Event Causality Extraction (ECE) aims to extract causal event pairs from text. Existing methods overlook the interplay between causal event pairs and their corresponding textual evidence (e.g., causal event mention pairs), and fail to effectively leverage global causal dependency information. To address these issues, we propose a Mention-Level Causal Evidence and Global Causal Graph Reasoning (MLCE-GCGR) framework to enhance ECE. First, we introduce an auxiliary Event Mention Causality Extraction (EMCE) task, which extracts causal event mention pairs, to provide evidence for the main ECE task, and design a Dual-Level Interaction Enhancement (DLIE) strategy to enhance the bidirectional interplay between event-level and mention-level causality. Second, we develop a Global Causal Graph Reasoning (GCGR) module that simulates human-like multi-turn reasoning, aiming to progressively refine the causal graph by capturing global dependencies among event mentions, types, and arguments. Experiments on four benchmark datasets show that our method outperforms state-of-the-art approaches. Moreover, by extracting causal event mention pairs as supporting evidence, our approach improves the interpretability of structured causality extraction.

Source-free domain adaptation (SFDA) adapts a pre-trained model from a labeled source domain to an unlabeled target domain without source data access, preserving privacy. While SFDA is common in computer vision, it remains largely unexplored in time series analysis, where existing methods struggle to capture temporal dynamics and often produce overconfident predictions on out-of-distribution samples. We propose MAsk And imPUte (MAPU), which tackles temporal consistency through a novel imputation task, where randomly masked time series signals are recovered within the learned embedding space. During adaptation, a dedicated temporal imputer guides the target model to generate features that maintain temporal consistency with source features. However, MAPU relies on standard softmax predictions, leading to overconfident predictions on target samples that fall outside the source domain's support. To address this limitation, we introduce Evidential-MAPU (E-MAPU), which leverages evidential uncertainty estimation to identify these out-of-support samples and adapts the feature extractor to map them closer to the source domain's support, while maintaining the classifier fixed. Extensive experiments on five real-world time series datasets demonstrate significant performance improvements over existing methods. Our approaches effectively handle various time series domain adaptation challenges while maintaining computational efficiency, achieving state-of-the-art performance through its uncertainty-aware adaptation strategy.

Rationalization, a data-centric framework, aims to build self-explanatory models to explain the prediction outcome by generating a subset of human-intelligible pieces of the input data. It involves a cooperative game model where a generator generates the most human-intelligible parts of the input (i.e., rationales), followed by a predictor that makes predictions based on these generated rationales. Conventional rationalization methods typically impose constraints via regularization terms to calibrate or penalize undesired generation. However, these methods are suffering from a problem called mode collapse, in which the predictor produces correct predictions yet the generator consistently outputs rationales with collapsed patterns. Moreover, existing studies are typically designed separately for specific collapsed patterns, lacking a unified consideration. In this paper, we systematically revisit cooperative rationalization from a novel game-theoretic perspective and identify the fundamental cause of this problem: the generator no longer tends to explore new strategies to uncover informative rationales, ultimately leading the system to converge to a suboptimal game equilibrium (correct predictionsv.scollapsed rationales). To solve this problem, we then propose a novel approach, Game-theoreticPolicyOptimization orientedRATionalization (PoRat), which progressively introduces policy interventions to address the game equilibrium in the cooperative game process, thereby guiding the model toward a more optimal solution state. We theoretically analyse the cause of such a suboptimal equilibrium and prove the feasibility of the proposed method. Furthermore, we validate our method on nine widely used real-world datasets and two synthetic settings, wherePoRatachieves up to 8.1% performance improvements over existing state-of-the-art methods. Code and data are available atanonymous.4open.science/r/Rationalization-PORAT-ECE9.

Source-Free Domain Adaptation (SFDA) aims to adapt a pre-trained model from an annotated source domain to an unlabelled target domain without accessing the source data, thereby preserving data privacy. While existing SFDA methods have proven effective in reducing reliance on source data, they struggle to perform well on multivariate time series (MTS) due to their failure to consider the intrinsic spatial correlations inherent in MTS data. These spatial correlations are crucial for accurately representing MTS data and preserving invariant information across domains. To address this challenge, we propose Temporal Restoration and Spatial Rewiring (TERSE), a novel and concise SFDA method tailored for MTS data. Specifically, TERSE comprises a customized spatial-temporal feature encoder designed to capture the underlying spatial-temporal characteristics, coupled with both temporal restoration and spatial rewiring tasks to reinstate latent representations of the temporally masked time series and the spatially masked correlated structures. During the target adaptation phase, the target encoder is guided to produce spatially and temporally consistent features with the source domain by leveraging the source pre-trained temporal restoration and spatial rewiring networks. Therefore, TERSE can effectively model and transfer spatial-temporal dependencies across domains, facilitating implicit feature alignment. In addition, as the first approach to simultaneously consider spatial-temporal consistency in MTS-SFDA, TERSE can also be integrated as a versatile plug-and-play module into established SFDA methods. Extensive experiments on three real-world time series datasets demonstrate the effectiveness and versatility of our approach. Our code is available at https://github.com/Tokenmw/TERSE-master.

