Fraud detection is a key research topic for e-commerce, addressing challenges like dynamic heterogeneity and interlinked fraudulent patterns. Existing efforts include rule-based and machine learning systems, but graph-based approaches are increasingly critical. This paper presents the first systematic review of fraud detection in real-world e-commerce environment like eBay, leveraging multi-source data such as transaction logs and user behavior, dealing with challenges of information heterogeneity, scalability, graph dynamics, explainability, and adaptability. We also highlight eBay's efforts in designing explainable fraud detection systems with graph neural networks (GNNs) tailored to deployment needs and offer insights and recommendations for advancing research.

This work demonstrates how archives of daylight satellite data can be utilized to inform urban studies about the micro-regional, annual building-stock evolution over longer time spans. We predict building footprint and height on a 5-meter resolution to generate an 11-year panel data set for Beijing. In order to achieve this, we employ a U-Net algorithm on multiple scenes of the same region per year from the RapidEye legacy archive. We introduce a post-processing pipeline to address prediction variation over time with a changepoint detection algorithm, Pruned Exact Linear Time (PELT). The method is straightforward to use and does not require any longitudinal labels that are hard to obtain as input. We also compare our results with well-known data sources such as the Global Human Settlement Layer (GHSL) and show our strengths and contributions in modeling annual building volume change.

Traditional fraud detection approaches often use linking entities, such as device, email, and address, to identify fraudulent transactions and users. However, as fraud methods continue to evolve and escalate, the fraudsters can fabricate the involved entities and thus hide their real intent. To make fraud detection more robust, we incorporate user behaviors in the pipeline and consider biometric characteristics that are difficult to forge. In this work, we conduct a detailed study of how user behavior data can help identify and prevent fraudulent activity in e-commerce. We present Multi-Modal Behavioral Transformer (MMBT), where we combine both inner-page behavioral data, such as mouse trajectory, and inter-page behavioral data, such as page view sequences. We propose to construct mouse trajectory data as an image, treat each mouse position as a pixel in the image, convert the image into small patches, and hence transform the mouse trajectory into patch index sequences. Our experimental results on real-word data show that MMBT significantly outperforms baselines - the precision@recall = 0.1 increases by up to 7%. In addition, we have built an online pipeline to operationalize our model. In production, the 99th percentile latency is maintained below 500 milliseconds, allowing the platform to initiate rapid response measures and prevent potential losses.

This paper presents a hierarchical classification system that automatically categorizes a scholarly publication, using its abstract, into a three-tier hierarchical label set (discipline, field, subfield) in a multiclass setting. This system enables a holistic categorization of research activities in the mentioned hierarchy in terms of knowledge production through articles and impact through citations, permitting those activities to fall into multiple categories. The classification system distinguishes 44 disciplines, 718 fields, and 1,485 subfields among 160 million abstract snippets in Microsoft Academic Graph (version 2018-05-17). We used batch training in a modularized and distributed fashion to address and allow for interdisciplinary and interfield classifications in single-label and multilabel settings. In total, we have conducted 3,140 experiments in all considered models (Convolutional Neural Networks, Recurrent Neural Networks, and Transformers). The classification accuracy is >90% in 77.13% and 78.19% of the single-label and multilabel classifications, respectively. We examine the advantages of our classification by its ability to better align research texts and output with disciplines, to adequately classify them in an automated way, and to capture the degree of interdisciplinarity. The proposed system (a set of pre-trained models) can serve as a backbone for an interactive system for indexing scientific publications in the future.

