Enhancing sequential recommendation with contrastive Generative Adversarial Network

Sequential recommendation models a user’s historical sequence to predict future items. Existing studies utilize deep learning methods and contrastive learning for data augmentation to alleviate data sparsity. However, these existing methods cannot learn accurate high-quality item representations while augmenting data. In addition, they usually ignore data noise and user cold-start issues. To solve the above issues, we investigate the possibility of Generative Adversarial Network (GAN) with contrastive learning for sequential recommendation to balance data sparsity and noise. Specifically, we propose a new framework, Enhanced Contrastive Learning with Generative Adversarial Network for Sequential Recommendation (ECGAN-Rec), which models the training process as a GAN and recommendation task as the main task of the discriminator. We design a sequence augmentation module and a contrastive GAN module to implement both data-level and model-level augmentations. In addition, the contrastive GAN learns more accurate high-quality item representations to alleviate data noise after data augmentation. Furthermore, we propose an enhanced Transformer recommender based on GAN to optimize the performance of the model. Experimental results on three open datasets validate the efficiency and effectiveness of the proposed model and the ability of the model to balance data noise and data sparsity. Specifically, the improvement of ECGAN-Rec in two evaluation metrics (HR@N and NDCG@N) compared to the state-of-the-art model performance on the Beauty, Sports and Yelp datasets are 34.95%, 36.68%, and 13.66%, respectively. Our implemented model is available via https://github.com/nishawn/ECGANRec-master.     

gDART: Improving rumor verification in social media with Discrete Attention Representations

Due to the harmful impact of fabricated information on social media, many rumor verification techniques have been introduced in recent years. Advanced techniques like multi-task learning (MTL), shared-private models suffer from many strategic limitations that restrict their capability of veracity identification on social media. These models are often reliant on multiple tasks for the primary targeted objective. Even the most recent deep neural network (DNN) models like VRoC, Hierarchical-PSV, StA-HiTPLAN etc. based on VAE, GCN, Transformer respectively with improved modification are able to perform good on veracity identification task but with the help of additional auxiliary information, mostly. However, their rise is still not substantial with respect to the proposed model even though the proposed model is not using any additional information. To come up with an improved DNN model architecture, we introduce globally Discrete Attention Representations from Transformers (gDART). Discrete-Attention mechanism in gDART is capable of capturing multifarious correlations veiled among the sequence of words which existing DNN models including Transformer often overlook. Our proposed framework uses a Branch-CoRR Attention Network to extract highly informative features in branches, and employs Feature Fusion Network Component to identify deep embedded features and use them to make enhanced identification of veracity of an unverified claim. Moreover, to achieve its goal, gDART is not dependent on any costly auxiliary resource but on an unsupervised learning process. Extensive experiments reveal that gDART marks a considerable performance gain in veracity identification task over state-of-the-art models on two real world rumor datasets. gDART reports a gain of 36.76%, 40.85% on standard benchmark metrics.     

Dynamic intent-aware iterative denoising network for session-based recommendation

Session-based recommendation aims to predict items that a user will interact with based on historical behaviors in anonymous sessions. It has long faced two challenges: (1) the dynamic change of user intents which makes user preferences towards items change over time; (2) the uncertainty of user behaviors which adds noise to hinder precise preference learning. They jointly preclude recommender system from capturing real intents of users. Existing methods have not properly solved these problems since they either ignore many useful factors like the temporal information when building item embeddings, or do not explicitly filter out noisy clicks in sessions. To tackle above issues, we propose a novel Dynamic Intent-aware Iterative Denoising Network (DIDN) for session-based recommendation. Specifically, to model the dynamic intents of users, we present a dynamic intent-aware module that incorporates item-aware, user-aware and temporal-aware information to learn dynamic item embeddings. A novel iterative denoising module is then devised to explicitly filter out noisy clicks within a session. In addition, we mine collaborative information to further enrich the session semantics. Extensive experimental results on three real-world datasets demonstrate the effectiveness of the proposed DIDN. Specifically, DIDN obtains improvements over the best baselines by 1.66%, 1.75%, and 7.76% in terms of P@20 and 1.70%, 2.20%, and 10.48% in terms of MRR@20 on all datasets.     

