Urban region embedding, which learns dense vector representations for urban zones, plays a foundational role in data-driven urban intelligence. These representations are critical for downstream applications like public safety management and infrastructure development, requiring nuanced understanding of urban functionality. A core challenge remains effective fusion of multi-view data (e.g., human mobility flows and static regional attributes) into unified zone representations. To this end, we propose MVJC, a Multi-view Joint Learning and Contrastive Learning framework, which employs: (1) Multi-view Joint Learning (MVJL) layer to model intra-view dependencies to extract view-specific features and (2) Multi-view Contrastive Learning (MVCL) layer to perform cross-region aggregation to derive consensus representations while capturing the regional complementarity. We further introduce a structure-aware contrastive loss that mitigates false negatives by aligning representations through region topology instead of instance identity. Extensive experiments on New York City datasets demonstrate MVJC’s superiority: it reduces crime prediction MAE by 9.1% (vs. 66.9 baseline) and improves land use clustering F-measure by 55.6% (vs. 0.45 baseline) over state-of-the-art method, which is attributed to MVJC’s synergy of joint and contrastive learning, yielding representations that are simultaneously predictive and semantically discriminative.

Large Language Models (LLMs) gain substantial reasoning and decision-making capabilities from thought structures. However, existing methods such as Tree of Thought and Retrieval Augmented Thoughts often fall short in complex tasks due to the limitations of insufficient local retrieval of factual knowledge and inadequate global selection of strategies. These limitations make it challenging for these methods to balance factual accuracy and comprehensive logical optimization effectively. To address these limitations, we introduce the Retrieval Augmented Thought Tree (RATT), a novel thought structure that considers both overall logical soundness and factual correctness at each step of the thinking process. Specifically, at every point of a thought branch, RATT performs planning and lookahead to explore and evaluate multiple potential reasoning steps, and integrate the fact-checking ability of Retrieval-Augmented Generation (RAG) with LLM's ability to assess overall strategy. Through this combination of factual knowledge and strategic feasibility, the RATT adjusts and integrates the thought tree structure to search for the most promising branches within the search space. This thought structure significantly enhances the model's coherence in logical inference and efficiency in decision-making, and thus increases the limit of the capacity of LLM to generate reliable inferences and decisions based on thought structures. A broad range of experiments on different types of tasks showcases that the RATT structure significantly outperforms existing methods in factual correctness and logical coherence.

In the rapidly developing field of automatic text generation and understanding, the quality of input data has been shown to be a key factor affecting the efficiency and accuracy of large language model (LLM) output. With the advent of advanced tools such as ChatGPT, input refinement work has mainly focused on prompt engineering. However, existing methods are often too dependent on specific contexts and are easily affected by individual expert experience and potential biases, limiting their wide applicability in diverse real-world applications. To address this problem, this study develops an Reinforced Token-Level Input Refinement, called RTLIR. We choose to optimize the input data at the fine-grained level of tokens, cleverly preserving the original text structure. Operationally, each state is defined by the token set of the current text, and each action is a binary decision process to decide whether to retain a specific token information. The agent automatically calculates and determines the selection probability of each token based on the current state, thereby optimizing the entire decision process. Through continuous exploration and learning, the agent can autonomously learn to identify the key inputs that have the greatest impact on the generation results and achieve refinement of the input data. In addition, RTLIR is a plug-and-play, LLM-agnostic module that can be used for a wide range of tasks and models. Experimental results show that RTLIR improves the performance of LLM in various input scenarios and tasks, with an average accuracy increase of 6%.

Autonomous underwater vehicle (AUV) is crucial for marine applications such as ocean data collection, pollution monitoring, and navigation. However, their limited energy resources constrain their operational duration, posing a significant challenge for long-term operations. Due to the complex and unpredictable nature of the underwater environment, AUVs allocate energy to their sensing systems to sense the surrounding environment and avoid obstacles. Existing methods focus on reducing energy consumption on AUV computing and movement, neglecting sensing energy consumption and few attempts have been made to balance the AUV energy and sensing ability with a flexible sensing system. Along these lines, we consider both AUV energy consumption and flexible sensing abilities, and propose a deep reinforcement learning-based method to Reduce Energy Consumption by AUV Sensing system (RECS). Specifically, we build an AUV sensing system in a 2-dimension space, with controllable 8-direction sensing abilities to collect the environment information dynamically. Then we divide the underwater environment into several areas and assign weights on the edges of areas based on the AUV planned path. Additionally, we dynamically switch the sensors in different directions and radii to sense the edges of the area where the AUV is located. The Artificial Potential Field (APF) method is employed to re-plan the AUV path to avoid obstacles and reach the target point effectively. Experimental results demonstrate that compared to full sensors on, our method reduces energy consumption by 53.48% and is capable of generalizing to varying environments and varying sensing system radii.

