The need to safeguard data privacy and adhere to regulations such as GDPR creates data silos and has prompted the emergence and widespread adoption of techniques for distributed databases. To effectively explore the value of data across multiple organizations, techniques for data management, data analysis and data functionality from distributed databases have been proposed. Recently, Vertical Federated Learning (VFL) has become a solution with growing interests, which enables collaborative model training when data features are partitioned into multiple parts and are held by different parties. However, typical VFL methods heavily rely on private set intersection (PSI) to align data before training and only utilize aligned data for training. In this work, we provide a theoretical analysis to show that unaligned data actually contains valuable and rich features, and a thoughtful design that harnesses the potential of unaligned samples to significantly improve the performance of VFL models. Regrettably, many existing methods simply discard unaligned data, resulting in an irrecoverable loss of performance. To address this data sacrifice problem, we introduce the concept of data mixture, which enables the utilization of both aligned and unaligned data during training. Building upon the data mixture idea, we present FedMix, the first on-the-fly and distribution-agnostic framework designed to boost the performance of VFL models by leveraging unaligned data. A data seasoning approach is also designed to utilize auxiliary data lacking label information. Evaluations on diverse datasets under different settings demonstrate the effectiveness of the proposed FedMix compared with various SOTA approaches. FedMix achieves up to 15% model performance improvement and 30.5 hours time cost reduction.

Online learning to rank (OLTR), which directly optimizes the ranker with interactive user feedback, has gained considerable attention in both academia and industry. However, most current approaches suffer from the inefficiency of heuristic exploration strategies, which can seriously hurt users' experience. Furthermore, the existing OLTR solutions fail to learn from the cost-effective logged data, blocking their usage in the real industrial system. To handle the above issues, we in this paper introduce a new OLTR framework LT2R, namely Learning To online Learning to Rank. LT2R aims to study an efficient parameterized exploration strategy, by which a ranker could converge to the optimal ranking with as few exploration steps as possible. Specifically, we formulate the OLTR task as a typical Markov Decision Process and introduce an online reinforcement learning algorithm with a multi-round cumulative reward to guarantee fast convergence. Moreover, we contribute an offline learning algorithm for LT2R to exploit the knowledge from the historical searching logs, which can provide a fair warm-up model for its industrial deployment. Extensive experiments on both benchmark datasets and Baidu search engine have demonstrated its superiority over state-of-the-art methods.

Trajectory anomaly detection, aiming to estimate the anomaly risk of trajectories given the Source-Destination (SD) pairs, has become a critical problem for many real-world applications. Existing solutions directly train a generative model for observed trajectories and calculate the conditional generative probability$P(T \vert C)$as the anomaly risk, where$T$and$C$represent the trajectory and SD pair respectively. However, we argue that the observed trajectories are confounded by road network preference which is a common cause of both SD distribution and trajectories. Existing methods ignore this issue limiting their generalization ability on out-of-distribution trajectories. In this paper, we define the debiased trajectory anomaly detection problem and propose a causal implicit generative model, namely CausalTAD, to solve it. CausalTAD adopts do-calculus to eliminate the confounding bias of road network preference and estimates$P(T\vert do(C))$as the anomaly criterion. Extensive experiments show that CausalTadcan not only achieve superior performance on trained trajectories but also generally improve the performance of out-of-distribution data, with improvements of 2.1% ~ 5.7% and 10.6% ~ 32.7% respectively.

The page presentation biases in the information retrieval system, especially on the click behavior, is a well-known challenge that hinders improving ranking models' performance with implicit user feedback. Unbiased Learning to Rank~(ULTR) algorithms are then proposed to learn an unbiased ranking model with biased click data. However, most existing algorithms are specifically designed to mitigate position-related bias, e.g., trust bias, without considering biases induced by other features in search result page presentation(SERP), e.g. attractive bias induced by the multimedia. Unfortunately, those biases widely exist in industrial systems and may lead to an unsatisfactory search experience. Therefore, we introduce a new problem, i.e., whole-page Unbiased Learning to Rank(WP-ULTR), aiming to handle biases induced by whole-page SERP features simultaneously. It presents tremendous challenges: (1) a suitable user behavior model (user behavior hypothesis) can be hard to find; and (2) complex biases cannot be handled by existing algorithms. To address the above challenges, we propose a Bias Agnostic whole-page unbiased Learning to rank algorithm, named BAL, to automatically find the user behavior model with causal discovery and mitigate the biases induced by multiple SERP features with no specific design. Experimental results on a real-world dataset verify the effectiveness of the BAL.