Test-time adaptation aims to adapt pre-trained deep neural networks using solely online unlabelled test data during inference. Although TTA has shown promise in visual applications, its potential in time series contexts remains largely unexplored. Existing TTA methods, originally designed for visual tasks, may not effectively handle the complex temporal dynamics of real-world time series data, resulting in suboptimal adaptation performance. To address this gap, we propose Augmented Contrastive Clustering with Uncertainty-aware Prototyping (ACCUP), a straightforward yet effective TTA method for time series data. Initially, our approach employs augmentation ensemble on the time series data to capture diverse temporal information and variations, incorporating uncertainty-aware prototypes to distill essential characteristics. Additionally, we introduce an entropy comparison scheme to selectively acquire more confident predictions, enhancing the reliability of pseudo labels. Furthermore, we utilize augmented contrastive clustering to enhance feature discriminability and mitigate error accumulation from noisy pseudo labels, promoting cohesive clustering within the same class while facilitating clear separation between different classes. Extensive experiments conducted on three real-world time series datasets demonstrate the effectiveness and generalization potential of the proposed method, advancing the underexplored realm of TTA for time series data. Our code is available at https://github.com/Tokenmw/ACCUP-main.

Key-value databases with Log Structured Merge tree are increasingly favored by modern applications. Apart from supporting fast lookup on primary key, efficient queries on non-key attributes are also highly demanded by many of these applications. To enhance query performance, many auxiliary structures like secondary indexing and filters have been developed. However, existing auxiliary structures suffer from three limitations. First, creating filter for every disk component has low lookup efficiency as all components need to be searched during query processing. Second, current secondary index design requires primary table access to fetch the data entries for each output primary key from the index. This indirect entries fetching process involves significant point lookup overhead in the primary table and hence hinders the query performance. Last, maintaining the consistency between the secondary index and the primary table is challenging due to the out-of-place update mechanism of the LSM-tree. To overcome the limitations in existing auxiliary structures for non-key attributes queries, this paper proposes a novel secondary index framework, NEXT, for LSM-based key-value storage system. NEXT utilizes a two-level structure which is integrated with the primary table. In particular, NEXT proposes to create secondary index blocks on each LSM disk component to map the secondary attributes to their corresponding data blocks. In addition, NEXT introduces a global index component which is created on top of all secondary index blocks to direct the secondary index operation to the target secondary index blocks. Finally, NEXT adopts two optimization strategies to further improve the query performance. We implement NEXT on RocksDB and experimentally evaluate its performance against existing methods. Experiments on both static and mixed workloads demonstrate that NEXT outperforms existing methods for different types of non-key attributes.

Attributed Question Answering (AQA) aims to provide both a trustworthy answer and a reliable attribution report for a given question. Retrieval is a widely adopted approach, including two general paradigms: Retrieval-Then-Read (RTR) and post-hoc retrieval. Recently, Large Language Models (LLMs) have shown remarkable proficiency, prompting growing interest in AQA among researchers. However, RTR-based AQA often suffers from irrelevant knowledge and rapidly changing information, even when LLMs are adopted, while post-hoc retrievalbased AQA struggles with comprehending long-form answers with complex logic, and precisely identifying the content needing revision and preserving the original intent. To tackle these problems, this paper proposes an Atomic fact decompositionbased Retrieval and Editing (ARE) framework, which decomposes the generated long-form answers into molecular clauses and atomic facts by the instruction-tuned LLMs. Notably, the instruction-tuned LLMs are fine-tuned using a well-constructed dataset, generated from large scale Knowledge Graphs (KGs). This process involves extracting one-hop neighbors from a given set of entities and transforming the result into coherent long-form text. Subsequently, ARE leverages a search engine to retrieve evidences related to atomic facts, inputting these evidences into an LLM-based verifier to determine whether the facts require expansion for re-retrieval or editing. Furthermore, the edited facts are backtracked into the original answer, with evidence aggregated based on the relationship between molecular clauses and atomic facts. Extensive evaluations demonstrate the superior performance of our proposed method over the state-of-the-arts on several datasets, with an additionally proposed new metricAttrpfor evaluating the precision of evidence attribution.

Source-free domain adaptation (SFDA) aims to adapt a pretrained model from a labeled source domain to an unlabeled target domain without access to the source domain data, preserving source domain privacy. Despite its prevalence in visual applications, SFDA is largely unexplored in time series applications. The existing SFDA methods that are mainly designed for visual applications may fail to handle the temporal dynamics in time series, leading to impaired adaptation performance. To address this challenge, this paper presents a simple yet effective approach for source-free domain adaptation on time series data, namely MAsk and imPUte (MAPU). First, to capture temporal information of the source domain, our method performs random masking on the time series signals while leveraging a novel temporal imputer to recover the original signal from a masked version in the embedding space. Second, in the adaptation step, the imputer network is leveraged to guide the target model to produce target features that are temporally consistent with the source features. To this end, our MAPU can explicitly account for temporal dependency during the adaptation while avoiding the imputation in the noisy input space. Our method is the first to handle temporal consistency in SFDA for time series data and can be seamlessly equipped with other existing SFDA methods. Extensive experiments conducted on three real-world time series datasets demonstrate that our MAPU achieves significant performance gain over existing methods. Our code is available at: https://github.com/mohamedr002/MAPU_SFDA_TS.

