The spread of fake news harms individuals and presents a critical social challenge that must be addressed. Although numerous algorithmic and insightful features have been developed to detect fake news, many of these features can be manipulated with style-conversion attacks, especially with the emergence of advanced language models, making it more difficult to differentiate from genuine news. This study proposes adversarial style augmentation, AdStyle, designed to train a fake news detector that remains robust against various style-conversion attacks. The primary mechanism involves the strategic use of LLMs to automatically generate a diverse and coherent array of style-conversion attack prompts, enhancing the generation of particularly challenging prompts for the detector. Experiments indicate that our augmentation strategy significantly improves robustness and detection performance when evaluated on fake news benchmark datasets.

Large language models have revolutionized text generation, offering significant benefits while also posing threats to society, such as copyright infringement and misinformation. To prevent harmful use, the task of detecting machine-generated content has become an important research topic, though it remains particularly challenging across diverse content domains. This paper presents DGRM, an innovative add-on module designed to improve the domain generalization capability of existing machine-generated text detectors. Our model consists of two training components. (1) Feature disentanglement separates a text's embedding into target-specific and common attributes, thereby enhancing semantic domain generalization across different content domains. (2) Feature regularization applies constraints to these attributes to extract additional target-relevant information and ensure detection consistency under syntactic perturbations—thus achieving syntactic domain generalization. Evaluation over multiple datasets demonstrates that incorporating our module substantially improves the detection of machine-generated text across semantically and syntactically diverse domains. We hope our work contributes to mitigating the harmful use of language models.

The task of assigning internationally accepted commodity codes (aka HS codes) to traded goods is a critical function of customs offices. Like court decisions made by judges, this task follows the doctrine of precedent and can be nontrivial even for experienced officers. Together with the Korea Customs Service (KCS), we propose a first-ever explainable decision supporting model that suggests the most likely subheadings (i.e., the first six digits) of the HS code. The model also provides reasoning for its suggestion in the form of a document that is interpretable by customs officers. We evaluated the model using 5,000 cases that recently received a classification request. The results showed that the top-3 suggestions made by our model had an accuracy of 93.9% when classifying 925 challenging subheadings. A user study with 32 customs experts further confirmed that our algorithmic suggestions accompanied by explainable reasonings, can substantially reduce the time and effort taken by customs officers for classification reviews.

While measuring socioeconomic indicators is critical for local governments to make informed policy decisions, such measurements are often unavailable at fine-grained levels like municipality. This study employs deep learning-based predictions from satellite images to close the gap. We propose a method that assigns a socioeconomic score to each satellite image by capturing the distributional behavior observed in larger areas based on the ground truth. We train an ordinal regression scoring model and adjust the scores to follow the common power law within and across regions. Evaluation based on official statistics in South Korea shows that our method outperforms previous models in predicting population and employment size at both the municipality and grid levels. Our method also demonstrates robust performance in districts with uneven development, suggesting its potential use in developing countries where reliable, fine-grained data is scarce.

Information presented in Wikipedia articles must be attributable to reliable published sources in the form of references. This study examines over 5 million Wikipedia articles to assess the reliability of references in multiple language editions. We quantify the cross-lingual patterns of the perennial sources list, a collection of reliability labels for web domains identified and collaboratively agreed upon by Wikipedia editors. We discover that some sources (or web domains) deemed untrustworthy in one language (i.e., English) continue to appear in articles in other languages. This trend is especially evident with sources tailored for smaller communities. Furthermore, non-authoritative sources found in the English version of a page tend to persist in other language versions of that page. We finally present a case study on the Chinese, Russian, and Swedish Wikipedias to demonstrate a discrepancy in reference reliability across cultures. Our finding highlights future challenges in coordinating global knowledge on source reliability.

Customs officials across the world encounter huge volumes of transactions. Associated with customs transactions is customs fraud-the intentional manipulation of goods declarations to avoid taxes and duties. Due to limited manpower, the customs offices can only manually inspect a small number of declarations, necessitating the automation of customs fraud detection by machine learning techniques. The limited availability of manually inspected ground truth data makes it essential for the ML approach to generalize well on unseen data. However, current customs fraud detection models are not well suited or designed for this setting. In this work, we propose GraphFC (Graph Neural networks for Customs Fraud), a model-agnostic, domain-specific, graph neural network based customs fraud detection model that is designed to work in a real-world setting with limited ground truth data. Extensive experimentation using real customs data from two countries demonstrates that GraphFC generalizes well over unseen data and outperforms various baselines and other models by a large margin.

