Search

Constructing commonsense knowledge graphs (CKGs) has attracted wide research attention due to its significant importance in cognitive intelligence. Nevertheless, existing CKGs are typically oriented to English, limiting the research in non-English languages. Meanwhile, the emergence of foundation models like ChatGPT and GPT-4 has shown promising intelligence with the help of reinforcement learning from human feedback. Under the background, in this paper, we utilize foundation models to construct a Chinese CKG, named Snowman. Specifically, we distill different types of commonsense head items from ChatGPT, and continue to use it to collect tail items with respect to the head items and pre-defined relations. Based on the preliminary analysis, we find the negative commonsense knowledge distilled by ChatGPT achieves lower human acceptance compared to other knowledge. Therefore, we design a simple yet effective self-instruct filtering strategy to filter out invalid negative commonsense. Overall, the constructed Snowman covers more than ten million Chinese commonsense triples, making it the largest Chinese CKG. Moreover, human studies show the acceptance of Snowman achieves 90.6\%, indicating the high-quality triples distilled by the cutting-edge foundation model. We also conduct experiments on commonsense knowledge models to show the usability and effectiveness of our Snowman., tech report

Deep learning and symbolic learning are two frequently employed methods in Sequential Recommendation (SR). Recent neural-symbolic SR models demonstrate their potential to enable SR to be equipped with concurrent perception and cognition capacities. However, neural-symbolic SR remains a challenging problem due to open issues like representing users and items in logical reasoning. In this paper, we combine the Deep Neural Network (DNN) SR models with logical reasoning and propose a general framework named Sequential Recommendation with Probabilistic Logical Reasoning (short for SR-PLR). This framework allows SR-PLR to benefit from both similarity matching and logical reasoning by disentangling feature embedding and logic embedding in the DNN and probabilistic logic network. To better capture the uncertainty and evolution of user tastes, SR-PLR embeds users and items with a probabilistic method and conducts probabilistic logical reasoning on users' interaction patterns. Then the feature and logic representations learned from the DNN and logic network are concatenated to make the prediction. Finally, experiments on various sequential recommendation models demonstrate the effectiveness of the SR-PLR.

The self-attention mechanism, which equips with a strong capability of modeling long-range dependencies, is one of the extensively used techniques in the sequential recommendation field. However, many recent studies represent that current self-attention based models are low-pass filters and are inadequate to capture high-frequency information. Furthermore, since the items in the user behaviors are intertwined with each other, these models are incomplete to distinguish the inherent periodicity obscured in the time domain. In this work, we shift the perspective to the frequency domain, and propose a novel Frequency Enhanced Hybrid Attention Network for Sequential Recommendation, namely FEARec. In this model, we firstly improve the original time domain self-attention in the frequency domain with a ramp structure to make both low-frequency and high-frequency information could be explicitly learned in our approach. Moreover, we additionally design a similar attention mechanism via auto-correlation in the frequency domain to capture the periodic characteristics and fuse the time and frequency level attention in a union model. Finally, both contrastive learning and frequency regularization are utilized to ensure that multiple views are aligned in both the time domain and frequency domain. Extensive experiments conducted on four widely used benchmark datasets demonstrate that the proposed model performs significantly better than the state-of-the-art approaches., 11 pages, 7 figures, The 46th International ACM SIGIR Conference on Research and Development in Information Retrieval

Contrastive Learning (CL) performances as a rising approach to address the challenge of sparse and noisy recommendation data. Although having achieved promising results, most existing CL methods only perform either hand-crafted data or model augmentation for generating contrastive pairs to find a proper augmentation operation for different datasets, which makes the model hard to generalize. Additionally, since insufficient input data may lead the encoder to learn collapsed embeddings, these CL methods expect a relatively large number of training data (e.g., large batch size or memory bank) to contrast. However, not all contrastive pairs are always informative and discriminative enough for the training processing. Therefore, a more general CL-based recommendation model called Meta-optimized Contrastive Learning for sequential Recommendation (MCLRec) is proposed in this work. By applying both data augmentation and learnable model augmentation operations, this work innovates the standard CL framework by contrasting data and model augmented views for adaptively capturing the informative features hidden in stochastic data augmentation. Moreover, MCLRec utilizes a meta-learning manner to guide the updating of the model augmenters, which helps to improve the quality of contrastive pairs without enlarging the amount of input data. Finally, a contrastive regularization term is considered to encourage the augmentation model to generate more informative augmented views and avoid too similar contrastive pairs within the meta updating. The experimental results on commonly used datasets validate the effectiveness of MCLRec., 11 Pages,8 figures