Unsupervised domain adaptation methods aim at generalizing well on unlabeled test data that may have a different (shifted) distribution from the training data. Such methods are typically developed on image data, and their application to time series data is less explored. Existing works on time series domain adaptation suffer from inconsistencies in evaluation schemes, datasets, and backbone neural network architectures. Moreover, labeled target data are often used for model selection, which violates the fundamental assumption of unsupervised domain adaptation. To address these issues, we develop a benchmarking evaluation suite ( AdaTime ) to systematically and fairly evaluate different domain adaptation methods on time series data. Specifically, we standardize the backbone neural network architectures and benchmarking datasets, while also exploring more realistic model selection approaches that can work with no labeled data or just a few labeled samples. Our evaluation includes adapting state-of-the-art visual domain adaptation methods to time series data as well as the recent methods specifically developed for time series data. We conduct extensive experiments to evaluate 11 state-of-the-art methods on five representative datasets spanning 50 cross-domain scenarios. Our results suggest that with careful selection of hyper-parameters, visual domain adaptation methods are competitive with methods proposed for time series domain adaptation. In addition, we find that hyper-parameters could be selected based on realistic model selection approaches. Our work unveils practical insights for applying domain adaptation methods on time series data and builds a solid foundation for future works in the field. The code is available at github.com/emadeldeen24/AdaTime .

Keyphrases can concisely describe the high-level topics discussed in a document that usually possesses hierarchical topic structures. Thus, it is crucial to understand the hierarchical topic structures and employ it to guide the keyphrase identification. However, integrating the hierarchical topic information into a deep keyphrase generation model is unexplored. In this paper, we focus on how to effectively exploit the hierarchical topic to improve the keyphrase generation performance (HTKG). Specifically, we propose a novel hierarchical topic-guided variational neural sequence generation method for keyphrase generation, which consists of two major modules: a neural hierarchical topic model that learns the latent topic tree across the whole corpus of documents, and a variational neural keyphrase generation model to generate keyphrases under hierarchical topic guidance. Finally, these two modules are jointly trained to help them learn complementary information from each other. To the best of our knowledge, this is the first attempt to leverage the neural hierarchical topic to guide keyphrase generation. The experimental results demonstrate that our method significantly outperforms the existing state-of-the-art methods across five benchmark datasets.

Index selection remains one of the most challenging problems in relational database management systems. To find an optimum index configuration for a workload, accurately and efficiently quantifying the benefits of each candidate index configuration is indispensable. As materializing each index configuration candidate and physically executing queries are infeasible, most of index tuners rely on the cost estimations from optimizer with "what-if" API. However, "what-if" based index benefit estimations have the following two limitations. Firstly, they generate significant errors, which compromise index recommendation quality. Secondly, generating query plans and benefit estimations for each candidate index configuration takes a considerable amount of time. To address the two challenges in index selection, we propose an effective end-to-end machine learning based index benefit estimator. In particular, we propose novel feature extraction and encoding techniques that do not rely on "what-if" call to generate query plan for each index configuration candidate. In addition, we design an attention mechanism to address index interaction issue and aggregate the impacts of different query operations. Finally, we leverage transfer learning technique to improve the estimator's learning ability for adaption to new database. Comprehensive experiments are conducted on different workloads, and extensive experimental results show that our proposed method outperforms "what-if" based index benefit estimations in terms of accuracy and efficiency. In addition, integrating our method into existing index selection algorithms can significantly improve index recommendation quality.

Real-world networks often exist with multiple views, where each view describes one type of interaction among a common set of nodes. For example, on a video-sharing network, while two user nodes are linked, if they have common favorite videos in one view, then they can also be linked in another view if they share common subscribers. Unlike traditional single-view networks, multiple views maintain different semantics to complement each other. In this article, we propose M ulti-view coll A borative N etwork E mbedding (MANE), a multi-view network embedding approach to learn low-dimensional representations. Similar to existing studies, MANE hinges on diversity and collaboration—while diversity enables views to maintain their individual semantics, collaboration enables views to work together. However, we also discover a novel form of second-order collaboration that has not been explored previously, and further unify it into our framework to attain superior node representations. Furthermore, as each view often has varying importance w.r.t. different nodes, we propose MANE , an attention -based extension of MANE, to model node-wise view importance. Finally, we conduct comprehensive experiments on three public, real-world multi-view networks, and the results demonstrate that our models consistently outperform state-of-the-art approaches.

Data in the form of graphs are prevalent, ranging from biological and social networks to citation graphs and the Web. In particular, most real-world graphs are heterogeneous, containing objects of multiple types, which present new opportunities for many problems on graphs. Consider a typical proximity search problem on graphs, which boils down to measuring the proximity between two given nodes. Most earlier studies on homogeneous or bipartite graphs only measure a generic form of proximity, without accounting for different “semantic classes”-for instance, on a social network two users can be close for different reasons, such as being classmates or family members, which represent two distinct semantic classes. Learning these semantic classes are made possible on heterogeneous graphs through the concept of metagraphs. In this study, we identify metagraphs as a novel and effective means to characterize the common structures for a desired class of proximity. Subsequently, we propose a family of metagraph-based proximity, and employ a learning-to-rank technique that automatically learns the right parameters to suit the desired semantic class. In terms of efficiency, we develop a symmetry-based matching algorithm to speed up the computation of metagraph instances. Empirically, extensive experiments reveal that our metagraph-based proximity substantially outperforms the best competitor by more than 10 percent, and our matching algorithm can reduce matching time by more than half. As a further generalization, we aim to derive a general node and edge representation for heterogeneous graphs, in order to support arbitrary machine learning tasks beyond proximity search. In particular, we propose the finer-grained anchored metagraph, which is capable of discriminating the roles of nodes within the same metagraph. Finally, further experiments on the general representation show that we can outperform the state of the art significantly and consistently across various machine learning tasks.