We present UrbanXplain, a language-driven urban planning system that combines real-time 3D rendering with explainable reasoning. UrbanXplain enables an integrated planning workflow driven entirely by natural language input. This includes steps from functional zoning to land use implementation. The system uses a large language models (LLMs) to perform spatial inference. It converts high-level planning goals into structured zoning commands, assigns building functions such as residential, commercial, or cultural, and creates layout plans that respect constraints like height, material, and accessibility. A Unity3D-based simulator renders the output design in real time, allowing users to explore and interact with the results. A key feature of UrbanXplain is its support for reasoning traceability. For each decision, the system displays its semantic parsing, zoning logic, and siting justifications. This ensures transparency and supports iterative refinement. We evaluate UrbanX-plain in three scenario-based experiments: 15 minute city design, green infrastructure planning, and energy efficient mixed-use layouts. These cases show LLMs supporting interpretable, adaptive, and cognitively accessible urban planning.

Multivariate time series forecasting in practical deployment faces a critical challenge termed Variable Subset Forecasting (VSF), where certain variables accessible during training are entirely missing during inference. This creates a stark discrepancy between the training (source domain with full variables) and inference (target domain with partial variables) environments, disrupting cross-variable dependencies and fragmenting global temporal patterns. Existing imputation methods, limited to transferring local knowledge (e.g., temporal neighbors or pairwise correlations), fail to capture essential global dynamics, leading to severe performance degradation under distribution shifts. To address these challenges, we redefine VSF as a cross-domain knowledge transfer problem and propose VIDA, a framework that systematically transfers Variable Invariant knowledge from complete to partial observations through Domain Adaptation. Key to our approach is (1) Global time-frequency joint representation learning, which encodes temporal dynamics via dilated convolutions and captures low-frequency spectral consistency using Fourier neural operators, and (2) Sinkhorn-regularized distribution alignment to bridge non-overlapping feature supports across domains via optimal transport. Unlike imputation-first methods, VIDA enforces task-driven consistency by jointly optimizing predictions on reconstructed and original data, ensuring the transferred knowledge directly enhances forecasting robustness. Extensive experiments across four real-world datasets show that VIDA outperforms state-of-the-art imputation methods by 25% on average with partially observed variables. This work establishes a new paradigm for variable-missing scenarios by unifying imputation and forecasting through principled knowledge transfer.

Evaluation of Massive Open Online Course (MOOC) quality is crucial to enhance the educational resources, benefiting user services, and enhancing students’ learning efficiency. Despite achieving encouraging results, current efforts are hindered by complex relationships between entities and individual varies. To address the above problem, in this article, we frame the issue as a task of learning course representations and proceed to develop an U ser-Centric H ypergraph R epresentation L earning ( UHRL ) for online course quality evaluation. In particular, we initially construct a MOOC hypergraph to depict the interactions and connections between the entities and use cross-hyperedge alignment to reveal the semantics of courses. And then we incorporate an attention mechanism in the information transmission process to ensure semantic integrity. Furthermore, to tackle the bias of users’ preference, our framework exploits mutual information for preserving the fairness of representation learning. Finally, our comprehensive experiments on three real-world datasets confirm the effectiveness of our approach compared to cutting-edge methods in evaluating online course quality across various performance metrics.

