Collapse

Topics: fraud detection credit card fraud privacy business process management electronic mail

Methods: computational algorithm experiment machine learning feature engineering literature study

Diversity Preference-Aware Link Recommendation for Online Social Networks

2023 | Information Systems Research | Citations: 0

Authors: Yin, Kexin; Fang, Xiao; Chen, Bintong; Sheng, Olivia R. Liu

Abstract: Link recommendation, which recommends links to connect unlinked online social ne ... Expand

Abstract: Link recommendation, which recommends links to connect unlinked online social network users, is a fundamental social network analytics problem with ample business implications. Existing link recommendation methods tend to recommend similar friends to a user but overlook the user’s diversity preference, although social psychology theories suggest the criticality of diversity preference to link recommendation performance. In recommender systems, a field related to link recommendation, a number of diversification methods have been proposed to improve the diversity of recommended items. Nevertheless, diversity preference is distinct from diversity studied by diversification methods. To address these research gaps, we define and operationalize the concept of diversity preference for link recommendation and propose a new link recommendation problem: the diversity preference-aware link recommendation problem. We then analyze key properties of the new link recommendation problem and develop a novel link recommendation method to solve the problem. Using two large-scale online social network data sets, we conduct extensive empirical evaluations to demonstrate the superior performance of our method over representative diversification methods adapted for link recommendation and state-of-the-art link recommendation methods. History: Eric Zheng, Senior Editor; Zhengrui Jiang, Associate Editor. Supplemental Material: The online appendices are available at https://doi.org/10.1287/isre.2022.1174. Collapse

Topics: recommender system social network resource allocation Google+

Methods: machine learning optimization model computational algorithm experiment integer programming

Collaborative group embedding and decision aggregation based on attentive influence of individual members: A group recommendation perspective

2023 | Decision Support Systems | Citations: 0

Authors: Yu, Li; Leng, Youfang; Zhang, Dongsong; He, Shuheng

Abstract: A key group decision making task is to aggregate individual preferences. Convent ... Expand

Abstract: A key group decision making task is to aggregate individual preferences. Conventional group decision methods adopt pre-defined and fixed strategies to aggregate individuals’ preferences, which can be ineffective due to the varying importance and influence of individual group members. Recent studies have proposed to assign different weights to individual members automatically based on the level of consistency of their ratings with group assessment outcomes. However, they ignored the high-order influence relationship among individual group members on group decision making. In this study, from a group recommendation perspective, we propose a novel collaborative Group Embedding and Decision Aggregation (GEDA) approach by leveraging the graph neural network technique to address those limitations. Specifically, GEDA first deploys a graph convolution operation on user-item interaction and group-item interaction graphs to generate embedding representations of members, groups, and items. A novel multi-attention (MA) module then learns each member’s decision weight by simul­ taneously considering the relationships among members for aggregating individual preferences into group preferences. The empirical evaluation using two real-world datasets demonstrates the advantage of the proposed GEDA model over the state-of-the-art group recommendation models. Collapse

Topics: recommender system decision making website

Methods: experiment deep learning graph neural network ablation analysis artificial neural network

Bring Me a Good One: Seeking High-potential Startups using Heterogeneous Venture Information Networks

2023 | International Conference on Information Systems | Citations: 0

Authors: zhang, shengming; Zhong, Hao; Ge, Yong; Xiong, Hui

Abstract: Identifying startups with the highest potential for success is a complex task, ... Expand

Abstract: Identifying startups with the highest potential for success is a complex task, necessitating the examination of various information sources, including firm demographics, management team composition, and financial performance. The effectiveness of existing methodologies, such as feature-based and network-topological approaches, is limited for predicting highpotential startups. In response, we propose a novel Venture Graph Neural Network (VenGNN) model, leveraging Heterogeneous Information Networks (HIN) and Graph Neural Networks (GNN) techniques to address the prediction problem. Specifically, we construct a Heterogeneous Venture Information Network (HVIN) using raw business data and deem the prediction as a node classification task. Our model integrates theory-guided semantic meta-paths, firm demographics, sampling-based self-attention, and centrality encoding to overcome certain constraints of existing GNNs. Our experimental analysis reveals that VenGNN outperforms state-of-the-art models by 15-20% across a wide range of performance metrics. Our study also includes a comprehensive interpretation analysis to provide investors with an essential understanding for better decision-making. Collapse

Topics: database system innovation management decision making data breach knowledge sharing

Methods: graph neural network experiment machine learning ablation analysis XGBoost

Theories: theory of economic growth