Short text classification has been widely explored in news tagging to provide more efficient search strategies and more effective search results for information retrieval. However, most existing studies, concentrating on long text classification, deliver unsatisfactory performance on short texts due to the sparsity issue and the insufficiency of labeled data. In this article, we propose a novel heterogeneous graph neural network-based method for semi-supervised short text classification, leveraging full advantage of limited labeled data and large unlabeled data through information propagation along the graph. Specifically, we first present a flexible heterogeneous information network (HIN) framework for modeling short texts, which can integrate any type of additional information and meanwhile capture their relations to address the semantic sparsity. Then, we propose Heterogeneous Graph Attention networks (HGAT) to embed the HIN for short text classification based on a dual-level attention mechanism, including node-level and type-level attentions. To efficiently classify new coming texts that do not previously exist in the HIN, we extend our model HGAT for inductive learning, avoiding re-training the model on the evolving HIN. Extensive experiments on single-/multi-label classification demonstrates that our proposed model HGAT significantly outperforms state-of-the-art methods across the benchmark datasets under both transductive and inductive learning.

Session-based recommendation (SBR) is a challenging task, which aims at recommending items based on anonymous behavior sequences. Almost all the existing solutions for SBR model user preference only based on the current session without exploiting the other sessions, which may contain both relevant and irrelevant item-transitions to the current session. This paper proposes a novel approach, called Global Context Enhanced Graph Neural Networks (GCE-GNN) to exploit item transitions over all sessions in a more subtle manner for better inferring the user preference of the current session. Specifically, GCE-GNN learns two levels of item embeddings from session graph and global graph, respectively: (i) Session graph, which is to learn the session-level item embedding by modeling pairwise item-transitions within the current session; and (ii) Global graph, which is to learn the global-level item embedding by modeling pairwise item-transitions over all sessions. In GCE-GNN, we propose a novel global-level item representation learning layer, which employs a session-aware attention mechanism to recursively incorporate the neighbors' embeddings of each node on the global graph. We also design a session-level item representation learning layer, which employs a GNN on the session graph to learn session-level item embeddings within the current session. Moreover, GCE-GNN aggregates the learnt item representations in the two levels with a soft attention mechanism. Experiments on three benchmark datasets demonstrate that GCE-GNN outperforms the state-of-the-art methods consistently.

This paper investigates the notion of learning user and item representations in non-Euclidean space. Specifically, we study the connection between metric learning in hyperbolic space and collaborative filtering by exploring Mobius gyrovector spaces where the formalism of the spaces could be utilized to generalize the most common Euclidean vector operations. Overall, this work aims to bridge the gap between Euclidean and hyperbolic geometry in recommender systems through metric learning approach. We propose HyperML (Hyperbolic Metric Learning), a conceptually simple but highly effective model for boosting the performance. Via a series of extensive experiments, we show that our proposed HyperML not only outperforms their Euclidean counterparts, but also achieves state-of-the-art performance on multiple benchmark datasets, demonstrating the effectiveness of personalized recommendation in hyperbolic geometry.

Topic model and word embedding reflect two perspectives of text semantics. Topic model maps documents into topic distribution space by utilizing word collocation patterns within and across documents, while word embedding represents words within a continuous embedding space by exploiting the local word collocation patterns in context windows. Clearly, these two types of patterns are complementary. In this paper, we propose a novel integration framework to combine the two representation methods, where topic information can be transmitted into corresponding semantic embedding structure. Based on this framework, we construct a Embedding Enhanced Topic Model (EETM), which can improve topic modeling and generate topic embeddings by leveraging the word embedding. Extensive experimental results show that EETM can learn high-quality document representations for common text analysis tasks across multiple data sets, indicating it is very effective for merging topic models with word embeddings.

This paper proposes Medley of Sub-Attention Networks (MoSAN), a new novel neural architecture for the group recommendation task. Group-level recommendation is known to be a challenging task, in which intricate group dynamics have to be considered. As such, this is to be contrasted with the standard recommendation problem where recommendations are personalized with respect to a single user. Our proposed approach hinges upon the key intuition that the decision making process (in groups) is generally dynamic, i.e., a user's decision is highly dependent on the other group members. All in all, our key motivation manifests in a form of an attentive neural model that captures fine-grained interactions between group members. In our MoSAN model, each sub-attention module is representative of a single member, which models a user's preference with respect to all other group members. Subsequently, a Medley of Sub-Attention modules is then used to collectively make the group's final decision. Overall, our proposed model is both expressive and effective. Via a series of extensive experiments, we show that MoSAN not only achieves state-of-the-art performance but also improves standard baselines by a considerable margin.

Traditional supervised keyphrase extraction models depend on the features of labelled keyphrases while prevailing unsupervised models mainly rely on structure of the word graph, with candidate words as nodes and edges capturing the co-occurrence information between words. However, systematically integrating all these multidimensional heterogeneous information into a unified model is relatively unexplored. In this paper, we focus on how to effectively exploit multidimensional information to improve the keyphrase extraction performance (MIKE). Specifically, we propose a random-walk parametric model, MIKE, that learns the latent representation for a candidate keyphrase that captures the mutual influences among all information, and simultaneously optimizes the parameters and ranking scores of candidates in the word graph. We use the gradient-descent algorithm to optimize our model and show the comprehensive experiments with two publicly-available WWW and KDD datasets in Computer Science. Experimental results demonstrate that our approach significantly outperforms the state-of-the-art graph-based keyphrase extraction approaches.

