With the rise of short-form videos, the mental impact on viewers has led to widespread consequences, prompting platforms to predict videos' impact on viewers' mental health. Subsequently, platforms can take intervention measures according to their community guidelines. Nevertheless, applicable predictive methods lack relevance to wellestablished medical knowledge, which outlines clinically proven external and environmental factors of mental disorders. To account for such medical knowledge, we resort to an emergent methodological discipline: seeded neural topic models (NTMs). However, existing seeded NTMs suffer from the limitations of single-origin topics, unknown topic sources, unclear seed supervision, and suboptimal convergence. To address those challenges, we develop a novel knowledge-guided NTM to predict a short-form video's suicidal thought impact on viewers. Extensive empirical analyses using two short-form video platforms prove that our method outperforms state-of-the-art benchmarks. Our method also discovers medically relevant topics from videos that are linked to suicidal thought impact. We contribute to information systems with a novel video analytics method that is generalizable to other video classification problems. Practically, our method can help platforms understand videos' suicidal thought impacts, thus moderating videos that violate their community guidelines.

Causal learning has garnered significant attention in recent years because it reveals the essential relationships that underpin phenomena and delineates the mechanisms by which the world evolves. Nevertheless, traditional causal learning methods face numerous challenges and limitations, including high-dimensional, unstructured variables, combinatorial optimization problems, unobserved confounders, selection biases, and estimation inaccuracies. Deep causal learning, which leverages deep neural networks, offers innovative insights and solutions for addressing these challenges. Although numerous deep learning-based methods for causal discovery and inference have been proposed, there remains a dearth of reviews examining the underlying mechanisms by which deep learning can enhance causal learning. In this article, we comprehensively review how deep learning can contribute to causal learning by tackling traditional challenges across three key dimensions: representation, discovery, and inference. We emphasize that deep causal learning is pivotal for advancing the theoretical frontiers and broadening the practical applications of causal science. We conclude by summarizing open issues and outlining potential directions for future research.

Scientific innovation increasingly depends on collaboration, yet the organizational structure that fosters breakthrough ideas remains poorly understood. Existing metrics - such as team size or compositional diversity - capture readily observable characteristics but not the deeper architecture of collaboration. We introduce Structural Diversity (SD): the extent to which a team bridges multiple distinct knowledge communities within its prior collaboration network. Using a century-scale dataset of 260 million scientific publications (1900-2025) and combining causal inference with a quasi-natural experiment based on a U.S. National Science Foundation policy change in 2012, we show that SD is a powerful and robust predictor of disruptive innovation, outperforming traditional team novelty indicators such as team freshness and edge density. Moreover, SD positively interacts with team size and is able to mitigate the well-known "curse of scale" by transforming scale from a liability into a resource for creative synthesis. We find that one mechanism underlying this effect is Disciplinary Integration (DI): teams with higher SD can more effectively combine heterogeneous knowledge into novel configurations. Our findings position SD as both a new theoretical construct and an actionable design principle for organizing scientific collaboration. By linking the architecture of team assembly to the dynamics of creative discovery, our work offers a structural explanation for how collective intelligence can be systematically engineered to foster disruptive innovation.

Current fake news detectors struggle with cross-platform generalization due to overfitting platform-specific features. We propose SHIELD, a framework that models platform-invariant interaction structures via a Social Reaction-Aware Heterogeneous Interaction Graph. Utilizing Hierarchical Attentive Aggregation, SHIELD extracts homogeneous signals from claim-reaction conflicts. Experiments on the MCFEND benchmark show SHIELD outperforms strong baselines, improving Macro-F1 by approximately 6% in unseen source detection. These results confirm that leveraging structural interaction indicators is essential for robust multi-source generalization.

Conversational recommendation systems face the fundamental challenge of autonomous dialogue control—determining when to gather information versus when to recommend. Existing approaches rely on predetermined rules or fixed dialogue patterns, lacking adaptability to diverse user communication styles. We propose ADRec (Autonomous Decision-making Recommendation agent), trained through behavior cloning from GPT-4 demonstrations. The agent learns to select among four structured actions (ask_user, call_ranker, explain, end_dialog) based on dialogue context, eliminating hand-crafted rules. By imitating expert strategies across diverse user scenarios, ADRec develops the capability to autonomously control conversation flow and adapt to different user types. Experiments on ReDial demonstrate that ADRec achieves strong recommendation success while significantly reducing dialogue length compared to fixed-strategy baselines, validating the effectiveness of learned autonomous dialogue control. The source code can be found at https://github.com/yylyylyyl/ADRec.

