An official website of the United States government
Cyberbullying has become one of the most pressing online risks for adolescents and has raised serious concerns in society. Recent years have witnessed a surge in research aimed at developing principled learning models to detect cyberbullying behaviors. These efforts have primarily focused on building a single generic classification model to differentiate bullying content from normal (non-bullying) content among all users. These models treat users equally and overlook idiosyncratic information about users that might facilitate the accurate detection of cyberbullying. In this paper, we propose a personalized cyberbullying detection framework, PI-Bully, that draws on empirical findings from psychology highlighting unique characteristics of victims and bullies and peer influence from like-minded users as predictors of cyberbullying behaviors. Our framework is novel in its ability to model peer influence in a collaborative environment and tailor cyberbullying prediction for each individual user. Extensive experimental evaluations on real-world datasets corroborate the effectiveness of the proposed framework.
more »
« less
Personalized Learning for Cyberbullying Detection
Cyberbullying has become one of the most pressing online risks for adolescents and has raised serious concerns in society. Traditional efforts are primarily devoted to building a single generic classification model for all users to differentiate bullying behaviors from the normal content [6, 3, 1, 2, 4]. Despite its empirical success, these models treat users equally and inevitably ignore the idiosyncrasies of users. Recent studies from psychology and sociology suggest that the occurrence of cyberbullying has a strong connection with the personality of victims and bullies embedded in the user-generated content, and the peer influence from like-minded users. In this paper, we propose a personalized cyberbullying detection framework PI-Bully with peer influence in a collaborative environment to tailor the prediction for each individual. In particular, the personalized classifier of each individual consists of three components: a global model that captures the commonality shared by all users, a personalized model that expresses the idiosyncratic personality of each specific user, and a third component that encodes the peer influence received from like-minded users. Most of the existing methods adopt a two-stage approach: they first apply feature engineering to capture the cyberbullying patterns and then employ machine learning classifiers to detect cyberbullying behaviors. However, building a personalized cyberbullying detection framework that is customized to each individual remains a challenging task, in large part because: (1) Social media data is often sparse, noisy and high-dimensional (2) It is important to capture the commonality shared by all users as well as idiosyncratic aspects of the personality of each individual for automatic cyberbullying detection; (3) In reality, a potential victim of cyberbullying is often influenced by peers and the influences from different users could be quite diverse. Hence, it is imperative to develop a way to encode the diversity of peer influence for cyberbullying detection. To summarize, we study a novel problem of personalized cyberbullying detection with peer influence in a collaborative environment, which is able to jointly model users' common features, unique personalities and peer influence to identify cyberbullying cases.
more »
« less
- Award ID(s):
- 1719722
- PAR ID:
- 10067396
- Date Published:
- Journal Name:
- Doctoral Consortium of the International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction and Behavior Representation in Modeling and Simulation
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Personalization has emerged as a critical research area in modern intelligent systems, focusing on mining users' behavioral history and adapting to their preferences for delivering tailored experiences. Despite the remarkable few-shot capabilities exhibited by black-box large language models (LLMs), the inherent opacity of their model parameters presents significant challenges in aligning the generated output with individual expectations. Existing solutions have primarily focused on prompt design to incorporate user-specific profiles and behaviors; however, such approaches often struggle to generalize effectively due to their inability to capture shared knowledge among all users. To address these challenges, we propose HYDRA, a model factorization framework that captures both user-specific behavior patterns from historical data and shared general knowledge among all users to deliver personalized generation. In order to capture user-specific behavior patterns, we first train a reranker to prioritize the most useful information from top-retrieved relevant historical records. By combining the prioritized history with the corresponding query, we train an adapter to align the output with individual user-specific preferences, eliminating the reliance on access to inherent model parameters of black-box LLMs. Both the reranker and the adapter can be decomposed into a base model with multiple user-specific heads, resembling a hydra. The base model maintains shared knowledge across users, while the multiple personal heads capture user-specific preferences. Experimental results demonstrate that \method outperforms existing state-of-the-art prompt-based methods by an average relative improvement of 9.01% across five diverse personalization tasks in the LaMP benchmark.more » « less
-
Collaborative systems design is a human-centered activity dependent on individual decision-making processes. Personality traits have been found to influence individual behaviors and tendencies to compete or cooperate. This paper investigates the effects of Big Five and Locus of Control personality traits on negotiated outcomes of a simplified collaborative engineering design task. Secondary data includes results from short-form personality inventories and outcomes of pair design tasks. The data includes ten sessions of four participants each, where each participant completes a sequence of 12 pair tasks involving design space exploration and negotiation. Regression analysis shows a statistically-significant relationship between Big Five and Locus of Control and total individual value accumulated across the 12 design tasks. Results show the Big Five, aggregating extraversion, agreeableness, conscientiousness, neuroticism, and intellect/imagination to a single factor, negatively affects individual value and internal Locus of Control positively affects individual value. Future work should consider a dedicated experiment to refine understanding of how personality traits influence collaborative systems design and propose interventions to improve collaborative design processes.more » « less
-
Collaborative infrastructure systems are vital for managing scarce resources, particularly where user behaviors influence system sustainability. This study examines the relationship between design of constructed water infrastructure and strategic behaviors, focusing on flood irrigation systems as an example of collaborative infrastructure. The objectives are to investigate 1) whether shared water infrastructure can be effectively modeled using the stag hunt game framework and 2) how network topology impacts the strategic stability of user cooperation. Flood irrigation relies on collective action, where users balance risks of collaboration failure against benefits of successful cooperation. This situation closely aligns with stag hunt dynamics, in which users choose between a higher-value but riskier collaborative strategy or a lower-value, safer independent option. A key challenge arises when users opt out, increasing the burden on remaining collaborators. We apply a game-theoretic model using risk dominance criteria to analyze stability across four distinct infrastructure topologies: linear, tree, bus, and star. Results identify star and bus topologies as Pareto efficient, where a bus topology offers greater economic efficiency through reduced infrastructure costs and a star topology enhances stability due to equitable distribution of influence and reduced dependencies. An agent-based simulation validates analytical findings by dynamically captures user interactions under uncertainty and showing a strong correlation with game-theoretic results. Consequently, this study confirms the applicability of stag hunt frameworks for analyzing collaborative water infrastructure and provides practical insights into how topology design can influence cooperative resilience. These findings enhance knowledge for sustainable improvement of collaborative infrastructure.more » « less
-
null (Ed.)Due to the prevalence and severe consequences of cyberbullying, numerous research works have focused on mining and analyzing social network data to understand cyberbullying behavior and then using the gathered insights to develop accurate classifiers to detect cyberbullying. Some recent works have been proposed to leverage the detection classifiers in a centralized cyberbullying detection system and send notifications to the concerned authority whenever a person is perceived to be victimized. However, two concerns limit the effectiveness of a centralized cyberbullying detection system. First, a centralized detection system gives a uniform severity level of alerts to everyone, even though individual guardians might have different tolerance levels when it comes to what constitutes cyberbullying. Second, the volume of data being generated by old and new social media makes it computationally prohibitive for a centralized cyberbullying detection system to be a viable solution. In this work, we propose BullyAlert, an android mobile application for guardians that allows the computations to be delegated to the hand-held devices. In addition to that, we incorporate an adaptive classification mechanism to accommodate the dynamic tolerance level of guardians when receiving cyberbullying alerts. Finally, we include a preliminary user analysis of guardians and monitored users using the data collected from BullyAlert usage.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