HCA: Hierarchical Compare Aggregate model for question retrieval in community question answering

We address the problem of finding similar historical questions that are semantically equivalent or relevant to an input query question in community question-answering (CQA) sites. One of the main challenges for this task is that questions are usually too long and often contain peripheral information in addition to the main goals of the question. To address this problem, we propose an end-to-end Hierarchical Compare Aggregate (HCA) model that can handle this problem without using any task-specific features. We first split questions into sentences and compare every sentence pair of the two questions using a proposed Word-Level-Compare-Aggregate model called WLCA-model and then the comparison results are aggregated with a proposed Sentence-Level-Compare-Aggregate model to make the final decision. To handle the insufficient training data problem, we propose a sequential transfer learning approach to pre-train the WLCA-model on a large paraphrase detection dataset. Our experiments on two editions of the Semeval benchmark datasets and the domain-specific AskUbuntu dataset show that our model outperforms the state-of-the-art models.     

Group event recommendation based on graph multi-head attention network combining explicit and implicit information

In event-based social networks (EBSN), group event recommendation has become an important task for groups to quickly find events that they are interested in. Existing methods on group event recommendation either consider just one type of information, explicit or implicit, or separately model the explicit and implicit information. However, these methods often generate a problem of data sparsity or of model vector redundancy. In this paper, we present a Graph Multi-head Attention Network (GMAN) model for group event recommendation that integrates the explicit and implicit information in EBSN. Specifically, we first construct a user-explicit graph based on the user's explicit information, such as gender, age, occupation and the interactions between users and events. Then we build a user-implicit graph based on the user's implicit information, such as friend relationships. The incorporated both explicit and implicit information can effectively describe the user's interests and alleviate the data sparsity problem. Considering that there may be a correlation between the user's explicit and implicit information in EBSN, we take the user's explicit vector representation as the input of the implicit information aggregation when modeling with graph neural networks. This unified user modeling can solve the aforementioned problem of user model vector redundancy and is also suitable for event modeling. Furthermore, we utilize a multi-head attention network to learn richer implicit information vectors of users and events from multiple perspectives. Finally, in order to get a higher level of group vector representation, we use a vanilla attention mechanism to fuse different user vectors in the group. Through experimenting on two real-world Meetup datasets, we demonstrate that GMAN model consistently outperforms state-of-the-art methods on group event recommendation.     

SAT-Geo: A social sensing based content-only approach to geolocating abnormal traffic events using syntax-based probabilistic learning

Social sensing has become an emerging and pervasive sensing paradigm to collect timely observations of the physical world from human sensors. In this paper, we study the problem of geolocating abnormal traffic events using social sensing. Our goal is to infer the location (i.e., geographical coordinates) of the abnormal traffic events by exploring the location entities from the content of social media posts. Two critical challenges exist in solving our problem: (i) how to accurately identify the location entities related to the abnormal traffic event from the content of social media posts? (ii) How to accurately estimate the geographic coordinates of the abnormal traffic event from the set of identified location entities? To address the above challenges, we develop a Social sensing based Abnormal Traffic Geolocalization (SAT-Geo) framework to accurately estimate the geographic coordinates of abnormal traffic events by exploring the syntax-based patterns in the content of social media posts and the geographic information associated with the location entities from the social media posts. We evaluate the SAT-Geo framework on three real-world Twitter datasets collected from New York City, Los Angeles, and London. Evaluation results demonstrate that SAT-Geo significantly outperforms state-of-the-art baselines by effectively identifying location entities related to the abnormal traffic events and accurately estimating the geographic coordinates of the events.     

个性化信息推荐服务中用户潜在兴趣挖掘研究

如何准确分析用户行为,向用户提供满意的网页信息,一直以来都是个性化信息推荐系统设计的目标。本文在分析现有个性化信息推荐模型的基础上,针对以往研究在推荐兴趣时仅根据语义相关度进行协助性信息推荐,而忽略用户行为规律所包含的潜在兴趣信息的不足,尝试提出一个结合Web语义挖掘和FP-tree规则发现技术的个性化信息推荐模型。该模型利用本体对语义的明确化描述,在挖掘用户行为信息时获取用户兴趣偏好的语义信息,并利用FP-tree技术根据以获取的语义信息推理出用户兴趣行为模式,从而在信息推荐时不仅能准确理解用户兴趣偏好,也能根据用户潜在兴趣规律,推荐给用户更全面的网页信息。     