Federated learning enables learning from decentralized data sources without compromising privacy, which makes it a crucial technique. However, it is vulnerable to model poisoning attacks, where malicious clients interfere with the training process. Previous defense mechanisms have focused on the server-side by using careful model aggregation, but this may not be effective when the data is not identically distributed or when attackers can access the information of benign clients. In this paper, we propose a new defense mechanism that focuses on the client-side, called FedDefender, to help benign clients train robust local models and avoid the adverse impact of malicious model updates from attackers, even when a server-side defense cannot identify or remove adversaries. Our method consists of two main components: (1) attack-tolerant local meta update and (2) attack-tolerant global knowledge distillation. These components are used to find noise-resilient model parameters while accurately extracting knowledge from a potentially corrupted global model. Our client-side defense strategy has a flexible structure and can work in conjunction with any existing server-side strategies. Evaluations of real-world scenarios across multiple datasets show that the proposed method enhances the robustness of federated learning against model poisoning attacks.

Algorithmic fairness has become an important machine learning problem, especially for mission-critical Web applications. This work presents a self-supervised model, called DualFair, that can debias sensitive attributes like gender and race from learned representations. Unlike existing models that target a single type of fairness, our model jointly optimizes for two fairness criteria—group fairness and counterfactual fairness—and hence makes fairer predictions at both the group and individual levels. Our model uses contrastive loss to generate embeddings that are indistinguishable for each protected group, while forcing the embeddings of counterfactual pairs to be similar. It then uses a self-knowledge distillation method to maintain the quality of representation for the downstream tasks. Extensive analysis over multiple datasets confirms the model’s validity and further shows the synergy of jointly addressing two fairness criteria, suggesting the model’s potential value in fair intelligent Web applications.

Wikipedia plays a crucial role in the integrity of the Web. This work analyzes the reliability of this global encyclopedia through the lens of its references. We operationalize the notion of reference quality by defining reference need (RN), i.e., the percentage of sentences missing a citation, and reference risk (RR), i.e., the proportion of non-authoritative references. We release Citation Detective, a tool for automatically calculating the RN score, and discover that the RN score has dropped by 20 percent point in the last decade, with more than half of verifiable statements now accompanying references. The RR score has remained below 1% over the years as a result of the efforts of the community to eliminate unreliable references. We propose pairing novice and experienced editors on the same Wikipedia article as a strategy to enhance reference quality. Our quasi-experiment indicates that such a co-editing experience can result in a lasting advantage in identifying unreliable sources in future edits. As Wikipedia is frequently used as the ground truth for numerous Web applications, our findings and suggestions on its reliability can have a far-reaching impact. We discuss the possibility of other Web services adopting Wiki-style user collaboration to eliminate unreliable content.

Individuals interacting in organizational settings involving varying levels of formal hierarchy naturally form a complex network of social ties having different tie valences (e.g., positive and negative connections). Social ties critically affect employees’ satisfaction, behaviors, cognition, and outcomes—yet identifying them solely through survey data is challenging because of the large size of some organizations or the often hidden nature of these ties and their valences. We present a novel deep learning model encompassing NLP and graph neural network techniques that identifies positive and negative ties in a hierarchical network. The proposed model uses human resource attributes as node information and web-logged work conversation data as link information. Our findings suggest that the presence of conversation data improves the tie valence classification by 8.91% compared to employing user attributes alone. This gain came from accurately distinguishing positive ties, particularly for male, non-minority, and older employee groups. We also show a substantial difference in conversation patterns for positive and negative ties with positive ties being associated with more messages exchanged on weekends, and lower use of words related to anger and sadness. These findings have broad implications for facilitating collaboration and managing conflict within organizational and other social networks.