Variable Subset Forecasting (VSF) refers to a unique scenario in multivariate time series forecasting, where available variables in the inference phase are only a subset of the variables in the training phase. VSF presents significant challenges as the entire time series may be missing, and neither inter- nor intra-variable correlations persist. Such conditions impede the effectiveness of traditional imputation methods, primarily focusing on filling in individual missing data points. Inspired by the principle of feature engineering that not all variables contribute positively to forecasting, we proposeTask-OrientedImputation forVSF(TOI-VSF), a novel framework shifts the focus from accurate data recovery to directly support the downstream forecasting task. TOI-VSF incorporates a self-supervised imputation module, agnostic to the forecasting model, designed to fill in missing variables while preserving the vital characteristics and temporal patterns of time series data. Additionally, we implement a joint learning strategy for imputation and forecasting, ensuring that the imputation process is directly aligned with and beneficial to the forecasting objective. Extensive experiments across four datasets demonstrate the superiority of TOI-VSF, outperforming baseline methods by 15% on average.

In the realm of human mobility, the decision-making process for selecting the next-visit location is intricately influenced by a trade-off between spatial and temporal constraints, which are reflective of individual needs and preferences. This trade-off, however, varies across individuals, making the modeling of these spatial-temporal dynamics a formidable challenge. To address the problem, in this work, we introduce the "Spatial-temporal Induced Hierarchical Reinforcement Learning" (STI-HRL) framework, for capturing the interplay between spatial and temporal factors in human mobility decision-making. Specifically, STI-HRL employs a two-tiered decision-making process: the low-level focuses on disentangling spatial and temporal preferences using dedicated agents, while the high-level integrates these considerations to finalize the decision. To complement the hierarchical decision setting, we construct a hypergraph to organize historical data, encapsulating the multi-aspect semantics of human mobility. We propose a cross-channel hypergraph embedding module to learn the representations as the states to facilitate the decision-making cycle. Our extensive experiments on two real-world datasets validate the superiority of STI-HRL over state-of-the-art methods in predicting users' next visits across various performance metrics.

University Course Timetabling Problem (UCTP) is a significant resource allocation challenge with NP-hard characteristics. As problem sizes increase, finding an optimal solution becomes increasingly complex. To address this, we propose an automated planning method using Reinforcement Learning (RL), which treats UCTP as a series of dynamic decision-making tasks. The RL agent acts as an automated planner, operating in a simulated environment that reflects the complexity and constraints of a university course. It adapts to changes in the timetable by evaluating the outcomes of its actions, continuously refining its strategy to identify and implement the most effective actions. During the explore phase, the agent tests new strategies using a priority-driven reward system and an experience replay mechanism. This approach replays high-value actions, enhancing solution diversity and reducing the search space. In the exploit phase, the agent leverages its accumulated knowledge to apply proven effective strategies. It also incorporates unbiased external evaluations to avoid local optima and ensure global optimization of its strategy. Extensive testing on multiple real-world datasets shows that our automated planning framework consistently outperforms existing methods in various complex scheduling scenarios. This confirms its effectiveness and practicality in meeting diverse scheduling needs.

In this paper, we study the problem of MOOC quality evaluation that is essential for improving the course materials, promoting students' learning efficiency, and benefiting user services. While achieving promising performances, current works still suffer from the complicated interactions and relationships of entities in MOOC platforms. To tackle the challenges, we formulate the problem as a course representation learning task based, and develop an Information-aware Graph Representation Learning(IaGRL) for multi-view MOOC quality evaluation. Specifically, We first build a MOOC Heterogeneous Network (HIN) to represent the interactions and relationships among entities in MOOC platforms. And then we decompose the MOOC HIN into multiple single-relation graphs based on meta-paths to depict multi-view semantics of courses. The course representation learning can be further converted to a multi-view graph representation task. Different from traditional graph representation learning, the learned course representations are expected to match the following three types of validity: (1) the agreement on expressiveness between the raw course portfolio and the learned course representations; (2) the consistency between the representations in each view and the unified representations; (3) the alignment between the course and MOOC platform representations. Therefore, we propose to exploit mutual information for preserving the validity of course representations. We conduct extensive experiments over real-world MOOC datasets to demonstrate the effectiveness of our proposed method.