Integration of big-data ERP and business analytics (BA)

Technology advancements in cloud computing, big data systems, No-SQL database, cognitive systems, deep learning, and other artificial intelligence techniques make the integration of traditional ERP transaction data and big data streaming from various social media platforms and Internet of Things (IOTs) into a unified analytics system not only feasible but also inevitable. Two steps are prominent for this integration. The first, coined as forming the big-data ERP, is the integration of traditional ERP transaction data and the big data and the second is to integrate the big-data ERP with business analytics (BA). As ERP implementers and BA users are facing various challenges, managers responsible for this big-data ERP-BA integration are also seriously challenged. To help them deal with these challenges, we develop the SIST model (including Strategic alignment, Intellectual and Social capital integration, and Technology integration) and propose that this integration is an evolving portfolio with various maturity levels for different business functions, likely leading to sustainable competitive advantages.     

Building user profiles based on sequences for content and collaborative filtering

Modeling user profiles is a necessary step for most information filtering systems – such as recommender systems – to provide personalized recommendations. However, most of them work with users or items as vectors, by applying different types of mathematical operations between them and neglecting sequential or content-based information. Hence, in this paper we study how to propose an adaptive mechanism to obtain user sequences using different sources of information, allowing the generation of hybrid recommendations as a seamless, transparent technique from the system viewpoint. As a proof of concept, we develop the Longest Common Subsequence (LCS) algorithm as a similarity metric to compare the user sequences, where, in the process of adapting this algorithm to recommendation, we include different parameters to control the efficiency by reducing the information used in the algorithm (preference filter), to decide when a neighbor is considered useful enough to be included in the process (confidence filter), to identify whether two interactions are equivalent (δ-matching threshold), and to normalize the length of the LCS in a bounded interval (normalization functions). These parameters can be extended to work with any type of sequential algorithm.We evaluate our approach with several state-of-the-art recommendation algorithms using different evaluation metrics measuring the accuracy, diversity, and novelty of the recommendations, and analyze the impact of the proposed parameters. We have found that our approach offers a competitive performance, outperforming content, collaborative, and hybrid baselines, and producing positive results when either content- or rating-based information is exploited.     

Dimensions as Virtual Items: Improving the predictive ability of top-N recommender systems

Traditionally, recommender systems for the web deal with applications that have two dimensions, users and items. Based on access data that relate these dimensions, a recommendation model can be built and used to identify a set of N items that will be of interest to a certain user. In this paper we propose a multidimensional approach, called DaVI (Dimensions as Virtual Items), that consists in inserting contextual and background information as new user–item pairs. The main advantage of this approach is that it can be applied in combination with several existing two-dimensional recommendation algorithms. To evaluate its effectiveness, we used the DaVI approach with two different top-N recommender algorithms, Item-based Collaborative Filtering and Association Rules based, and ran an extensive set of experiments in three different real world data sets. In addition, we have also compared our approach to the previously introduced combined reduction and weight post-filtering approaches. The empirical results strongly indicate that our approach enables the application of existing two-dimensional recommendation algorithms in multidimensional data, exploiting the useful information of these data to improve the predictive ability of top-N recommender systems.     

A Convolutional Attention Network for Unifying General and Sequential Recommenders

General recommenders and sequential recommenders are two modeling paradigms of recommender. The main focus of a general recommender is to identify long-term user preferences, while the user’s sequential behaviors are ignored and sequential recommenders try to capture short-term user preferences by exploring item-to-item relations, failing to consider general user preferences. Recently, better performance improvement is reported by combining these two types of recommenders. However, most of the previous works typically treat each item separately and assume that each user–item interaction in a sequence is independent. This may be a too simplistic assumption, since there may be a particular purpose behind buying the successive item in a sequence. In fact, a user makes a decision through two sequential processes, i.e., start shopping with a particular intention and then select a specific item which satisfies her/his preferences under this intention. Moreover, different users usually have different purposes and preferences, and the same user may have various intentions. Thus, different users may click on the same items with an attention on a different purpose. Therefore, a user’s behavior pattern is not completely exploited in most of the current methods and they neglect the distinction between users’ purposes and their preferences. To alleviate those problems, we propose a novel method named, CAN, which takes both users’ purposes and preferences into account for the next-item recommendation. We propose to use Purpose-Specific Attention Unit (PSAU) in order to discriminately learn the representations of user purpose and preference. The experimental results on real-world datasets demonstrate the advantages of our approach over the state-of-the-art methods.     