We study the human-in-the-loop customs inspection scenario, where an AI-assisted algorithm supports customs officers by recommending a set of imported goods to be inspected. If the inspected items are fraudulent, the officers can levy extra duties. These logs are then used as additional training data for the next iterations. Choosing to inspect suspicious items first leads to an immediate gain in customs revenue, yet such inspections may not bring new insights for learning dynamic traffic patterns. On the other hand, inspecting uncertain items can help acquire new knowledge, which will be used as a supplementary training resource to update the selection systems. Based on multiyear customs datasets from three countries, we demonstrate that some degree of exploration is necessary to cope with domain shifts in the trade data. The results show that a hybrid strategy of selecting likely fraudulent and uncertain items will eventually outperform the exploitation-only strategy.

Modern climate models offer simulation results that provide unprecedented details at the local level. However, even with powerful supercomputing facilities, their computational complexity and associated costs pose a limit on simulation resolution that is needed for agile planning of resource allocation, parameter calibration, and model reproduction. As regional information is vital for policymakers, data from coarse-grained resolution simulations undergo the process of "statistical downscaling" to generate higher-resolution projection at a local level. We present a new method for downscaling climate simulations called GINE (Geospatial INformation Encoded statistical downscaling). To preserve the characteristics of climate simulation data during this process, our model applies the latest computer vision techniques over topography-driven spatial and local-level information. The comprehensive evaluations on 2x, 4x, and 8x resolution factors show that our model substantially improves performance in terms of RMSE and the visual quality of downscaled data.

Web data and computational models can play important roles in analyzing cultural trends. The current study presents an analysis of 23,492 sneaker images and metadata collected from a global reselling shop, StockX.com. Based on data encompassing 22 years from 1999 to 2020, we propose a sneaker design index that helps track changes in the design characteristics of sneakers using a contrastive learning method. Our data suggest that sneaker designs have been employing brighter colors and lower hue and saturation values over time. We also observe how popular brands have continued to build their unique identities in shape-related design space. The embedding analysis also predicts which sneakers will likely see a high premium in the reselling market, suggesting viable algorithm-driven investment and design strategies. The current work is one of the first publicly available studies to analyze product design evolution over a long historical period and has implications for the novel use of Web data to understand cultural patterns that are otherwise difficult to assess.

The COVID-19 pandemic has been the single most important global agenda in the past two years. In addition to its health and economic impacts, it has affected people’s psychological states, including a rise in depression and domestic violence. We traced how the overall emotional states of individual Twitter users changed before and after the pandemic. Our data, including more than 9 million tweets posted by 9,493 users, suggest that the threat posed by the virus did not upset the emotional equilibrium of social media. In early 2020, COVID-related tweets skyrocketed in number and were filled with negative emotions; however, this emotional outburst was short-lived. We found that users who had expressed positive emotions in the pre-COVID period remained positive after the initial outbreak, while the opposite was true for those who regularly expressed negative emotions. Individuals achieved such emotional consistency by selectively focusing on emotion-reinforcing topics. The implications are discussed in light of an emotionally motivated confirmation bias, which we conceptualize as emotion bubbles that demonstrate the public’s resilience to a global health risk.

The online retail market shows a sharp increase in traffic during holiday sales. The ability to distinguish customers who will likely purchase is critical for provisioning traffic and for providing cost-effective promotions. This paper uniquely studies the browsing and purchasing behaviors of online shoppers during a yearly sale event in China, the world’s largest online marketplace. Based on 31 million action logs gathered from wide residential areas, we characterize the steps leading to purchases and determine their precursors. We investigate the effect of time (e.g., date, time of date), environment (e.g., platform, viewed category), and action (e.g., session time, clicks, sequence) on purchases. Action cues from shopping behaviors can be used for early detection. While most shoppers start with strong intentions to purchase, yet the moment of ordering comes rather impulsively within 30 seconds to several minutes of browsing. The predictive accuracy reaches as a high AUC of 0.924. The findings in this paper provide an understanding of traffic during mega sale events that can help online shops plan and provide a better user experience for upcoming shopping festivals.