In this commemorative article, five past editors-in-chief of IEEE Intelligent Systems reflect on the magazine’s first 40 years and share their memories.

Personality identification plays important roles in understanding user behavior and offering foresight ability for downstream applications. The key challenge is how to address the scarcity of labeled personality data. Recently, some studies have adopted data augmentation and prompt learning to perform personality identification. However, they still heavily require a large amount of labeled data to learn an appropriate distance strategy, which limits the generalization and flexibility of the model. This study proposes a knowledge-enhanced hierarchical heterogeneous graph model, which adopts a global multi-view graph node encoding to acquire comprehensive personality features and their inherent associations, where three types of knowledge including part-of-speech (POS) tag, entity, and Linguistic Inquiry and Word Count (LIWC) are introduced. Then, a hierarchical heterogeneous graph with a “post-word-diverse knowledge” structure is constructed for each post to obtain enhanced representation. Finally, a relation guided representation optimization that considers intra-user relationships and inter-label relationships is further developed to learn more discriminative semantic representation. Experimental results on three widely used datasets demonstrate that the model outperforms state-of-the-art methods when training with only 100 samples (approximately 1% of the total data set).

Board game offers a unique platform for exploring the capabilities of artificial intelligence in decision-making. It demands long-term strategic planning and opponent behaviors to refine decision-making. Since the success of AlphaGo family, learning agents have become pivotal methods for board game. However, current learning agents rarely incorporate planning or build interactive loops with opponents' behaviors in decision-making. This paper proposes a novel planning-based model (BG-Planner) for strategic decision-making and long-term planning in board game. We propose a Graphplan-style network with alternating action and proposition layers to predict actions and assess wining rate. Further, an opponent modeling strategy is incorporated to predict opponent behaviors, assist decision-making and reduce competitive uncertainty. We also introduce a knowledge-based search tactic to enhance BG-Planner’s learning. Experimental results demonstrate that BG-Planner enhances the quality and efficiency of decision-making in the Gomoku game. It shows potential to improve deep planning strategies in decision-making intelligence.

Large language models (LLMs) represent a novel technological species in the realm of general intelligence. Their problem-solving approach is not based on “first principles” (logocentrism) but rather on empirical learning from observed data. LLMs possess the ability to extract intuitive knowledge from vast amounts of data, enabling them to offer flexible and effective solutions in the face of complex and dynamic scenarios. The general intelligence characteristics of LLMs are mainly reflected in three aspects: technologically, they exhibit common sense, deep reasoning, strong generalization, and natural human-computer interaction; in terms of intelligence, they demonstrate memory-driven core features, powerful data-driven learning capabilities, and exceptional generalization abilities; in terms of thought, they possess highly humanlike cognitive traits such as contextual understanding, analogy, and intuitive reasoning. These capabilities collectively suggest that LLMs can adapt to a wide range of complex, open-ended scenarios, presenting a stark contrast to traditional models that emphasize formal logic, quantitative analysis, and narrowly defined problem structures. As such, the rise of LLMs is likely to drive significant shifts in AI theory and application, potentially redefining how intelligent systems approach decision-making, strategic reasoning, and contextual understanding in uncertain and dynamic environments.