Efficient generation of spatiotemporal relationships from spatial data streams and static data

Recently, a massive amount of position-annotated data is being generated in a stream fashion. Also, massive amounts of static data including spatial features are collected and made available. In the Internet of Things (IoT) environments, various applications can get benefits by utilizing spatial data streams and static data. Therefore, IoT applications typically require stream processing and reasoning capabilities that extract information from low-level data. Particularly for sophisticated stream processing and reasoning, spatiotemporal relationship (SR) generation from spatial data streams and static data must be preceded. However, existing techniques mostly focus solely on direct processing of sensing data or generation of spatial relationships from static data. In this paper, we first address the importance of SRs between spatial data streams and static data and then propose an efficient approach of deriving SRs in real-time. We design a novel R-tree-based index with Representative Rectangles (RRs) called R3 index and devise an algorithm that leverages relationships and distances between RRs to generate SRs. To verify the effectiveness and efficiency of the proposed approach, we performed experiments using real-world datasets. Through the results of the experiments, we confirmed the superiority of the proposed approach.     

Personalization of tagging systems

Social media systems have encouraged end user participation in the Internet, for the purpose of storing and distributing Internet content, sharing opinions and maintaining relationships. Collaborative tagging allows users to annotate the resulting user-generated content, and enables effective retrieval of otherwise uncategorised data. However, compared to professional web content production, collaborative tagging systems face the challenge that end-users assign tags in an uncontrolled manner, resulting in unsystematic and inconsistent metadata.This paper introduces a framework for the personalization of social media systems. We pinpoint three tasks that would benefit from personalization: collaborative tagging, collaborative browsing and collaborative search. We propose a ranking model for each task that integrates the individual user’s tagging history in the recommendation of tags and content, to align its suggestions to the individual user preferences. We demonstrate on two real data sets that for all three tasks, the personalized ranking should take into account both the user’s own preference and the opinion of others.     

基于用户动态画像的学术新媒体信息精准推荐模型研究

【目的/意义】构建基于用户动态画像的学术新媒体信息精准推荐模型是满足学术新媒体用户对学术信息 资源精准化、个性化与专业化的要求,同时也是提高学术信息流转效率以及价值增值的有效途径。【方法/过程】在 探究学术新媒体信息流转模型的基础上,进一步分析学术新媒体用户需求与分层画像,重构学术新媒体用户画像 步骤,构建基于用户动态画像的学术新媒体信息精准推荐模型。【结果/结论】基于用户动态画像的学术新媒体信息 精准推荐模型能够实现学术信息资源与用户的精准对接,提升用户忠诚度,更好地服务科研工作者的学术活动。 【创新/局限】从理论框架角度分析与构建学术新媒体信息推荐模型,后续将重点研究模型的技术实现与实践应用。     

TSVFN: Two-Stage Visual Fusion Network for multimodal relation extraction

Multimodal relation extraction is a critical task in information extraction, aiming to predict the class of relations between head and tail entities from linguistic sequences and related images. However, the current works are vulnerable to less relevant visual objects detected from images and are not able to sufficiently fuse visual information into text pre-trained models. To overcome these problems, we propose a Two-Stage Visual Fusion Network (TSVFN) that employs the multimodal fusion approach in vision-enhanced entity relation extraction. In the first stage, we design multimodal graphs, whose novelty lies mainly in transforming the sequence learning into the graph learning. In the second stage, we merge the transformer-based visual representation into the text pre-trained model by a multi-scale cross-model projector. Specifically, two multimodal fusion operations are implemented inside the pre-trained model respectively. We finally accomplish deep interaction of multimodal multi-structured data in two fusion stages. Extensive experiments are conducted on a dataset (MNRE), our model outperforms the current state-of-the-art method by 1.76%, 1.52%, 1.29%, and 1.17% in terms of accuracy, precision, recall, and F1 score, respectively. Moreover, our model also achieves excellent results under the condition of fewer samples.     