Fact-checking has become the de facto solution for fighting fake news online. This research brings attention to the unexpected and diminished effect of fact-checking due to cognitive biases. We experimented (66,870 decisions) comparing the change in users’ stance toward unproven claims before and after being presented with a hypothetical fact-checked condition. We found that, first, the claims tagged with the ‘Lack of Evidence’ label are recognized similarly as false information unlike other borderline labels, indicating the presence of uncertainty-aversion bias in response to insufficient information. Second, users who initially show disapproval toward a claim are less likely to correct their views later than those who initially approve of the same claim when opposite fact-checking labels are shown — an indication of disapproval bias. Finally, user interviews revealed that users are more likely to share claims with Divided Evidence than those with Lack of Evidence among borderline messages, reaffirming the presence of uncertainty-aversion bias. On average, we confirm that fact-checking helps users correct their views and reduces the circulation of falsehoods by leading them to abandon extreme views. Simultaneously, the presence of two biases reveals that fact-checking does not always elicit the desired user experience and that the outcome varies by the design of fact-checking messages and people’s initial view. These new observations have direct implications for multiple stakeholders, including platforms, policy-makers, and online users.

Self-supervised representation learning aims to generate effective representations for data instances without the need for manual labels, also known as unsupervised embedding learning, which has been a critical challenge in many existing semi-supervised and supervised learning tasks. This paper proposes a new self-supervised learning approach, called Super-AND, which extends the memory-based pretraining method AND model [13]. Super-AND has its unique set of losses that combines data augmentation in neighborhood discovery for more accurate anchor selection in embedding learning and further presents an adversarial training manner to learn more confident embeddings under the unsupervised setting. Experimental results exhibit that Super-AND outperforms all existing state-of-the-art self-supervised representation learning approaches and achieves an accuracy of 89.2% on the image classification task for CIFAR-10.

Reliable and timely measurements of economic activities are fundamental for understanding economic development and designing government policies. However, many developing countries still lack reliable data. In this paper, we introduce a novel approach for measuring economic development from high-resolution satellite images in the absence of ground truth statistics. Our method consists of three steps. First, we run a clustering algorithm on satellite images that distinguishes artifacts from nature (siCluster). Second, we generate a partial order graph of the identified clusters based on the level of economic development, either by human guidance or by low-resolution statistics (siPog). Third, we use a CNN-based sorter that assigns differentiable scores to each satellite grid based on the relative ranks of clusters (siScore). The novelty of our method is that we break down a computationally hard problem into sub-tasks, which involves a human-in-the-loop solution. With the combination of unsupervised learning and the partial orders of dozens of urban vs. rural clusters, our method can estimate the economic development scores of over 10,000 satellite grids consistently with other baseline development proxies (Spearman correlation of 0.851). This efficient method is interpretable and robust; we demonstrate how to apply our method to both developed (e.g., South Korea) and developing economies (e.g., Vietnam and Malawi).

Intentional manipulation of invoices that lead to undervaluation of trade goods is the most common type of customs fraud to avoid ad valorem duties and taxes. To secure government revenue without interrupting legitimate trade flows, customs administrations around the world strive to develop ways to detect illicit trades. This paper proposes DATE, a model of Dual-task Attentive Tree-aware Embedding, to classify and rank illegal trade flows that contribute the most to the overall customs revenue when caught. The strength of DATE comes from combining a tree-based model for interpretability and transaction-level embeddings with dual attention mechanisms. To accurately identify illicit transactions and predict tax revenue, DATE learns simultaneously from illicitness and surtax of each transaction. With a five-year amount of customs import data with a test illicit ratio of 2.24%, DATE shows a remarkable precision of 92.7% on illegal cases and a recall of 49.3% on revenue after inspecting only 1% of all trade flows. We also discuss issues on deploying DATE in Nigeria Customs Service, in collaboration with the World Customs Organization.

Precision psychiatry is a new research field that uses advanced data mining over a wide range of neural, behavioral, psychological, and physiological data sources for classification of mental health conditions. This study presents a computational framework for predicting sleep efficiency of insomnia sufferers. A smart band experiment is conducted to collect heterogeneous data, including sleep records, daily activities, and demographics, whose missing values are imputed via Improved Generative Adversarial Imputation Networks (Imp-GAIN). Equipped with the imputed data, we predict sleep efficiency of individual users with a proposed interpretable LSTM-Attention (LA Block) neural network model. We also propose a model, Pairwise Learning-based Ranking Generation (PLRG), to rank users with high insomnia potential in the next day. We discuss implications of our findings from the perspective of a psychiatric practitioner. Our computational framework can be used for other applications that analyze and handle noisy and incomplete time-series human activity data in the domain of precision psychiatry.