Large Language Models (LLMs) have demonstrated transformative potential in reshaping the world. As these models are pretrained on general corpora, they often require domain-specific fine-tuning to optimize performance in specialized business applications. Due to their massive scale, parameter-efficient fine-tuning (PEFT) methods are widely used to reduce training costs. Among them, hybrid PEFT methods that combine multiple PEFT techniques have achieved the best performance. However, existing hybrid PEFT methods face two main challenges when fine-tuning LLMs for specialized applications: (1) relying on point estimates, lacking the ability to quantify uncertainty for reliable decision-making, and (2) struggling to dynamically adapt to emerging data, lacking the ability to suit real-world situations. We propose Bayesian Hybrid Parameter-Efficient Fine-Tuning (BH-PEFT), a novel method that integrates Bayesian learning into hybrid PEFT. BH-PEFT combines Adapter, LoRA, and prefix-tuning to fine-tune feedforward and attention layers of the Transformer. By modeling learnable parameters as distributions, BH-PEFT enables uncertainty quantification. We further propose a Bayesian dynamic fine-tuning approach where the last posterior serves as the prior for the next round, enabling effective adaptation to new data. We evaluated BH-PEFT on business tasks such as sentiment analysis, news categorization, and commonsense reasoning. Results show that our method outperforms existing PEFT baselines, enables uncertainty quantification for more reliable decisions, and improves adaptability in dynamic scenarios. This work contributes to business analytics and data science by proposing a novel BH-PEFT method and dynamic fine-tuning approach that support uncertainty-aware and adaptive decision-making in real-world situations.

Large language models (LLMs) have achieved significant progress in mathematical reasoning, especially in elementary math. However, they remain indisposed on tackling complex questions at high-school or college levels, which put forward a more advanced requirement of mastering relevant mathematical theorems. For we humans, whether selecting the appropriate theorems according to the provided question is a crucial factor affecting the quality of the ultimate solutions, yet which has been neglected by previous research in the field of LLM reasoning. In this paper, we propose a novel approach to enhance the LLM's capability of utilizing the mathematical theorems to specific problems, which we refer to as Theorem Rationale (TR). To this end, a new dataset encompassing problem-theorem-solution triples is deliberately established for transferring principles of TR. Furthermore, we develop an evolving strategy to boost hierarchical instructions oriented on the theorems to alleviate difficulty in acquiring the curated data and facilitate the digestion of theorem application from various perspectives. Evaluations on a wide range of public datasets exhibit that the model fine-tuned with our dataset achieves consistent improvements at varying mathematical levels compared to the backbone. And further ablation studies illustrate the effectiveness of our proposed evolutionary strategies on enhancing the model's capability of math problem-solving. Overall, extensive experiments reveal the potential of our proposed method which highlights the significance of aligning the problems with the concrete theorems for LLMs to alleviate hallucination and improve the models' mathematical reasoning capabilities.

Planning is a central concern in artificial intelligence research, which automatically generates sequences of actions to achieve goals for many fields such as logistics and robotics. Recently, neural networks have gained significant attention in the planning domain due to their powerful representation and learning capabilities. However, current neural planners often rely on extensive training data and lack sufficient feedback mechanisms, resulting in limited adaptability to complex planning problems. This paper introduces a Symbolic Incubator for Training Planners (SITP), designed to autonomously and efficiently acquire planning capabilities through the integration of advanced recognition-based feedback mechanisms. SITP leverages a symbol recognizer to establish the feedback mechanism that integrates the strengths of neural and symbolic systems, enhancing planning capabilities and enabling learning from limited data. To enhance flexibility and generality, we design three distinct training mechanisms that demonstrate the superior planning capabilities of SITP at different training stages and adapt to the specific requirements of different planning tasks. Empirical evaluations demonstrate the remarkable performance of SITP across various baseline planners and tasks. The proposed SITP is an effective incubator, featuring advanced feedback and flexible training mechanisms to enhance planner capabilities. Codes and data will be released upon acceptance.

Positional bias (PB), manifesting as non-uniform sensitivity across different contextual locations, significantly impairs long-context comprehension and processing capabilities. Previous studies have addressed PB either by modifying the underlying architectures or by employing extensive contextual awareness training. However, the former approach fails to effectively eliminate the substantialperformance disparities, while the latter imposes significant data and computational overhead. To address PB effectively, we introduce Pos2Distill, a position to position knowledge distillation framework. Pos2Distill transfers the superior capabilities from advantageous positions to less favorable ones, thereby reducing the huge performance gaps. The conceptual principle is to leverage the inherent, position-induced disparity to counteract the PB itself. We identify distinct manifestations of PB under retrieval and reasoning paradigms, thereby designing two specialized instantiations: Pos2Distill-R1 and Pos2Distill-R2 respectively, both grounded in this core principle. By employing the Pos2Distill approach, we achieve enhanced uniformity and significant performance gains across all contextual positions in long-context retrieval and reasoning tasks. Crucially, both specialized systems exhibit strong cross-task generalization mutually, while achieving superior performance on their respective tasks.