Adaptive dictionary and structure learning for unsupervised feature selection

Unsupervised feature selection is very attractive in many practical applications, as it needs no semantic labels during the learning process. However, the absence of semantic labels makes the unsupervised feature selection more challenging, as the method can be affected by the noise, redundancy, or missing in the originally extracted features. Currently, most methods either consider the influence of noise for sparse learning or think over the internal structure information of the data, leading to suboptimal results. To relieve these limitations and improve the effectiveness of unsupervised feature selection, we propose a novel method named Adaptive Dictionary and Structure Learning (ADSL) that conducts spectral learning and sparse dictionary learning in a unified framework. Specifically, we adaptively update the dictionary based on sparse dictionary learning. And, we also introduce the spectral learning method of adaptive updating affinity matrix. While removing redundant features, the intrinsic structure of the original data can be retained. In addition, we adopt matrix completion in our framework to make it competent for fixing the missing data problem. We validate the effectiveness of our method on several public datasets. Experimental results show that our model not only outperforms some state-of-the-art methods on complete datasets but also achieves satisfying results on incomplete datasets.     

Information extraction with automatic knowledge expansion

POSIE (POSTECH Information Extraction System) is an information extraction system which uses multiple learning strategies, i.e., SmL, user-oriented learning, and separate-context learning, in a question answering framework. POSIE replaces laborious annotation with automatic instance extraction by the SmL from structured Web documents, and places the user at the end of the user-oriented learning cycle. Information extraction as question answering simplifies the extraction procedures for a set of slots. We introduce the techniques verified on the question answering framework, such as domain knowledge and instance rules, into an information extraction problem. To incrementally improve extraction performance, a sequence of the user-oriented learning and the separate-context learning produces context rules and generalizes them in both the learning and extraction phases. Experiments on the “continuing education” domain initially show that the F1-measure becomes 0.477 and recall 0.748 with no user training. However, as the size of the training documents grows, the F1-measure reaches beyond 0.75 with recall 0.772. We also obtain F-measure of about 0.9 for five out of seven slots on “job offering” domain.     

Efficient federated item similarity model for privacy-preserving recommendation

Previous federated recommender systems are based on traditional matrix factorization, which can improve personalized service but are vulnerable to gradient inference attacks. Most of them adopt model averaging to fit the data heterogeneity of federated recommender systems, requiring more training costs. To address privacy and efficiency, we propose an efficient federated item similarity model for the heterogeneous recommendation, called FedIS, which can train a global item-based collaborative filtering model to eliminate user feature dependencies. Specifically, we extend the neural item similarity model to the federated model, where each client only locally optimizes the shared item feature matrix. We then propose a fast-convergent federated aggregation method inspired by meta-learning to address heterogeneous user updates and accelerate the convergence of global training. Furthermore, we propose a two-stage perturbation method to protect both local training and transmission while reducing communication costs. Finally, extensive experiments on four real-world datasets validate that FedIS can provide more competitive performance on federated recommendations. Our proposed method also shows significant training efficiency with less performance degradation.     

Meta-prompt based learning for low-resource false information detection

The wide spread of false information has detrimental effects on society, and false information detection has received wide attention. When new domains appear, the relevant labeled data is scarce, which brings severe challenges to the detection. Previous work mainly leverages additional data or domain adaptation technology to assist detection. The former would lead to a severe data burden; the latter underutilizes the pre-trained language model because there is a gap between the downstream task and the pre-training task, which is also inefficient for model storage because it needs to store a set of parameters for each domain. To this end, we propose a meta-prompt based learning (MAP) framework for low-resource false information detection. We excavate the potential of pre-trained language models by transforming the detection tasks into pre-training tasks by constructing template. To solve the problem of the randomly initialized template hindering excavation performance, we learn optimal initialized parameters by borrowing the benefit of meta learning in fast parameter training. The combination of meta learning and prompt learning for the detection is non-trivial: Constructing meta tasks to get initialized parameters suitable for different domains and setting up the prompt model’s verbalizer for classification in the noisy low-resource scenario are challenging. For the former, we propose a multi-domain meta task construction method to learn domain-invariant meta knowledge. For the latter, we propose a prototype verbalizer to summarize category information and design a noise-resistant prototyping strategy to reduce the influence of noise data. Extensive experiments on real-world data demonstrate the superiority of the MAP in new domains of false information detection.