This study investigates factors that may determine satisfaction in customer service operations. We utilized more than 170,000 online chat sessions between customers and agents to identify characteristics of chat sessions that incurred dissatisfying experience. Quantitative data analysis suggests that sentiments or moods conveyed in online conversation are the most predictive factor of perceived satisfaction. Conversely, other session related meta data (such as that length, time of day, and response time) has a weaker correlation with user satisfaction. Knowing in advance what can predict satisfaction allows customer service staffs to identify potential weaknesses and improve the quality of service for better customer experience.

How does one develop a new online community that is highly engaging to each user and promotes social interaction? A number of websites offer friend-finding features that help users bootstrap social networks on the website by copying links from an established network like Facebook or Twitter. This paper quantifies the extent to which such social bootstrapping is effective in enhancing a social experience of the website. First, we develop a stylised analytical model that suggests that copying tends to produce a giant connected component (i.e., a connected community) quickly and preserves properties such as reciprocity and clustering, up to a linear multiplicative factor. Second, we use data from two websites, Pinterest and Last.fm, to empirically compare the subgraph of links copied from Facebook to links created natively. We find that the copied subgraph has a giant component, higher reciprocity and clustering, and confirm that the copied connections see higher social interactions. However, the need for copying diminishes as users become more active and influential. Such users tend to create links natively on the website, to users who are more similar to them than their Facebook friends. Our findings give new insights into understanding how bootstrapping from established social networks can help engage new users by enhancing social interactivity.

The problem of identifying rumors is of practical importance especially in online social networks, since information can diffuse more rapidly and widely than the offline counterpart. In this paper, we identify characteristics of rumors by examining the following three aspects of diffusion: temporal, structural, and linguistic. For the temporal characteristics, we propose a new periodic time series model that considers daily and external shock cycles, where the model demonstrates that rumor likely have fluctuations over time. We also identify key structural and linguistic differences in the spread of rumors and non-rumors. Our selected features classify rumors with high precision and recall in the range of 87% to 92%, that is higher than other states of the arts on rumor classification.

Modern Massively Multiplayer Online Role-Playing Games (MMORPGs) provide lifelike virtual environments in which players can conduct a variety of activities including combat, trade, and chat with other players. While the game world and the available actions therein are inspired by their offline counterparts, the games' popularity and dedicated fan base are testaments to the allure of novel social interactions granted to people by allowing them an alternative life as a new character and persona. In this paper we investigate the phenomenon of "gender swapping," which refers to players choosing avatars of genders opposite to their natural ones. We report the behavioral patterns observed in players of Fairyland Online, a globally serviced MMORPG, during social interactions when playing as in-game avatars of their own real gender or gender-swapped. We also discuss the effect of gender role and self-image in virtual social situations and the potential of our study for improving MMORPG quality and detecting online identity frauds.

The way in which social conventions emerge in communities has been of interest to social scientists for decades. Here we report on the emergence of a particular social convention on Twitter---the way to indicate a tweet is being reposted and attributing the content to its source. Despite being invented at different times and having different adoption rates, only two variations became widely adopted. In this paper we describe this process in detail, highlighting the factors that come into play in deciding which variation individuals will adopt. Our classification analysis demonstrates that the date of adoption and the number of exposures are particularly important in the adoption process, while personal features (such as the number of followers and join date) and the number of adopter friends have less discriminative power in predicting adoptions. We discuss implications of these findings in the design of future Web applications and services.

Online social networking sites like MySpace, Facebook, and Flickr have become a popular way to share and disseminate content. Their massive popularity has led to viral marketing techniques that attempt to spread content, products, and ideas on these sites. However, there is little data publicly available on viral propagation in the real world and few studies have characterized how information spreads over current online social networks.