In recommendation scenarios, there are two long-standing challenges, i.e., selection bias and data sparsity, which lead to a significant drop in prediction accuracy for both Click-Through Rate (CTR) and post-click Conversion Rate (CVR) tasks. To cope with these issues, existing works emphasize on leveraging Multi-Task Learning (MTL) frameworks (Category 1) or causal debiasing frameworks (Category 2) to incorporate more auxiliary data in the entire exposure/inference space D or debias the selection bias in the click/training space O. However, these two kinds of solutions cannot effectively address the not-missing-at-random problem and debias the selection bias in O to fit the inference in D. To fill the research gaps, we propose a Direct entire-space Causal Multi-Task framework, namely DCMT, for post-click conversion prediction in this paper. Specifically, inspired by users' decision process of conversion, we propose a new counterfactual mechanism to debias the selection bias in D, which can predict the factual CVR and the counterfactual CVR under the soft constraint of a counterfactual prior knowledge. Extensive experiments demonstrate that our DCMT can improve the state-of-the-art methods by an average of 1.07% in terms of CVR AUC on the five offline datasets and 0.75% in terms of PV-CVR on the online A/B test (the Alipay Search). Such improvements can increase millions of conversions per week in real industrial applications, e.g., the Alipay Search., Comment: 13 pages; Accepted by ICDE 2023

Sequential recommendation (SR) aims to model user preferences by capturing behavior patterns from their item historical interaction data. Most existing methods model user preference in the time domain, omitting the fact that users' behaviors are also influenced by various frequency patterns that are difficult to separate in the entangled chronological items. However, few attempts have been made to train SR in the frequency domain, and it is still unclear how to use the frequency components to learn an appropriate representation for the user. To solve this problem, we shift the viewpoint to the frequency domain and propose a novel Contrastive Enhanced \textbf{SLI}de Filter \textbf{M}ixEr for Sequential \textbf{Rec}ommendation, named \textbf{SLIME4Rec}. Specifically, we design a frequency ramp structure to allow the learnable filter slide on the frequency spectrums across different layers to capture different frequency patterns. Moreover, a Dynamic Frequency Selection (DFS) and a Static Frequency Split (SFS) module are proposed to replace the self-attention module for effectively extracting frequency information in two ways. DFS is used to select helpful frequency components dynamically, and SFS is combined with the dynamic frequency selection module to provide a more fine-grained frequency division. Finally, contrastive learning is utilized to improve the quality of user embedding learned from the frequency domain. Extensive experiments conducted on five widely used benchmark datasets demonstrate our proposed model performs significantly better than the state-of-the-art approaches. Our code is available at https://github.com/sudaada/SLIME4Rec.

Sequential Recommendation aims to predict the next item based on user behaviour. Recently, Self-Supervised Learning (SSL) has been proposed to improve recommendation performance. However, most of existing SSL methods use a uniform data augmentation scheme, which loses the sequence correlation of an original sequence. To this end, in this paper, we propose a Learnable Model Augmentation self-supervised learning for sequential Recommendation (LMA4Rec). Specifically, LMA4Rec first takes model augmentation as a supplementary method for data augmentation to generate views. Then, LMA4Rec uses learnable Bernoulli dropout to implement model augmentation learnable operations. Next, self-supervised learning is used between the contrastive views to extract self-supervised signals from an original sequence. Finally, experiments on three public datasets show that the LMA4Rec method effectively improves sequential recommendation performance compared with baseline methods.

Rare-earth-containing ultrastable Y zeolite (ReUSY) was modified by oxalic acid solution leaching treatment and applied in industrial units for catalytic olefin removal from aromatic hydrocarbons. The porous structure and amount of acidity of the modified ReUSY (denoted as ReUSY-x, where x indicated the amount of oxalic acid in solution) could be tuned by different concentrations of oxalic acid solution, and the ReUSY-x samples exhibited different catalytic performances. Based on the best catalytic performance of ReUSY-25, an industrial catalyst was prepared and applied in industrial units for catalytic olefin removal. The industrial catalyst exhibited excellent activity and regeneration stability with a long lifetime of about 2 years, which was about 13 times longer than that of activated clay.

Graph Convolution Network (GCN) has been widely applied in recommender systems for its representation learning capability on user and item embeddings. However, GCN is vulnerable to noisy and incomplete graphs, which are common in real world, due to its recursive message propagation mechanism. In the literature, some work propose to remove the feature transformation during message propagation, but making it unable to effectively capture the graph structural features. Moreover, they model users and items in the Euclidean space, which has been demonstrated to have high distortion when modeling complex graphs, further degrading the capability to capture the graph structural features and leading to sub-optimal performance. To this end, in this paper, we propose a simple yet effective Quaternion-based Graph Convolution Network (QGCN) recommendation model. In the proposed model, we utilize the hyper-complex Quaternion space to learn user and item representations and feature transformation to improve both performance and robustness. Specifically, we first embed all users and items into the Quaternion space. Then, we introduce the quaternion embedding propagation layers with quaternion feature transformation to perform message propagation. Finally, we combine the embeddings generated at each layer with the mean pooling strategy to obtain the final embeddings for recommendation. Extensive experiments on three public benchmark datasets demonstrate that our proposed QGCN model outperforms baseline methods by a large margin., 13 pages, 7 figures, 6 tables. Submitted to ICDE 2022