In this article, we study knowledge concept recommendation in Massive Open Online Courses (MOOCs) in an explainable manner. Knowledge concepts, composing course units (e.g., videos) in MOOCs, refer to topics and skills that students are expected to master. Compared to traditional course recommendation in MOOCs, knowledge concepts recommendation has drawn more attention because students’ interests over knowledge concepts can better revealstudents’ real intention in a more refined granularity. However, there are three unique challenges in knowledge concept recommendation: (1) How to design an appropriate data structure to capture complex relationships between knowledge concepts, course units, and other participants (e.g., students, teachers)? (2) How to model interactions between students and knowledge concepts? (3) How to make explainable recommendation results to students? To tackle these challenges, we formulate the knowledge concept recommendation as a reinforcement learning task integrated with MOOC knowledge graph (KG). Specifically, we first construct MOOC KG as the environment to capture all the relationships and behavioral histories by considering all the entities (e.g., students, teachers, videos, courses, and knowledge concepts) on the MOOC provider. Then, to model the interactions between students and knowledge concepts, we train an agent to mimic students’ learning behavioral patterns facing the complex environment. Moreover, to provide explainable recommendation results, we generate recommended knowledge concepts in the format of a path from MOOC KG to indicate semantic reasons. Finally, we conduct extensive experiments on a real-world MOOC dataset to demonstrate the effectiveness of our proposed method.

Accurate identification of drug-target interactions (DTIs) plays a crucial role in drug discovery. Compared with traditional experimental methods that are labor-intensive and time-consuming, computational methods are more and more popular in recent years. Conventional computational methods almost simply view heterogeneous networks which integrate diverse drug-related and target-related dataset instead of fully exploring drug and target similarities. In this paper, we propose a new method, named DTIHNC, for $\mathbf{D}$rug-$\mathbf{T}$arget $\mathbf{I}$nteraction identification, which integrates $\mathbf{H}$eterogeneous $\mathbf{N}$etworks and $\mathbf{C}$ross-modal similarities calculated by relations between drugs, proteins, diseases and side effects. Firstly, the low-dimensional features of drugs, proteins, diseases and side effects are obtained from original features by a denoising autoencoder. Then, we construct a heterogeneous network across drug, protein, disease and side-effect nodes. In heterogeneous network, we exploit the heterogeneous graph attention operations to update the embedding of a node based on information in its 1-hop neighbors, and for multi-hop neighbor information, we propose random walk with restart aware graph attention to integrate more information through a larger neighborhood region. Next, we calculate cross-modal drug and protein similarities from cross-scale relations between drugs, proteins, diseases and side effects. Finally, a multiple-layer convolutional neural network deeply integrates similarity information of drugs and proteins with the embedding features obtained from heterogeneous graph attention network. Experiments have demonstrated its effectiveness and better performance than state-of-the-art methods. Datasets and a stand-alone package are provided on Github with website https://github.com/ningq669/DTIHNC.

Accurate identification of drug-target interactions (DTIs) play a crucial role in drug discovery. Conventional computational methods almost simply view heterogeneous networks which integrate diverse drug-related and target-related dataset instead of fully explored drug and protein similarities. In this paper, we propose a new method, named HGSDTI. Firstly, the low-dimensional features of drugs, proteins, diseases and side-effects are obtained by a denoising autoencoder. Then, we construct a heterogeneous network across drug, protein, disease and side-effect nodes, and a three-layer graph convolutional network (GCN) is applied to learn the neighbor topology information and integrate the low-dimensional features of nodes. Next, we calculate multi-modal drug similarities and protein similarities from multi-scale relations between drugs, proteins, diseases and side-effects. Finally, a multiple-layer convolutional neural network (CNN) deeply integrate similarity information of drugs and proteins with the neighbor topology information. Experiments have demonstrated its effectiveness and better performance than state-of-the-art methods.

The COVID-19 pandemic forces schools to move teaching online and stimulates the development of online tutoring platforms.Although online tutoring platforms provide students the access to learning materials and tools anytime and anywhere, the quality of studies is impeded by the fact that students learn by watching videos, which lacks interactions between teachers and students.Such dilemma prevents us from respectively understanding and improving the online learning patterns and efficiency of students.To achieve this goal, we need to solve three challenges: (1) How can we quantify the study quality of online learning?(2) How can we design an appropriate data structure to describe online study behaviors?(3) How can we model the online study behaviors to better mine online study patterns?To address the challenges, we first propose a new measurement to quantify the online study quality from the perspective of study engagement.We then define a study behavior sequence to describe online study behaviors.The study behavior at each timestamp is an event of a video lecture watching behavior type, such as, watching, dragging forward and dragging backward.Moreover, we develop a neural hawkes process framework (namely EduHawkes ) for online study behavior modeling.The EduHawkes is a novel hierarchical encode-decode architecture with simultaneously optimizing the study behavior prediction task (event-level) and the study quality prediction task (course-level).In the experiments, we apply EduHawkes to the applications of study quality prediction and flippant student identification in order to demonstrate the improved performances of our proposed method on modeling online study behaviors.

