Precise user and item embedding learning is the key to building a successful recommender system. Traditionally, Collaborative Filtering (CF) provides way learn embeddings from user-item interaction history. However, performance limited due sparseness of behavior data. With emergence online social networks, systems have been proposed utilize each user's local neighbors' preferences alleviate data sparsity for better modeling. We argue that, platform, her potential influenced by trusted users,...

The problem of point interest (POI) recommendation is to provide personalized recommendations places interests, such as restaurants, for mobile users. Due its complexity and connection location based social networks (LBSNs), the decision process a user choose POI complex can be influenced by various factors, preferences, geographical influences, mobility behaviors. While there are some studies on recommendations, it lacks integrated analysis joint effect multiple factors. To this end, in...

The problem of point interest (POI) recommendation is to provide personalized recommendations places, such as restaurants and movie theaters. increasing prevalence mobile devices location based social networks (LBSNs) poses significant new opportunities well challenges, which we address. decision process for a user choose POI complex can be influenced by numerous factors, personal preferences, geographical considerations, mobility behaviors. This further complicated the connection LBSNs...

Dynamic Graph Neural Networks (DGNNs) have become one of the most promising methods for traffic speed forecasting. However, when adapting DGNNs forecasting, existing approaches are usually built on a static adjacency matrix (no matter predefined or self-learned) to learn spatial relationships among different road segments, even if impact two segments can be changeable dynamically during day. Moreover, future cannot only related with current speed, but also affected by other factors such as...

Graph Convolutional Network (GCN) has been widely applied in transportation demand prediction due to its excellent ability capture non-Euclidean spatial dependence among station-level or regional demands. However, most of the existing research, graph convolution was implemented on a heuristically generated adjacency matrix, which could neither reflect real relationships stations accurately, nor multi-level demands adaptively. To cope with above problems, this paper provides novel...

Taxi and sharing bike bring great convenience to urban transportation. A lot of efforts have been made improve the efficiency taxi service or system by predicting next-period pick-up drop-off demand. Different from existing research, this paper is motivated following two facts: 1) From a micro view, an observed spatial demand at any time slot could be decomposed as combination many hidden bases; 2) macro multiple transportation demands are strongly correlated with each other, both spatially...

Bike sharing systems, aiming at providing the missing links in public transportation are becoming popular urban cities. In an ideal bike network, station locations usually selected a way that there balanced pick-ups and drop-offs among stations. This can help avoid expensive re-balancing operations maintain high user satisfaction. However, it is challenging task to develop such efficient system with appropriate locations. Indeed, demand influenced by multiple factors of surrounding...

The rapid adoption of mobile messaging Apps has enabled us to collect massive amount encrypted Internet traffic messaging. classification this into different types in-App service usages can help for intelligent network management, such as managing bandwidth budget and providing quality services. Traditional approaches rely on packet inspection, parsing HTTP headers. However, are increasingly using secure protocols, HTTPS SSL, transmit data. This imposes significant challenges the...

Point of interest (POI) recommendation, which provides personalized recommendation places to mobile users, is an important task in location-based social networks (LBSNs). However, quite different from traditional interest-oriented merchandise POI more complex due the timing effects: we need examine whether fits a user's availability. While there are some prior studies included temporal effect into recommendations, they overlooked compatibility between time-varying popularity POIs and regular...

Image based social networks are among the most popular networking services in recent years. With a tremendous amount of images uploaded everyday, understanding users' preferences on user-generated and making recommendations have become an urgent need. In fact, many hybrid models been proposed to fuse various kinds side information (e.g., image visual representation, network) user-item historical behavior for enhancing recommendation performance. However, due unique characteristics user...

Urban functions refer to the purposes of land use in cities where each zone plays a distinct role and cooperates with other serve people’s various life needs. Understanding helps solve variety urban related problems, such as increasing traffic capacity enhancing location-based service. Therefore, it is beneficial investigate how learn representations city zones terms functions, for better supporting analytic applications. To this end, paper, we propose framework vector representation...