The identification of network motifs has important applications in numerous domains, such as pattern detection in biological networks and graph analysis in digital circuits. However, mining network motifs is computationally challenging, as it requires enumerating subgraphs from a real-life graph, and computing the frequency of each subgraph in a large number of random graphs. In particular, existing solutions often require days to derive network motifs from biological networks with only a few thousand vertices. To address this problem, this paper presents a novel study on network motif discovery using Graphical Processing Units (GPUs). The basic idea is to employ GPUs to parallelize a large number of subgraph matching tasks in computing subgraph frequencies from random graphs, so as to reduce the overall computation time of network motif discovery. We explore the design space of GPU-based subgraph matching algorithms, with careful analysis of several crucial factors (such as branch divergences and memory coalescing) that affect the performance of GPU programs. Based on our analysis, we develop a GPU-based solution that (i) considerably differs from existing CPU-based methods in how it enumerates subgraphs, and (ii) exploits the strengths of GPUs in terms of parallelism while mitigating their limitations in terms of the computation power per GPU core. With extensive experiments on a variety of biological networks, we show that our solution is up to two orders of magnitude faster than the best CPU-based approach, and is around 20 times more cost-effective than the latter, when taking into account the monetary costs of the CPU and GPUs used.

With the rapid proliferation of social media, increasingly more people express their opinions and reviews (user-generated content (UGC)) on recent news articles through various online services, such as news portals, forums, discussion groups, and microblogs. Clearly, identifying hot topics that users greatly care about can improve readers’ news browsing experience and facilitate research into interaction analysis between news and UGC. Furthermore, it is of great benefit to public opinion monitoring and management for both industry and government agencies. However, it is extremely time consuming, if not impossible, to manually examine the large amount of available social content. In this article, we formally define the news comment alignment problem and propose a novel framework that: (1) automatically extracts topics from a given news article and its associated comments, (2) identifies and extends positive examples with different degrees of confidence using three methods (i.e., hypersphere, density, and cluster chain), and (3) completes the alignment between news sentences and comments through a weighted-SVM classifier. Extensive experiments show that our proposed framework significantly outperforms state-of-the-art methods.

Given ubiquitous graph data such as the Web and social networks, proximity search on graphs has been an active research topic. The task boils down to measuring the proximity between two nodes on a graph. Although most earlier studies deal with homogeneous or bipartite graphs only, many real-world graphs are heterogeneous with objects of various types, giving rise to different semantic classes of proximity. For instance, on a social network two users can be close for different reasons, such as being classmates or family members, which represent two distinct classes of proximity. Thus, it becomes inadequate to only measure a “generic” form of proximity as previous works have focused on. In this paper, we identify metagraphs as a novel and effective means to characterize the common structures for a desired class of proximity. Subsequently, we propose a family of metagraph-based proximity, and employ a supervised technique to automatically learn the right form of proximity within its family to suit the desired class. As it is expensive to match (i.e., find the instances of) a metagraph, we propose the novel approaches of dual-stage training and symmetry-based matching to speed up. Finally, our experiments reveal that our approach is significantly more accurate and efficient. For accuracy, we outperform the baselines by 11% and 16% in NDCG and MAP, respectively. For efficiency, dual-stage training reduces the overall matching cost by 83%, and symmetry-based matching further decreases the cost of individual metagraphs by 52%.

Bipartite graphs have been proven useful in modeling a wide range of relationship networks. Finding the maximum edge biclique within a bipartite graph is a well-known problem in graph theory and data mining, with numerous real-world applications across different domains. We propose a probabilistic algorithm for finding the maximum edge biclique using a Monte Carlo subspace clustering approach. Extensive experimentation with both artificial and real-world datasets shows that the algorithm is significantly better than the state-of-the-art technique. We prove that there are solid theoretical reasons for the algorithm's efficacy that manifest in a polynomial complexity of time and space.

Multi-Task Learning (MTL) can enhance the classifier's generalization performance by learning multiple related tasks simultaneously. Conventional MTL works under the offline or batch learning setting and suffers from the expensive training cost together with the poor scalability. To address such inefficiency issues, online learning technique has been applied to solve MTL problems. However, most existing algorithms for online MTL constrain task relatedness into a presumed structure via a single weight matrix, a strict restriction that does not always hold in practice. In this paper, we propose a general online MTL framework that overcomes this restriction by decomposing the weight matrix into two components: the first component captures the correlative structure among tasks in a low-rank subspace, and the second component identifies the personalized patterns for the outlier tasks. A projected gradient scheme is devised to learn such components adaptively. Theoretical analysis shows that the proposed algorithm can achieve a sub-linear regret with respect to the best linear model in hindsight. Experimental investigation on a number of real-world datasets also verifies the efficacy of our approach.

Instagram, an online photo-sharing platform, has gained increasing popularity. It allows users to take photos, apply digital filters and share them with friends instantaneously by using mobile devices.Instagram provides users with the functionality to associate their photos with points of interest, and it thus becomes feasible to study the association between points of interest and Instagram photos. However, no previous work studies the association. In this paper, we propose to study the problem of mapping Instagram photos to points of interest. To understand the problem, we analyze Instagram datasets, and report our findings, which also characterize the challenges of the problem. To address the challenges, we propose to model the mapping problem as a ranking problem, and develop a method to learn a ranking function by exploiting the textual, visual and user information of photos. To maximize the prediction effectiveness for textual and visual information, and incorporate the users' visiting preferences, we propose three subobjectives for learning the parameters of the proposed ranking function. Experimental results on two sets of Instagram data show that the proposed method substantially outperforms existing methods that are adapted to handle the problem.