Pre-trained language representation models (PLMs) such as BERT and Enhanced Representation through kNowledge IntEgration (ERNIE) have been integral to achieving recent improvements on various downstream tasks, including information retrieval. However, it is nontrivial to directly utilize these models for the large-scale web search due to the following challenging issues: (1) the prohibitively expensive computations of massive neural PLMs, especially for long texts in the web document, prohibit their deployments in the web search system that demands extremely low latency; (2) the discrepancy between existing task-agnostic pre-training objectives and the ad hoc retrieval scenarios that demand comprehensive relevance modeling is another main barrier for improving the online retrieval and ranking effectiveness; and (3) to create a significant impact on real-world applications, it also calls for practical solutions to seamlessly interweave the resultant PLM and other components into a cooperative system to serve web-scale data. Accordingly, we contribute a series of successfully applied techniques in tackling these exposed issues in this work when deploying the state-of-the-art Chinese pre-trained language model, i.e., ERNIE, in the online search engine system. We first present novel practices to perform expressive PLM-based semantic retrieval with a flexible poly-interaction scheme and cost-efficiently contextualize and rank web documents with a cheap yet powerful Pyramid-ERNIE architecture. We then endow innovative pre-training and fine-tuning paradigms to explicitly incentivize the query-document relevance modeling in PLM-based retrieval and ranking with the large-scale noisy and biased post-click behavioral data. We also introduce a series of effective strategies to seamlessly interwoven the designed PLM-based models with other conventional components into a cooperative system. Extensive offline and online experimental results show that our proposed techniques are crucial to achieving more effective search performance. We also provide a thorough analysis of our methodology and experimental results.

With the emerging of smartphones, mobile games have attracted billions of players and occupied most of the share for game companies. On the other hand, mobile game cheating, aiming to gain improper advantages by using programs that simulate the players' inputs, severely damages the game's fairness and harms the user experience. Therefore, detecting mobile game cheating is of great importance for mobile game companies. Many PC game-oriented cheating detection methods have been proposed in the past decades, however, they can not be directly adopted in mobile games due to the concern of privacy, power, and memory limitations of mobile devices. Even worse, in practice, the cheating programs are quickly updated, leading to the label scarcity for novel cheating patterns. To handle such issues, we in this paper introduce a mobile game cheating detection framework, namely FCDGame, to detect the cheats under the few-shot learning framework. FCDGame only consumes the screen sensor data, recording users' touch trajectories, which is less sensitive and more general for almost all mobile games. Moreover, a Hierarchical Trajectory Encoder and a Cross-pattern Meta Learner are designed in FCDGame to capture the intrinsic characters of mobile games and solve the label scarcity problem, respectively. Extensive experiments on two real online games show that FCDGame achieves almost 10% improvements in detection accuracy with only few fine-tuned samples.

The pre-trained language models (PLMs), such as BERT and ERNIE, have achieved outstanding performance in many natural language understanding tasks. Recently, PLMs-based Information Retrieval models have also been investigated and showed substantially state-of-the-art effectiveness, e.g., MORES, PROP and ColBERT. Moreover, most of the PLMs-based rankers only focus on a single level relevance matching (e.g., character-level), while ignore the other granularity information (e.g., words and phrases), which easily lead to the ambiguity of query understanding and inaccurate matching issues in web search.

This paper studies the task of detecting bots for online mobile games. Considering the fact of lacking labeled cheating samples and restricted available data in the real detection systems, we aim to study the finger operations captured by screen sensors to infer the potential bots in an unsupervised way. In detail, we introduce a Transformer-style detection model, namely FingFormer. It studies the finger operations in the format of graph structure in order to capture the spatial and temporal relatedness between the two hands’ operations. To optimize the model in an unsupervised way, we introduce two contrastive learning strategies to refine both finger moving patterns and players’ operation habits. We conduct extensive experiments under different experimental environments, including the synthetic dataset, the offline dataset, as well as the large-scale online data flow from three mobile games. The multi-facet experiments illustrate the proposed model is both effective and general to detect the bots for different mobile games.