The study of cities faces a core challenge: the absence of data that are simultaneously high-resolution, large-scale, and longitudinal. Only combining these three aspects reveals detailed (almost building-level) changes while covering vast urban areas consistently over time and promises advancing our understanding of the driving mechanisms of spatial agglomeration. We present a novel approach that leverages computer-vision techniques on Sentinel satellite imagery to generate detailed building-volume data throughout 106 cities in China over a six-year period (2018-2023). We validate the model by assessing building-volume density in out-of-sample cities. Additionally, we compare our results to nightlight-luminosity data, a frequently utilized remote-sensing resource for tracking density and human activity, and demonstrate how the proposed method and data drastically improve the measurement of urban density. The proposed method provides researchers in the social sciences at large with access to large and exponentially growing archives of customary daylight-satellite imagery either through direct use of the provided dataset or through adaptation of the model with new data.

Time-series information on building stock is of paramount importance to study cities in a host of disciplines ranging from economics to urban planning. Such data are lacking in a consistently measured way and especially among dynamically growing cities in developing countries. Due to their rapid change, building stock data in these cities can offer insights into the determinants and consequences of urbanization. To be able to analyze urban structures effectively, the building stock needs to be measured with sufficient detail-at a resolution that makes individual buildings or small conglomerates thereof visible-and it needs to consider building height (or volume) with a satisfactory scope across cities to cover both large numbers and multi-year sequences of data. This study aims to develop a comprehensive pipeline for predicting building volume-including both footprint and height-across 1,537 urban areas in mainland China, covering more than 60% of the Chinese population over a seven-year period (2017-2023). With the advancement of deep learning in remote sensing, we can leverage stateof-the-art techniques to efficiently produce large-scale data for Chinese cities across years, which could be very time-consuming with traditional remote-sensing techniques. We compare the performance of several deep learning architectures for the task at hand. We demonstrate that the best performing approach leads to credible metrics of both footprint and height predictions and performs very competitively with respect to existing building-volume predictions. We also benchmark our results against other data sources such as real-estate listings and demonstrate the out-of-sample prediction capability of the proposed model.

Temporal graph neural networks (TGNNs) outperform regular GNNs by incorporating time information into graph-based operations. However, TGNNs adopt specialized models (e.g., TGN, TGAT, and APAN ) and require tailored training frameworks (e.g., TGL and ETC). In this paper, we propose TF-TGN, which uses Transformer decoder as the backbone model for TGNN to enjoy Transformer's codebase for efficient training. In particular, Transformer achieves tremendous success for language modeling, and thus the community developed high-performance kernels (e.g., flash-attention and memory-efficient attention) and efficient distributed training schemes (e.g., PyTorch FSDP, DeepSpeed, and Megatron-LM). We observe that TGNN resembles language modeling, i.e., the message aggregation operation between chronologically occurring nodes and their temporal neighbors in TGNNs can be structured as sequence modeling. Beside this similarity, we also incorporate a series of algorithm designs including suffix infilling, temporal graph attention with self-loop, and causal masking self-attention to make TF-TGN work. During training, existing systems are slow in transforming the graph topology and conducting graph sampling. As such, we propose methods to parallelize the CSR format conversion and graph sampling. We also adapt Transformer codebase to train TF-TGN efficiently with multiple GPUs. We experiment with 9 graphs and compare with 2 state-of-the-art TGNN training frameworks. The results show that TF-TGN can accelerate training by over 2.20 while providing comparable or even superior accuracy to existing SOTA TGNNs. TF-TGN is available at https://github.com/qianghuangwhu/TF-TGN.

Datasets for Paper "BenchTemp: A General Benchmark for Evaluating Temporal Graph Neural Networks" URL: https://github.com/qianghuangwhu/benchtemp Openreview: https://openreview.net/forum?id=rnZm2vQq31 There are 19 (15+4) benchmark temporal graph datasets: reddit, wikipedia, mooc, lastfm, enron, SocialEvo, uci, CollegeMsg, TaobaoSmall, CanParl, Contacts, Flights, UNtrade, USLegis, UNvote, DGraphFin, TaobaoLarge, YoutubeReddit, YoutubeRedditLarge Each dataset has three files: 1. ml_{data_name}.csv - the csv file of the Temporal Graph. This file have five columns with properties: 'u': the id of the user. 'i': the id of the item. 'ts': the timestamp of the interaction (edge) between the user and the item. 'label': the label of the interaction (edge). 'idx': the index of the interaction (edge). For example: ,u,i,ts,label,idx 0,1,2,0.0,0.0,1 1,1,3,0.0,0.0,2 2,1,4,0.0,0.0,3 2. ml_{data_name}.npy - the edge features corresponding to the interactions (edges) in the the Temporal Graph.. 3. ml_{data_name}_node.npy - the initialization node features of the Temporal Graph.