CDR (Call Detail Records) data is more easily available than other network related data (such as GPS data) as most telecommunications service providers (TSPs) maintain such data. By analyzing it one can find mobility patterns of most of the population thus leading to efficient urban planning, disease and traffic control, etc. But its granularity is low as the latitude and longitude (lat-lon) of a cell tower is used as the current location of all mobile phones that are connected to the cell tower at that time. Granularity can range between 10s of metres to several kms depending on population density of a locality. This is one reason why, although there are many existing systems on visualizing mobility of people based on GPS data, there is hardly any existing system for CDR. We develop a Mobility Visualization System (MoVis) for visualizing mobility of people from their CDR records. First of all, given the CDR records of a user, we determine her stay regions (places where she stays for a significant amount of time). Trajectories of phone events (and lat-lon of cell towers) between stay regions are extracted as her trips. Start and end times of a trip are estimated using linear extrapolation. Based on the start and end times, temporal patterns are extracted. Trips with sufficient number of intermediate points are mapped to transport network that consists of train lines, bus routes and expressways. We use Kernel density estimation to visualize the most common path for a given origin and destination. Based on this we create a round-the-clock visualization of mobility of people over the entire city separately for weekdays and weekends. At the end we show the validation results.

The identification of network motifs has important applications in numerous domains, such as pattern detection in biological networks and graph analysis in digital circuits. However, mining network motifs is computationally challenging, as it requires enumerating subgraphs from a real-life graph, and computing the frequency of each subgraph in a large number of random graphs. In particular, existing solutions often require days to derive network motifs from biological networks with only a few thousand vertices. To address this problem, this paper presents a novel study on network motif discovery using Graphical Processing Units (GPUs). The basic idea is to employ GPUs to parallelize a large number of subgraph matching tasks in computing subgraph frequencies from random graphs, so as to reduce the overall computation time of network motif discovery. We explore the design space of GPU-based subgraph matching algorithms, with careful analysis of several crucial factors that affect the performance of GPU programs. Based on our analysis, we develop a GPU-based solution that (i) considerably differs from existing CPU-based methods, and (ii) exploits the strengths of GPUs in terms of parallelism while mitigating their limitations in terms of the computation power per GPU core. With extensive experiments on a variety of biological networks, we show that our solution is up to two orders of magnitude faster than the best CPU-based approach, and is around 20 times more cost-effective than the latter, when taking into account the monetary costs of the CPU and GPUs used.

Traditional online learning algorithms are designed for vector data only, which assume that the labels of all the training examples are provided. In this paper, we study graph classification where only limited nodes are chosen for labelling by selective sampling. Particularly, we first adapt a spectral-based graph regularization technique to derive a novel online learning linear algorithm which can handle graph data, although it still queries the labels of all nodes and thus is not preferred, as labelling is typically time-consuming. To address this issue, we then propose a new confidence-based query method for selective sampling. The theoretical result shows that our online learning algorithm with a fraction of queried labels can achieve a mistake bound comparable with the one learning on all labels of the nodes. In addition, the algorithm based on our proposed query strategy can achieve a mistake bound better than the one based on other query methods. However, our algorithm is conservative to update the model whenever error happens, which obviously wastes training labels that are valuable for the model. To take advantage of these labels, we further propose a novel aggressive algorithm, which can update the model aggressively even if no error occurs. The theoretical analysis shows that our aggressive approach can achieve a mistake bound better than its conservative and fully-supervised counterpart, with substantially fewer queried times. We empirically evaluate our algorithm on several real-world graph datasets and the experimental results demonstrate that our method is highly effective.

Cost-Sensitive Online Classification is recently proposed to directly online optimize two well-known cost-sensitive measures: (i) maximization of weighted sum of sensitivity and specificity, and (ii) minimization of weighted misclassification cost. However, the previous existing learning algorithms only utilized the first order information of the data stream. This is insufficient, as recent studies have proved that incorporating second order information could yield significant improvements on the prediction model. Hence, we propose a novel cost-sensitive online classification algorithm with adaptive regularization. We theoretically analyzed the proposed algorithm and empirically validated its effectiveness with extensive experiments. We also demonstrate the application of the proposed technique for solving several online anomaly detection tasks, showing that the proposed technique could be an effective tool to tackle cost-sensitive online classification tasks in various application domains.

With the rapid growth of location-based social networks, Point of Interest (POI) recommendation has become an important research problem. However, the scarcity of the check-in data, a type of implicit feedback data, poses a severe challenge for existing POI recommendation methods. Moreover, different types of context information about POIs are available and how to leverage them becomes another challenge. In this paper, we propose a ranking based geographical factorization method, called Rank-GeoFM, for POI recommendation, which addresses the two challenges. In the proposed model, we consider that the check-in frequency characterizes users' visiting preference and learn the factorization by ranking the POIs correctly. In our model, POIs both with and without check-ins will contribute to learning the ranking and thus the data sparsity problem can be alleviated. In addition, our model can easily incorporate different types of context information, such as the geographical influence and temporal influence. We propose a stochastic gradient descent based algorithm to learn the factorization. Experiments on publicly available datasets under both user-POI setting and user-time-POI setting have been conducted to test the effectiveness of the proposed method. Experimental results under both settings show that the proposed method outperforms the state-of-the-art methods significantly in terms of recommendation accuracy.

Differential privacy has recently become very promising in achieving data privacy guarantee. Typically, one can achieve ε-differential privacy by adding noise based on Laplace distribution to a query result. To reduce the noise magnitude for higher accuracy, various techniques have been proposed. They generally require high computational complexity, making them inapplicable to large-scale datasets. In this paper, we propose a novel orthogonal mechanism (OM) to represent a query set Q with a linear combination of a new query set Q, where Q consists of orthogonal query sets and is derived by exploiting the correlations between queries in Q. As a result of orthogonality of the derived queries, the proposed technique not only greatly reduces computational complexity, but also achieves better accuracy than the existing mechanisms. Extensive experimental results demonstrate the effectiveness and efficiency of the proposed technique.