User identity linkage (UIL) task aims to infer the identical users between different social networks/platforms. Existing models leverage the labeled inter-linkages or high-quality user attributes to make predictions. Nevertheless, it is often difficult or even impossible to obtain such information in real-world applications. To this end, we in this paper focus on studying an Anonymized User Identity Linkage (AUIL) problem wherein neither labeled anchor users nor attributes are available. To handle such a practical and challenging task, we propose a novel and concise unsupervised embedding method, VCNE, by utilizing the network structural information. Concretely, considering the inherent properties of structural diversity in the AUIL problem, we introduce a variational cross-network embedding learning framework to jointly study the Gaussian embeddings instead of the existing deterministic embedding from the angle of vector space. The multi-facet experiments on both real-world and synthetic datasets demonstrate that VCNE not only outperforms all baselines to a large extent but also be more robust to the different-level diversities and sparsities of the networks.

Heterogeneous information networks (HINs) have been successfully applied into several fields to accomplish complex data analytics, such as bibliography, bioinformatics, NLP, etc. In the meantime, network embedding at present has emerged as a convenient tool to mine and learn from networked data. As a result, it is of interest to develop HIN embedding methods. Despite recent breakthroughs in HIN embedding methods, little research attention has been paid to exploit the relation semantics in HINs and further integrate it to improve the embedding quality. Considering the sophisticated correlations in HINs, we in this paper propose a novel HIN embedding method LRHNE to yield latent-relation enhanced embeddings for nodes. Our work mainly involves three contributions: i) we verify that the latent relation can promote the embedding quality indeed through a real-world dataset, then a novel graph inception network is proposed to extract the latent relational features under the guidance of partial prior knowledge; ii) taking into account the existing structure information and inferred latent relation knowledge, we propose a cross-aligned variational graph autoencoder to extract and further fuse both the structure and latent relational features into the embeddings; and iii) we perform extensive experiments to validate our proposed LRHNE, and experimental results show that our LRHNE can significantly outperform state-of-the-art methods. The multi-facet inspections also exhibit our method is robust and hyper-parameter insensitive, therefore, our method can serve as a radical tool to tackle the relation-sophisticated HINs.

Collective classification, as an important technique to study networked data, aims to exploit the label autocorrelation for a group of inter-connected entities with complex dependencies. As the emergence of various heterogeneous information networks (HINs), collective classification at present is confronting several severe challenges stemming from the heterogeneity of HINs, such as complex relational hierarchy, potential incompatible semantics and node-context relational semantics. To address the challenges, in this paper, we propose a novel heterogeneous graph convolutional network-based deep learning model, called HGCN, to collectively categorize the entities in HINs. Our work involves three primary contributions: i) HGCN not only learns the latent relations from the relation-sophisticated HINs via multi-layer heterogeneous convolutions, but also captures the semantic incompatibility among relations with properly-learned edge-level filter parameters; ii) to preserve the fine-grained relational semantics of different-type nodes, we propose a heterogeneous graph convolution to directly tackle the original HINs without any in advance transforming the network from heterogeneity to homogeneity; iii) we perform extensive experiments using four real-world datasets to validate our proposed HGCN, the multi-facet results show that our proposed HGCN can significantly improve the performance of collective classification compared with the state-of-the-art baseline methods.

Recently, data mining through analyzing the complex structure and diverse relationships on multi-network has attracted much attention in both academia and industry. One crucial prerequisite for this kind of multi-network mining is to map the nodes across different networks, i.e., so-called network alignment. In this paper, we propose a cross-network embedding method CrossMNA for multi-network alignment problem through investigating structural information only. Unlike previous methods focusing on pair-wise learning and holding the topology consistent assumption, our proposed CrossMNA considers the multi-network scenarios which involve at least two types of networks with diverse network structures. CrossMNA leverages the cross-network information to refine two types of node embedding vectors, i.e., inter-vector for network alignment and intra-vector for other downstream network analysis tasks. Finally, we verify the effectiveness and efficiency of our proposed method using several real-world datasets. The extensive experiments show that our CrossMNA can significantly outperform the existing baseline methods on multi-network alignment task, and also achieve better performance for link prediction task with less memory usage.