Knowledge Graph (KG) has attracted more and more companies' attention for its ability to connect different types of data in meaningful ways and support rich data services. However, the data isolation problem limits the performance of KG and prevents its further development. That is, multiple parties have their own KGs but they cannot share with each other due to regulation or competition reasons. Therefore, how to conduct privacy preserving KG becomes an important research question to answer. That is, multiple parties conduct KG related tasks collaboratively on the basis of protecting the privacy of multiple KGs. To date, there is few work on solving the above KG isolation problem. In this paper, to fill this gap, we summarize the open problems for privacy preserving KG in data isolation setting and propose possible solutions for them. Specifically, we summarize the open problems in privacy preserving KG from four aspects, i.e., merging, query, representation, and completion. We present these problems in details and propose possible technical solutions for them. Moreover, we present three privacy preserving KG-aware applications and simply describe how can our proposed techniques be applied into these applications.

Nowadays, massive geo-tagged records are generated on the social media. These records are useful when the users intend to plan a trip and are interested in some specific topics along the trip. With such redundant records, a publish/subscribe system has been designed to allow the users who are interested in certain information (i.e., the subscribers) to receive messages from some message generators (i.e., the publishers). Existing efforts on publish/subscribe mainly focus on the textual content or the spatial location of the subscribers, while leaving the consideration of incorporating the subscribers’ moving behaviors and temporal information. Therefore, in this paper, we propose a Time-aware Path-based Publish/Subscribe (TPPS) model, where we propose a filtering-verification framework that contains two kinds of filters, i.e., time-aware location-based filter and time-aware region-based filter, with considering both temporal information and moving behaviors, and filtering unrelated subscriptions for each message. We evaluate the efficiency of our approach on a real-world dataset and the experimental results demonstrate the superiority of our method in both efficiency and effectiveness.

The prevalence of GPS-enabled devices and wireless communication technologies has led to myriads of spatial trajectories describing the movement history of moving objects. While a substantial research effort has been undertaken on the spatio-temporal features of trajectory data, recent years have witnessed the flourish of location-based web applications (i.e., Foursquare, Facebook), enriching the traditional trajectory data by associating locations with activity information, called activity trajectory. These trajectory data contain a wealth of activity information and offer unprecedented opportunities for heightening our understanding about human behaviors. In this paper, we propose a novel framework, called TEH (Trajectory Embedding and Hashing), to mine the similarity among users based on their activity trajectories. Such user similarity is of great importance for individuals to effectively retrieve the information with high relevance. With the time being separated into several slots according to the activity-based temporal distribution, we utilize trajectory embedding technique to mine the sequence property of the activity trajectories by treating them as paragraphs. Then a hash-based method is presented to reduce the dimensions for improving the efficiency of users’ similarity calculation. Finally, extensive experiments on a real activity trajectory dataset demonstrate the effectiveness and efficiency of the proposed methods.

Remote sensing image segmentation is the basis of image pattern recognition. It is significant for the application and analysis of remote sensing images. Clustering analysis as a non-supervised learning method is widely used in the segmentation of remote sensing images. It has made good results in the segmentation of low-resolution and moderate-resolution remote sensing images. As the improvement of image resolution, however, they have problems in the segmentation of high-resolution remote sensing images. In this paper we propose an agglomerative hierarchical clustering based high-resolution remote sensing image segmentation algorithm. The segmentation experiments show that the result of this algorithm is better than the K-Meanspsila and is close to the results of artificial extraction.

Graph is a useful tool to model complicated data structures. One important task in graph analysis is assessing node proximity based on graph topology. Recently, Random Walk with Restart (RWR) tends to pop up as a promising measure of node proximity, due to its proliferative applications in e.g. recommender systems, and image segmentation. However, the best-known algorithm for computing RWR resorts to a large LU matrix factorization on an entire graph, which is cost-inhibitive. In this paper, we propose hybrid techniques to efficiently compute RWR. First, a novel divide-and-conquer paradigm is designed, aiming to convert the large LU decomposition into small triangular matrix operations recursively on several partitioned subgraphs. Then, on every subgraph, a “sparse accelerator” is devised to further reduce the time of RWR without any sacrifice in accuracy. Our experimental results on real and synthetic datasets show that our approach outperforms the baseline algorithms by at least one constant factor without loss of exactness.