Board game offers a unique platform for exploring the capabilities of artificial intelligence in decision-making. It demands long-term strategic planning and opponent behaviors to refine decision-making. Since the success of AlphaGo family, learning agents have become pivotal methods for board game. However, current learning agents rarely incorporate planners or build interactive loops with opponents’ behaviors in decision-making. This paper proposes a novel planning-based model (BG-Planner) for strategic decision-making and long-term planning in board game. We propose a Graphplan-style network with alternating action and proposition layers to predict actions and assess wining rate. Further, an opponent modeling strategy is incorporated to predict opponent behaviors, assist decision-making and reduce competitive uncertainty. We also introduce a knowledge-based search tactic to enhance BG-Planner’s learning. Experimental results demonstrate that BG-Planner enhances the quality and efficiency of decision-making in the Gomoku game. It shows potential to improve deep planning strategies in decision-making intelligence.

Introduction Return-to-acute-care metrics, such as early emergency department (ED) visits, are key indicators of healthcare quality, with ED returns following surgery often considered avoidable and costly events. Proactively identifying patients at high risk of ED return can support quality improvement efforts, allowing interventions to target vulnerable patients. With its predictive capabilities, machine learning (ML) has shown potential in forecasting various clinical outcomes but remains underutilised in orthopaedic trauma. This study uses a random forest model to predict 90-day ED return in orthopaedic trauma patients, aiming to identify high-risk individuals and elucidate risk factors associated with returns. This study hypothesised that a highly accurate model could be developed to predict patients at high risk of ED return within 90 days of surgery.Purpose To develop and validate an ML model that predicts 90-day ED returns after orthopaedic trauma surgery using input data readily available in the electronic health record.Methods This is a retrospective model development and validation study. The study used data from a registry that includes information on all orthopaedic surgeries conducted at a level 1 academic medical centre. Patients who underwent orthopaedic trauma between 1 January 2017 and 1 March 2023 were identified using common procedural terminology code. The model used demographic, comorbid and perioperative variables. Return to the ED was captured as a binary outcome. Model performance was evaluated using the area under the receiver operator curve (AUROC).Results A total of 12 069 patients met the inclusion criteria. Patients were predominantly female (53%) and white (70%), with a median age of 55. The 90-day ED return rate was 14% (table 1). The random forest model identified body mass index, distance from the patient's residence to the hospital, age, length of hospital stay and complexity of procedure (work relative value unit) as significant predictors of ED return, each accounting for greater than 10% of the total importance across all features in the model (table 2). Further, the model displayed strong discrimination of patients returning to the ED (AUROC=0.74) (figure 1).Conclusions The random forest model demonstrated predictive discrimination of 90-day ED returns. Critical predictors such as patient distance from the hospital suggest considering geographical and socioeconomic factors in postdischarge care planning. Operational factors such as length of stay or complexity of the procedure also predicted return to the ED. The study lays the groundwork for future predictive models in clinical decision-making and healthcare resource utilisation.Level of evidence Level III, retrospective model development and validation study.

Spatiotemporal Graph Learning (SGL) under Zero-Inflated Distribution (ZID) is crucial for urban risk management tasks, including crime prediction and traffic accident profiling. However, SGL models are vulnerable to adversarial attacks, compromising their practical utility. While adversarial training (AT) has been widely used to bolster model robustness, our study finds that traditional AT exacerbates performance disparities between majority and minority classes under ZID, potentially leading to irreparable losses due to underreporting critical risk events. In this paper, we first demonstrate the smaller top-k gradients and lower separability of minority class are key factors contributing to this disparity. To address these issues, we propose MinGRE, a framework for Minority Class Gradients and Representations Enhancement. MinGRE employs a multi-dimensional attention mechanism to reweight spatiotemporal gradients, minimizing the gradient distribution discrepancies across classes. Additionally, we introduce an uncertainty-guided contrastive loss to improve the inter-class separability and intra-class compactness of minority representations with higher uncertainty. Extensive experiments demonstrate that the MinGRE framework not only significantly reduces the performance disparity across classes but also achieves enhanced robustness compared to existing baselines. These findings underscore the potential of our method in fostering the development of more equitable and robust models.