This paper provides a first milestone in measuring the floorspace of buildings (that is, building footprint and height) for 40 major Chinese cities. The intent is to maximize city coverage and, eventually provide longitudinal data. Doing so requires building on imagery that is of a medium-fine-grained granularity, as larger cross sections of cities and longer time series for them are only available in such format. We use a multi-task object segmenter approach to learn the building footprint and height in the same framework in parallel: (1) we determine the surface area is covered by any buildings (the square footage of occupied land); (2) we determine floorspace from multi-image representations of buildings from various angles to determine the height of buildings. We use Sentinel-1 and -2 satellite images as our main data source. The benefits of these data are their large cross-sectional and longitudinal scope plus their unrestricted accessibility. We provide a detailed description of our data, algorithms, and evaluations. In addition, we analyze the quality of reference data and their role for measuring the building floorspace with minimal error. We conduct extensive quantitative and qualitative analyses with Shenzhen as a case study using our multi-task learner. Finally, we conduct correlation studies between our results (on both pixel and aggregated urban area levels) and nightlight data to gauge the merits of our approach in studying urban development. Our data and codebase are publicly accessible under https://gitlab.ethz.ch/raox/urban-satellite-public-v2.

We present SAINE, an Scientific Annotation and Inference ENgine based on a set of standard open-source software, such as Label Studio and MLflow. We show that our annotation engine can benefit the further development of a more accurate classification. Based on our previous work on hierarchical discipline classifications, we demonstrate its application using SAINE in understanding the space for scholarly publications. The user study of our annotation results shows that user input collected with the help of our system can help us better understand the classification process. We believe that our work will help to foster greater transparency and better understand scientific research. Our annotation and inference engine can further support the downstream meta-science projects. We welcome collaboration and feedback from the scientific community on these projects. The demonstration video can be accessed from https://youtu.be/yToO-G9YQK4. A live demo website is available at https://app.heartex.com/user/signup/?token=e2435a2f97449fa1 upon free registration.

This paper presents a hierarchical classification system that automatically categorizes a scholarly publication using its abstract into a three-tier hierarchical label set (discipline, field, subfield) in a multi-class setting. This system enables a holistic categorization of research activities in the mentioned hierarchy in terms of knowledge production through articles and impact through citations, permitting those activities to fall into multiple categories. The classification system distinguishes 44 disciplines, 718 fields and 1,485 subfields among 160 million abstract snippets in Microsoft Academic Graph (version 2018-05-17). We used batch training in a modularized and distributed fashion to address and allow for interdisciplinary and interfield classifications in single-label and multi-label settings. In total, we have conducted 3,140 experiments in all considered models (Convolutional Neural Networks, Recurrent Neural Networks, Transformers). The classification accuracy is > 90% in 77.13% and 78.19% of the single-label and multi-label classifications, respectively. We examine the advantages of our classification by its ability to better align research texts and output with disciplines, to adequately classify them in an automated way, and to capture the degree of interdisciplinarity. The proposed system (a set of pre-trained models) can serve as a backbone to an interactive system for indexing scientific publications in the future.