Knowledge about population distribution of planning areas helps in making urban development decisions. Two important criteria are: "where do people live?" and "where do they work?" In this paper we propose methods to find home and workplaces from mobile network data. Home and work places are essential for discovery of mobility profiles of users. Validation of home and workplace prediction is not straight forward. We validate our methods using correlation with external data. Validation results show that even though a single cellular provider has only a portion of the entire population as its users, distribution of home and work places predicted using its mobile network data match that of government statistics. On the basis of this matching, we can have faith in distributions of more difficult statistics extracted from mobile network data which are difficult to obtain from external sources. We implemented an interactive system to show various distributions such as people living and working in different planning areas, and people working in different job sectors such as manufacturing. Interesting relationships are found by calculating joint distributions, e.g., Where do people, living in a planning area, work, and vice versa. Planning areas are ranked by the average distance traveled from home to work. Another interesting fact we extract is balance. Balance of a planning area is high if people live and work there, it is low if people living in a planning area work in other planning areas. We extend these statistics to regions which consist of many planning areas. The goal of this interactive system is to understand location semantics and mobility of users to aid in making urban development decisions. A video recording with subtitles is uploaded in http://www.youtube.com/watch?v=mo-7-DsCymw.

We present methods to predict and validate home and work places of anonymized users using their mobile network data. Knowledge of home and work place of a user is essential in order to find his (and overall population) mobility profiles. There are many methods that predict home and work places using GPS data. But unlike GPS data, mobile network data using GSM do not provide the exact location of a phone event. We use a novel criterion that combines an extracted feature from mobile data (i.e., Inactivity - no phone event for a given period of time) with open source data about location category % (i.e., Streetdirectory.com) to predict home location. Results show that the new criterion gives better prediction accuracy than inactivity alone. We predict work place using the idea that one goes to her work place on most of the weekdays but rarely on weekends. We validate our methods by comparing against the ground truth obtained from open source data. Validation results show that our proposed methods are about 25% more accurate than existing methods both for home and work place predictions.

One major problem of using location data collected from mobile cellular networks for mobility modelling is the oscillation phenomenon. An oscillation occurs when a mobile phone intermittently switches between cell towers instead of connecting to the nearest cell tower. For the purpose of mobility modeling, the location data needs to be cleansed to approximate the mobile device's actual location. However, this constitutes a challenge because the mobile device's true location is not known. In this paper, we study the oscillation resolution problem. We propose an algorithm framework called DECRE (Detect, Expand, Check, Remove) to detect and remove oscillation logs. To make informed decisions DECRE includes four steps: Detect, to identify log sequences that may contain oscillation using a few heuristics based on the concepts of stable period and moving at impossible speed, Expand, to look before and after suspicious records to gain more information, Check, to check whether a cell tower is observed repeatedly (which is a strong indication of oscillation), and Remove, resolving oscillation by selecting a cell tower to approximate the mobile device's actual location. Our experimental results on travel diaries show that our oscillation resolution approach is able to remove records that are far from mobile device's ground-truth locations, improve the quality of the location data, and performs better than an existing method. Our performance study on large scale cell tower data shows that the MapReduce implementation of our approach is able to process 1 Terabyte of cell tower data in five hours using a small cluster.

Traditional active learning encounters a cold start issue when very few labelled examples are present for learning a decent initial classifier. Its poor quality subsequently affects selection of the next query and stability of the iterative learning process, resulting in more annotation effort from a domain expert. To address this issue, this paper presents a novel class augmentation technique, which enhances each class's representation which initially consists of only limited set of labelled examples. Our augmentation employs a connectivity-based influence computation algorithm with an incorporated decaying mechanism for the unlabelled samples. Besides augmentation, our method also introduces structure preserving oversampling to correct class imbalance. Extensive experiments on ten publicly available data sets demonstrate the effectiveness of our proposed method over existing state-of-the-art methods. Moreover, our proposed modules perform at the fundamental data level without any requirement to modify the well-established standard machine learning tools.

This paper proposes a novel Integrated Oversampling (INOS) method that can handle highly imbalanced time series classification. We introduce an enhanced structure preserving oversampling (ESPO) technique and synergistically combine it with interpolation-based oversampling. ESPO is used to generate a large percentage of the synthetic minority samples based on multivariate Gaussian distribution, by estimating the covariance structure of the minority-class samples and by regularizing the unreliable eigen spectrum. To protect the key original minority samples, we use an interpolation-based technique to oversample a small percentage of synthetic population. By preserving the main covariance structure and intelligently creating protective variances in the trivial eigen dimensions, ESPO effectively expands the synthetic samples into the void area in the data space without being too closely tied with existing minority-class samples. This also addresses a key challenge for applying oversampling for imbalanced time series classification, i.e., maintaining the correlation between consecutive values through preserving the main covariance structure. Extensive experiments based on seven public time series data sets demonstrate that our INOS approach, used with support vector machines (SVM), achieved better performance over existing oversampling methods as well as state-of-the-art methods in time series classification.