In many recommender systems, users and items are associated with attributes, show preferences to items. The attribute information describes users'(items') characteristics has a wide range of applications, such as user profiling, item annotation, feature-enhanced recommendation. As annotating (item) attributes is labor intensive task, the values often incomplete missing values. Therefore, recommendation inference have become two main tasks in these platforms. Researchers long converged that...

In many recommender systems, users express item opinions through two kinds of behaviors: giving preferences and writing detailed reviews. As both behaviors reflect users' assessment items, review enhanced systems leverage these user to boost recommendation performance. On the one hand, researchers proposed better model embeddings with additional information for enhancing preference prediction accuracy. other some recent works focused on automatically generating reviews explanations related...

Knowledge graphs (KGs) have been widely used to improve recommendation accuracy. The multi-hop paths on KGs also enable reasoning, which is considered a crystal type of explainability. In this paper, we propose reinforcement learning framework for multi-level reasoning over KGs, leverages both ontology-view and instance-view model user interests. This ensures convergence more satisfying solution by effectively transferring high-level knowledge lower levels. Based the framework, path...

Recent studies have shown great promise in applying graph neural networks for multivariate time series forecasting, where the interactions of are described as a structure and variables represented nodes. Along this line, existing methods usually assume that (or adjacency matrix), which determines aggregation manner network, is fixed either by definition or self-learning. However, can be dynamic evolutionary real-world scenarios. Furthermore, quite different if they observed at scales. To...

The distribution shift in Time Series Forecasting (TSF), indicating series changes over time, largely hinders the performance of TSF models. Existing works towards time are mostly limited quantification and, more importantly, overlook potential between lookback and horizon windows. To address above challenges, we systematically summarize into two categories. Regarding windows as input-space output-space, there exist (i) intra-space shift, that within keeps shifted (ii) inter-space is...

Every day, our living city produces a tremendous amount of spatial-temporal data, involved with multiple sources from the individual scale to scale. Undoubtedly, such massive urban data can be explored for better and life, as what computing community has been dedicating in recent years. Nevertheless, existing studies are still facing challenges fusion well knowledge distillation specific applications. Moreover, there is lack full-featured user-friendly platforms both researchers developers...

Spatiotemporal data mining plays an important role in air quality monitoring, crowd flow modeling, and climate forecasting. However, the originally collected spatiotemporal real-world scenarios is usually incomplete due to sensor failures or transmission loss. imputation aims fill missing values according observed underlying dependence of them. The previous dominant models impute autoregressively suffer from problem error accumulation. As emerging powerful generative models, diffusion...

The spatio-temporal prediction task in the transportation network is core of solutions for various traffic problems. On one hand, mobility pattern can be reflected travel behavior crowd. In most tasks, importance often overlooked. other also has a variety predicting scenarios, including short-term and long-term prediction, relevant research cannot solve problems under two scenarios at same time. view problem existing work, we propose multi-pattern framework, MPGNNFormer. First, construct new...

It is traditionally a challenge for home buyers to understand, compare and contrast the investment values of real estates. While number estate appraisal methods have been developed value property, performances these limited by traditional data sources appraisal. However, with development new ways collecting estate-related mobile data, there potential leverage geographic dependencies estates enhancing Indeed, an can be from characteristics its own neighborhood (individual), nearby (peer),...

Collaborative Filtering (CF) is one of the most successful approaches for recommender systems. With emergence online social networks, recommendation has become a popular research direction. Most these models utilized each user's local neighbors' preferences to alleviate data sparsity issue in CF. However, they only considered neighbors user and neglected process that users' are influenced as information diffuses network. Recently, Graph Convolutional Networks~(GCN) have shown promising...

Ranking residential real estates based on investment values can provide decision making support for home buyers and thus plays an important role in estate marketplace. In this paper, we aim to develop methods ranking by mining users' opinions about from online user reviews offline moving behaviors (e.g., Taxi traces, smart card transactions, check-ins). While a variety of features could be extracted these data, are Interco related redundant. Thus, selecting good integrating the feature...