Tables have been an ever-existing structure to store data. There exist now different approaches to store tabular data physically. PDFs, images, spreadsheets, and CSVs are leading examples. Being able to parse table structures and extract content bounded by these structures is of high importance in many applications. In this paper, we devise TableParser, a system capable of parsing tables in both native PDFs and scanned images with high precision. We have conducted extensive experiments to show the efficacy of domain adaptation in developing such a tool. Moreover, we create TableAnnotator and ExcelAnnotator, which constitute a spreadsheet-based weak supervision mechanism and a pipeline to enable table parsing. We share these resources with the research community to facilitate further research in this interesting direction.

Detecting fraudulent transactions is an essential component to control risk in e-commerce marketplaces. Apart from rule-based and machine learning filters that are already deployed in production, we want to enable efficient real-time inference with graph neural networks (GNNs), which is useful to catch multihop risk propagation in a transaction graph. However, two challenges arise in the implementation of GNNs in production. First, future information in a dynamic graph should not be considered in message passing to predict the past. Second, the latency of graph query and GNN model inference is usually up to hundreds of milliseconds, which is costly for some critical online services. To tackle these challenges, we propose a Batch and Real-time Inception GrapH Topology (BRIGHT) framework to conduct an end-to-end GNN learning that allows efficient online real-time inference. BRIGHT framework consists of a graph transformation module (Two-Stage Directed Graph) and a corresponding GNN architecture (Lambda Neural Network). The Two-Stage Directed Graph guarantees that the information passed through neighbors is only from the historical payment transactions. It consists of two subgraphs representing historical relationships and real-time links, respectively. The Lambda Neural Network decouples inference into two stages: batch inference of entity embeddings and real-time inference of transaction prediction. Our experiments show that BRIGHT outperforms the baseline models by >2% in average w.r.t. precision. Furthermore, BRIGHT is computationally efficient for real-time fraud detection. Regarding end-to-end performance (including neighbor query and inference), BRIGHT can reduce the P99 latency by >75%. For the inference stage, our speedup is on average 7.8× compared to the traditional GNN.

The scientific publication output grows exponentially. Therefore, it is increasingly challenging to keep track of trends and changes. Understanding scientific documents is an important step in downstream tasks such as knowledge graph building, text mining, and discipline classification. In this workshop, we provide a better understanding of keyword and keyphrase extraction from the abstract of scientific publications.

Reasoning is a fundamental problem for computers and deeply studied in Artificial Intelligence. In this paper, we specifically focus on answering multi-hop logical queries on Knowledge Graphs (KGs). This is a complicated task because, in real-world scenarios, the graphs tend to be large and incomplete. Most previous works have been unable to create models that accept full First-Order Logical (FOL) queries, which include negative queries, and have only been able to process a limited set of query structures. Additionally, most methods present logic operators that can only perform the logical operation they are made for. We introduce a set of models that use Neural Networks to create one-point vector embeddings to answer the queries. The versatility of neural networks allows the framework to handle FOL queries with Conjunction ($\wedge$), Disjunction ($\vee$) and Negation ($\neg$) operators. We demonstrate experimentally the performance of our model through extensive experimentation on well-known benchmarking datasets. Besides having more versatile operators, the models achieve a 10\% relative increase over the best performing state of the art and more than 30\% over the original method based on single-point vector embeddings.

At online retail platforms, detecting fraudulent accounts and transactions is crucial to improve customer experience, minimize loss, and avoid unauthorized transactions. Despite the variety of different models for deep learning on graphs, few approaches have been proposed for dealing with graphs that are both heterogeneous and dynamic. In this paper, we propose DyHGN (Dynamic Heterogeneous Graph Neural Network) and its variants to capture both temporal and heterogeneous information. We first construct dynamic heterogeneous graphs from registration and transaction data from eBay. Then, we build models with diachronic entity embedding and heterogeneous graph transformer. We also use model explainability techniques to understand the behaviors of DyHGN-* models. Our findings reveal that modelling graph dynamics with heterogeneous inputs need to be conducted with "attention" depending on the data structure, distribution, and computation cost.