This paper presents a novel structure preserving over sampling (SPO) technique for classifying imbalanced time series data. SPO generates synthetic minority samples based on multivariate Gaussian distribution by estimating the covariance structure of the minority class and regularizing the unreliable eigen spectrum. By preserving the main covariance structure and intelligently creating protective variances in the trivial eigen feature dimensions, the synthetic samples expand effectively into the void area in the data space without being too closely tied with existing minority-class samples. Extensive experiments based on several public time series datasets demonstrate that our proposed SPO in conjunction with support vector machines can achieve better performances than existing over sampling methods and state-of-the-art methods in time series classification.

Learning from positive and unlabeled examples (PU learning) has been investigated in recent years as an alternative learning model for dealing with situations where negative training examples are not available. It has many real world applications, but it has yet to be applied in the data stream environment where it is highly possible that only a small set of positive data and no negative data is available. An important challenge is to address the issue of concept drift in the data stream environment, which is not easily handled by the traditional PU learning techniques. This paper studies how to devise PU learning techniques for the data stream environment. Unlike existing data stream classification methods that assume both positive and negative training data are available for learning, we propose a novel PU learning technique LELC (PU Learning by Extracting Likely positive and negative micro-Clusters) for document classification. LELC only requires a small set of positive examples and a set of unlabeled examples which is easily obtainable in the data stream environment to build accurate classifiers. Experimental results show that LELC is a PU learning method that can effectively address the issues in the data stream environment with significantly better speed and accuracy on capturing concept drift than the existing state-of-the-art PU learning techniques.

In this paper we introduce a novel technique for classifying text citations without labeled training examples. We first utilize the search results of a general search engine as original training data. We then proposed a mutually reinforcing learning algorithm (MRL) to mine the classification knowledge and to "clean" the training data. With the help of a set of established domain-specific ontological terms or keywords, the MRL mining step derives the relevant classification knowledge. The MRL cleaning step then builds a naive Bayes classifier based on the mined classification knowledge and tries to clean the training set. The MRL algorithm is iteratively applied until a clean training set is obtained. We show the effectiveness of the proposed technique in the classification of biomedical citations from a large medical literature database.

We study the problem of building text classifiers using positive and unlabeled examples. The key feature of this problem is that there is no negative example for learning. Recently, a few techniques for solving this problem were proposed in the literature. These techniques are based on the same idea, which builds a classifier in two steps. Each existing technique uses a different method for each step. We first introduce some new methods for the two steps, and perform a comprehensive evaluation of all possible combinations of methods of the two steps. We then propose a more principled approach to solving the problem based on a biased formulation of SVM, and show experimentally that it is more accurate than the existing techniques.

A commercial Web page typically contains many information blocks. Apart from the main content blocks, it usually has such blocks as navigation panels, copyright and privacy notices, and advertisements (for business purposes and for easy user access). We call these blocks that are not the main content blocks of the page the noisy blocks. We show that the information contained in these noisy blocks can seriously harm Web data mining. Eliminating these noises is thus of great importance. In this paper, we propose a noise elimination technique based on the following observation: In a given Web site, noisy blocks usually share some common contents and presentation styles, while the main content blocks of the pages are often diverse in their actual contents and/or presentation styles. Based on this observation, we propose a tree structure, called Style Tree, to capture the common presentation styles and the actual contents of the pages in a given Web site. By sampling the pages of the site, a Style Tree can be built for the site, which we call the Site Style Tree (SST). We then introduce an information based measure to determine which parts of the SST represent noises and which parts represent the main contents of the site. The SST is employed to detect and eliminate noises in any Web page of the site by mapping this page to the SST. The proposed technique is evaluated with two data mining tasks, Web page clustering and classification. Experimental results show that our noise elimination technique is able to improve the mining results significantly.

Internet search is one of the most important applications of the Web. A search engine takes the user's keywords to retrieve and to rank those pages that contain the keywords. One shortcoming of existing search techniques is that they do not give due consideration to the micro-structures of a Web page. A Web page is often populated with a number of small information units, which we call micro information units (MIU). Each unit focuses on a specific topic and occupies a specific area of the page. During the search, if all the keywords in the user query occur in a single MIU of a page, the top ranking results returned by a search engine are generally relevant and useful. However, if the query words scatter at different MIUs in a page, the pages returned can be quite irrelevant (which causes low precision). The reason for this is that although a page has information on individual MIUs, it may not have information on their intersections. In this paper, we propose a technique to solve this problem. At the off-line pre-processing stage, we segment each page to identify the MIUs in the page, and index the keywords of the page according to the MIUs in which they occur. In searching, our retrieval and ranking algorithm utilizes this additional information to return those most relevant pages. Experimental results show that this method is able to significantly improve the search precision.

Text categorization or classification is the automated assigning of text documents to pre-defined classes based on their contents. This problem has been studied in information retrieval, machine learning and data mining. So far, many effective techniques have been proposed. However, most techniques are based on some underlying models and/or assumptions. When the data fits the model well, the classification accuracy will be high. However, when the data does not fit the model well, the classification accuracy can be very low. In this paper, we propose a refinement approach to dealing with this problem of model misfit. We show that we do not need to change the classification technique itself (or its underlying model) to make it more flexible. Instead, we propose to use successive refinements of classification on the training data to correct the model misfit. We apply the proposed technique to improve the classification performance of two simple and efficient text classifiers, the Rocchio classifier and the naïve Bayesian classifier. These techniques are suitable for very large text collections because they allow the data to reside on disk and need only one scan of the data to build a text classifier. Extensive experiments on two benchmark document corpora show that the proposed technique is able to improve text categorization accuracy of the two techniques dramatically. In particular, our refined model is able to improve the naïve Bayesian or Rocchio classifier's prediction performance by 45% on average.