The discovery of high-performance materials is crucial for technological advancement. Inverse design using multi-agent systems (MAS) shows great potential for new material discovery. However, current MAS for materials research rely on predefined configurations and tools, limiting their adaptability and scalability. To address these limitations, we developed a planner driven multi-agent system (S1-MatAgent) which adopts a Planner-Executor architecture. Planner automatically decomposes complex materials design tasks, dynamically configures various tools to generate dedicated Executor agents for each subtask, significantly reducing reliance on manual workflow construction and specialized configuration. Applied to high-entropy alloy catalysts for hydrogen evolution reactions in alkaline conditions, S1-MatAgent completed full-cycle closed-loop design from literature analysis and composition recommendation to performance optimization and experimental validation. To tackle the deviations between designed materials and target, as well as high experimental verification costs, S1-MatAgent employs a novel composition optimization algorithm based on gradients of machine learning interatomic potential, achieving 27.7

Knowledge reasoning has been widely researched in knowledge graphs (KGs), but there has been relatively less research on hyper-relational KGs, which also plays an important role in downstream tasks. Existing reasoning methods on hyper-relational KGs are based on representation learning. Though this approach is effective, it lacks interpretability and ignores the graph structure information. In this paper, we make the first attempt at symbolic reasoning on hyper-relational KGs. We introduce rule extraction methods based on both individual facts and paths, and propose a rule-based symbolic reasoning approach, HyperPath. This approach is simple and interpretable, it can serve as a baseline model for symbolic reasoning in hyper-relational KGs. We provide experimental results on almost all datasets, including five large-scale datasets and seven sub-datasets of them. Experiments show that the expressive power of the proposed model is similar to simple neural networks like convolutional networks, but not as advanced as more complex networks such as Transformer and graph convolutional networks, which is consistent with the performance of symbolic methods on KGs. Furthermore, we also analyze the impact of rule length and hyperparameters on the model's performance, which can provide insights for future research in hypergraph symbolic reasoning.

Enterprise performance in real-world markets is shaped by dynamic factors, including competitors, collaborators, and hidden associates. Existing models struggle to capture the interplay between time-varying network dynamics and financial asset price movements. Traditional energy cascading models rely on static network assumptions, while deep learning approaches lack the incorporation of key network science principles such as structural balance and assortativity degree. To address these gaps, we propose the Dynamic Energy Cascading Model (DECM), a framework that models the propagation of business influence within dynamic enterprise networks. This method first constructs a dynamic enterprise association network, then applies an energy cascading mechanism to this network, utilizing the propagated energy metrics as predictive indicators for stock price forecasting. Unlike existing approaches, DECM uniquely integrates dynamic network properties and knowledge structures, such as structural balance and assortativity degree, to model the cascading effects of business influences on stock prices. Through extensive evaluations using data from S&P 500 companies, we demonstrate that DECM significantly outperforms conventional models in predictive precision. A key innovation of our work lies in identifying the critical role of assortativity degree in predicting stock price movements, which surpasses the impact of structural balance. These findings not only advance the theoretical understanding of enterprise performance dynamics but also provide actionable insights for policymakers and practitioners from a network science perspective.

Social media platforms has long served as open arenas where individuals discuss and change their opinions on various events, subsequently influencing the progression of these events. Public opinion, recognized as an important social signal, is instrumental in understanding the developmental patterns of social events and in guiding more informed responses. In light of this, we propose a sociologically-informed opinion prediction model, which integrates rich social interaction data with time series forecasting techniques using a graph neural network framework. This model, enriched by a sociological theoretical model, reflects the real-world dynamics of opinion evolution. Experimental results derived from three synthetic datasets and two real-world datasets indicate that incorporating user interaction data, along with more effective utilization of historical information, has led to a large improvement in the accuracy of opinion predictions. The source code and sample data for our study are available at https://github.com/RiikkaYang/SIGNN.