Predicting the next location a user tends to visit is an important task for applications like location-based advertising, traffic planning, and tour recommendation. We consider prediction problem semantic trajectory data, wherein each GPS record attached with text message that describes user's activity. In trajectories, confluence of spatiotemporal transitions textual messages indicates intents at fine granularity has great potential in improving accuracies. Nevertheless, existing methods...

Trajectory clustering, which aims at discovering groups of similar trajectories, has long been considered as a corner stone task for revealing movement patterns well facilitating higher-level applications like location prediction. While plethora trajectory clustering techniques have proposed, they often rely on spatiotemporal similarity measures that are not space- and time-invariant. As result, cannot detect clusters where the within-cluster occurs in different regions time periods. In this...

Trajectory similarity computation is a fundamental problem for various applications in trajectory data analysis. However, the high cost of existing measures has become key bottleneck analysis at scale. While there have been many research efforts reducing complexity, they are specific to one measure and often yield limited speedups. We propose NeuTraj accelerate computation. generic accommodate any fast compute given pair linear time. Furthermore, elastic collaborate with all spatial-based...

Recently, data mining through analyzing the complex structure and diverse relationships on multi-network has attracted much attention in both academia industry. One crucial prerequisite for this kind of is to map nodes across different networks, i.e., so-called network alignment. In paper, we propose a cross-network embedding method CrossMNA alignment problem investigating structural information only. Unlike previous methods focusing pair-wise learning holding topology consistent assumption,...

Computing trajectory similarities is a critical and fundamental task for various spatial-temporal applications, such as clustering, prediction, anomaly detection. Traditional similarity metrics, i.e. DTW Hausdorff, suffer from quadratic computation complexity, leading to their inability on large-scale data. To solve this problem, many representation learning techniques are proposed approximate the metric space while reducing complexity of computation. Nevertheless, these works designed based...

Abstract Trajectory clustering, which aims at discovering groups of similar trajectories, has long been considered as a corner stone task for revealing movement patterns well facilitating higher level applications such location prediction and activity recognition. Although plethora trajectory clustering techniques have proposed, they often rely on spatio‐temporal similarity measures that are not space time invariant. As result, cannot detect clusters where the within‐cluster occurs in...

Flight trajectory data plays a vital role in the traffic management community, especially for downstream tasks such as prediction, flight recognition, and anomaly detection. Existing works often utilize handcrafted features design models different individually, which heavily rely on domain expertise are hard to extend. We argue that analysis share same useful of trajectory. Jointly learning unified representation trajectories could be beneficial improving performance various tasks. However,...

Collective classification, as an important technique to study networked data, aims exploit the label autocorrelation for a group of inter-connected entities with complex dependencies. As emergence various heterogeneous information networks (HINs), collective classification at present is confronting several severe challenges stemming from heterogeneity HINs, such relational hierarchy, potential incompatible semantics and node-context semantics. To address challenges, in this paper, we propose...

Temporal data representing chronological observations of complex systems can be ubiquitously collected in smart industry, medicine, finance and etc. In the last decade, many tasks have been studied for mining temporal offered significant value various applications. Among these tasks, causal discovery aims to understand underlying generation mechanism has attracted much research attention. According whether is calibrated, existing approaches divided into two subtasks, i.e., multivariate...

Time series similarity computation is a fundamental primitive that underpins many time data analysis tasks. However, existing measures have high cost. While there has been much research effort for reducing the computational cost, such usually specific to one measure. We propose <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">NeuTS</small> ( <bold xmlns:xlink="http://www.w3.org/1999/xlink">Neu</b> ral metric learning...

Social media has became a critical manner for people to acquire information in daily life. Despite the great convenience, fake news can be widely spread through social networks, causing various adverse effects on people's lives. Detecting these or misinformations proved task and draws attentions from both governments individuals. Recently, many methods have been proposed solve this problem, but most of them rely body content news, ignoring context such as comments. We argue that comments...

In online advertising, marketing mix modeling (MMM) is employed to predict the gross merchandise volume (GMV) of brand shops and help decision-makers adjust budget allocation various advertising channels. Traditional MMM methods leveraging regression techniques can fail in handling complexity marketing. Although some efforts try encode causal structures for better prediction, they have strict restriction that are prior-known unchangeable. this paper, we define a new problem automatically...

In D2D communications, random contacts can be utilized to exchange data among nodes without the support from infrastructures or central control units. Because of huge quantity and high mobility nodes, scarcity available spectrum severely limits delivery capacity in communications. CR technology gives ability use idle licensed radio spectra networks improve capacity. The advantages opportunistic make communications an alternative that provides a complementary for big applications. However,...

Despite the recent and remarkable popularity of P2P IPTV systems, study their characteristics on largescale broadcasting is rare. In this paper, we have undertaken a measurement one most popular PPStream, during 29th Olympics broadcasting. We deployed our dedicated tool, PPS-Sniffer, under different network environments, collected extensive data large number events. evaluate playback delay, peering strategies towards local cluster. also investigate scheduling, its impact quality control...

User identity linkage (UIL) task aims to infer the identical users between different social networks/platforms. Existing models leverage labeled inter-linkages or high-quality user attributes make predictions. Nevertheless, it is often difficult even impossible obtain such information in real-world applications. To this end, we paper focus on studying an Anonymized Identity Linkage (AUIL) problem wherein neither anchor nor are available. handle a practical and challenging task, propose novel...

Video streaming uploading service over Vehicular Networks is very useful as it can support many applications. Due to the vehicle's high mobility and roadside access point (AP)'s sparse deployment, how provide a high-quality video with low-price charge still remains an open question. To address this issue, novel scheme based on vehicle moving prediction proposed, in which vehicle-to-infrastructure (V2I) vehicle-to-vehicle communications (V2V) are cooperated forward continuously from vehicles...

With the development of internet, there are billions short texts generated each day. However, accuracy large scale text classification is poor due to data sparseness. Traditional methods used use external dataset enrich representation document and solve sparsity problem. But which matches specific hard find. In this paper, we propose a framework problem without using dataset. Our deal with by making most semantic similarity words learned from training texts. First, learn word distributed...

The rapid evolution of large language models (LLMs) holds promise for reforming the methodology spatio-temporal data mining. However, current works evaluating understanding capability LLMs are somewhat limited and biased. These either fail to incorporate latest or only focus on assessing memorized knowledge. To address this gap, paper dissects LLMs' into four distinct dimensions: knowledge comprehension, reasoning, accurate computation, downstream applications. We curate several natural...

In Software-Defined Networking (SDN), Ternary Content Addressable Memory (TCAM) enables fast lookup with flexible wildcard rule patterns for flow tables, however, the scarcity and expensiveness of TCAM dramatically limit number rules that switches can support. Rule caching breaks table size constraint by appropriate combinations hardware software processing. Nevertheless, previous literatures, from viewpoint maximizing cache hit ratio, ignore update operations incurred replacement, which is...