Social recommendation tasks exploit social connections to enhance recommendation performance. To fully utilize each user's first-order and high-order neighborhood preferences, recent approaches incorporate influence diffusion process for better user preference modeling. Despite the superior performance of these models, they either neglect the latent individual interests hidden in the user-item interactions or rely on computationally expensive graph attention models to uncover the item-induced sub-relations, which essentially determine the influence propagation passages. Considering the sparse substructures are derived from original social network, we name them as partial relationships between users. We argue such relationships can be directly modeled such that both personal interests and shared interests can propagate along a few channels (or dimensions) of latent users' embeddings. To this end, we propose a partial relationship aware influence diffusion structure via a computationally efficient multi-channel encoding scheme. Specifically, the encoding scheme first simplifies graph attention operation based on a channel-wise sparsity assumption, and then adds an InfluenceNorm function to maintain such sparsity. Moreover, ChannelNorm is designed to alleviate the oversmoothing problem in graph neural network models. Extensive experiments on two benchmark datasets show that our method is comparable to state-of-the-art graph attention-based social recommendation models while capturing user interests according to partial relationships more efficiently.

Feature selection aims to select a subset of features to optimize the performances of downstream predictive tasks. Recently, multi-agent reinforced feature selection (MARFS) has been introduced to automate feature selection, by creating agents for each feature to select or deselect corresponding features. Although MARFS enjoys the automation of the selection process, MARFS suffers from not just the data complexity in terms of contents and dimensionality, but also the exponentially-increasing computational costs with regard to the number of agents. The raised concern leads to a new research question: Can we simplify the selection process of agents under reinforcement learning context so as to improve the efficiency and costs of feature selection? To address the question, we develop a single-agent reinforced feature selection approach integrated with restructured choice strategy. Specifically, the restructured choice strategy includes: 1) we exploit only one single agent to handle the selection task of multiple features, instead of using multiple agents. 2) we develop a scanning method to empower the single agent to make multiple selection/deselection decisions in each round of scanning. 3) we exploit the relevance to predictive labels of features to prioritize the scanning orders of the agent for multiple features. 4) we propose a convolutional auto-encoder algorithm, integrated with the encoded index information of features, to improve state representation. 5) we design a reward scheme that take into account both prediction accuracy and feature redundancy to facilitate the exploration process. Finally, we present extensive experimental results to demonstrate the efficiency and effectiveness of the proposed method.

We study the integration of reinforcement learning and spatial knowledge graph for incremental mobile user profiling, which aims to map mobile users to dynamically-updated profile vectors by incremental learning from a mixed-user event stream. After exploring many profiling methods, we identify a new imitation based criteria to better evaluate and optimize profiling accuracy. Considering the objective of teaching an autonomous agent to imitate a mobile user to plan next-visit based on the user's profile, the user profile is the most accurate when the agent can perfectly mimic the activity patterns of the user. We propose to formulate the problem into a reinforcement learning task, where an agent is a next-visit planner, an action is a POI that a user will visit next, and the state of environment is a fused representation of a user and spatial entities (e.g., POIs, activity types, functional zones). An event that a user takes an action to visit a POI, will change the environment, resulting into a new state of user profiles and spatial entities, which helps the agent to predict next visit more accurately. After analyzing such interactions among events, users, and spatial entities, we identify (1)semantic connectivity among spatial entities, and, thus, introduce a spatial Knowledge Graph (KG) to characterize the semantics of user visits over connected locations, activities, and zones. Besides, we identify (2) mutual influence between users and the spatial KG, and, thus, develop a mutual-updating strategy between users and the spatial KG, mixed with temporal context, to quantify the state representation that evolves over time. Along these lines, we develop a reinforcement learning framework integrated with spatial KG. The proposed framework can achieve incremental learning in multi-user profiling given a mixed-user event stream. Finally, we apply our approach to human mobility activity prediction and present extensive experiments to demonstrate improved performances.