Abstract The rapid proliferation of online misinformation poses a critical challenge to digital information ecosystems, driven by coordinated communities and dynamic network interactions. While existing detection models often treat social networks as static or focus narrowly on content, this study investigates the complex interplay between temporal bursts, structural persistence, and community dynamics. Utilizing a novel Misinformation Dynamics Testbed, we analyze a longitudinal Reddit dataset (2015–2024,... The rapid proliferation of online misinformation poses a critical challenge to digital information ecosystems, driven by coordinated communities and dynamic network interactions. While existing detection models often treat social networks as static or focus narrowly on content, this study investigates the complex interplay between temporal bursts, structural persistence, and community dynamics. Utilizing a novel Misinformation Dynamics Testbed, we analyze a longitudinal Reddit dataset (2015–2024, n=1,455) across multiple analytical dimensions. Statistical and network analyses reveal that misinformation engagement follows a heavily skewed, heavy-tailed distribution (Gini  0.71), with a pronounced visibility credibility mismatch (Spearman’s p  0.32). Temporal patterns are non stationary and episodic, characterized by event driven activity surges rather than linear trends. Structurally, the network exhibits scale free properties and densification driven growth, in which stable core communities show highly volatile interconnections. Cross community analysis uncovers a hybrid propagation mechanism: localized amplification within tightly knit echo chambers, coupled with controlled diffusion through strategic bridge nodes. These findings demonstrate that misinformation operates as a self reinforcing, temporally adaptive process rather than a purely contentdriven phenomenon. Consequently, effective mitigation requires a paradigm shift from static, post-level moderation to dynamic, system level interventions. Prioritizing high centrality hubs, monitoring bridge nodes, and deploying real time burst detection are essential for disrupting propagation pathways. This research underscores the necessity of integrated, network aware frameworks to address the evolving landscape of digital misinformation. Read More Read Less

Abstract The rapid proliferation of misleading information on social media poses significant challenges to digital ecosystems, driven by sensational narratives, emotional framing, and strategic engagement tactics. Existing research often examines linguistic patterns, user engagement, and propagation dynamics in isolation, limiting comprehensive understanding. This study introduces a unified analytical framework that integrates linguistic, behavioral, and contextual features to model and classify misleading versus authentic content on Reddit.... The rapid proliferation of misleading information on social media poses significant challenges to digital ecosystems, driven by sensational narratives, emotional framing, and strategic engagement tactics. Existing research often examines linguistic patterns, user engagement, and propagation dynamics in isolation, limiting comprehensive understanding. This study introduces a unified analytical framework that integrates linguistic, behavioral, and contextual features to model and classify misleading versus authentic content on Reddit. Analyzing a dataset of 2,344 posts across twelve subreddits, we engineered 86 multidimensional features encompassing linguistic structure, sentiment, stylistic markers, clickbait indicators, and TF-IDF representations. A Linear Support Vector Machine (LinearSVC) was employed for multi-class classification across true, satire, imposter, and misleading categories. Descriptive and statistical analyses revealed that authentic content attracts the highest average engagement, while misleading posts exhibit significantly greater verbosity and clickbait prevalence, strategically mimicking credible narratives. The baseline classifier achieved 88.7% weighted accuracy, effectively distinguishing satire and imposter content, yet struggled with misleading posts due to their deliberate lexical overlap with authentic news. Inferential tests confirmed statistically significant differences in engagement metrics, content length, and clickbait usage across categories. These findings demonstrate that virality is decoupled from factual accuracy and highlight the limitations of traditional machine learning in detecting nuanced deception. Future research should incorporate transformer-based architectures, temporal engagement modeling, and network propagation analysis to enhance robust misinformation detection in dynamic online communities. Read More Read Less

Abstract Retrieval-Augmented Generation (RAG) systems have significantly improved the factual grounding of large language models (LLMs). However, challenges remain in ensuring both semantic correctness and transparent attribution, as models may still produce hallucinated or unverified outputs. This paper proposes the Double- Lock Framework, a multi-layered architecture that integrates data engineering, linguistic attribution modeling, and dual validation mechanisms to mitigate hallucinations. A high-fidelity “Gold Standard” dataset is... Retrieval-Augmented Generation (RAG) systems have significantly improved the factual grounding of large language models (LLMs). However, challenges remain in ensuring both semantic correctness and transparent attribution, as models may still produce hallucinated or unverified outputs. This paper proposes the Double- Lock Framework, a multi-layered architecture that integrates data engineering, linguistic attribution modeling, and dual validation mechanisms to mitigate hallucinations. A high-fidelity “Gold Standard” dataset is constructed using strict filtering criteria based on factual consistency and hallucination labels. The framework introduces attribution markers to enhance interpretability and employs both supervised finetuning and reward-based optimization for model training. A two-stage validation mechanism combining regex-based linguistic checks and predictive AI scoring ensures both transparency and factual accuracy. Experimental evaluation on a 6,000-sample dataset demonstrates that while regex-based validation alone results in a 7.72% hallucination leakage rate, the proposed Double-Lock mechanism achieves 0% leakage. The framework effectively bridges the gap between retrieval and human-understandable knowledge synthesis, enabling the development of reliable and trustworthy AI systems. Read More Read Less