Pre-trained vision-language models like CLIP show great potential in learning representations that capture latent characteristics of users. A recently proposed method called Contextual Optimization (CoOp) introduces the concept of training prompt for adapting pre-trained vision-language models. Given the lightweight nature of this method, researchers have migrated the paradigm from centralized to decentralized system to innovate the collaborative training framework of Federated Learning (FL). However, current prompt training in FL mainly focuses on modeling user consensus and lacks the adaptation to user characteristics, leaving the personalization of prompt largely under-explored. Researches over the past few years have applied personalized FL (pFL) approaches to customizing models for heterogeneous users. Unfortunately, we find that with the variation of modality and training behavior, directly applying the pFL methods to prompt training leads to insufficient personalization and performance. To bridge the gap, we present pFedPrompt, which leverages the unique advantage of multimodality in vision-language models by learning user consensus from linguistic space and adapting to user characteristics in visual space in a non-parametric manner. Through this dual collaboration, the learned prompt will be fully personalized and aligned to the user’s local characteristics. We conduct extensive experiments across various datasets under the FL setting with statistical heterogeneity. The results demonstrate the superiority of our pFedPrompt against the alternative approaches with robust performance.
{"title":"pFedPrompt: Learning Personalized Prompt for Vision-Language Models in Federated Learning","authors":"Tao Guo, Song Guo, Junxiao Wang","doi":"10.1145/3543507.3583518","DOIUrl":"https://doi.org/10.1145/3543507.3583518","url":null,"abstract":"Pre-trained vision-language models like CLIP show great potential in learning representations that capture latent characteristics of users. A recently proposed method called Contextual Optimization (CoOp) introduces the concept of training prompt for adapting pre-trained vision-language models. Given the lightweight nature of this method, researchers have migrated the paradigm from centralized to decentralized system to innovate the collaborative training framework of Federated Learning (FL). However, current prompt training in FL mainly focuses on modeling user consensus and lacks the adaptation to user characteristics, leaving the personalization of prompt largely under-explored. Researches over the past few years have applied personalized FL (pFL) approaches to customizing models for heterogeneous users. Unfortunately, we find that with the variation of modality and training behavior, directly applying the pFL methods to prompt training leads to insufficient personalization and performance. To bridge the gap, we present pFedPrompt, which leverages the unique advantage of multimodality in vision-language models by learning user consensus from linguistic space and adapting to user characteristics in visual space in a non-parametric manner. Through this dual collaboration, the learned prompt will be fully personalized and aligned to the user’s local characteristics. We conduct extensive experiments across various datasets under the FL setting with statistical heterogeneity. The results demonstrate the superiority of our pFedPrompt against the alternative approaches with robust performance.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115389310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Intensive monitoring and anomaly diagnosis have become a knotty problem for modern microservice architecture due to the dynamics of service dependency. While most previous studies rely heavily on ample monitoring metrics, we raise a fundamental but always neglected issue - the diagnostic metric integrity problem. This paper solves the problem by proposing MicroCU – a novel approach to diagnose microservice systems using very sparse API logs. We design a structure named dynamic causal curves to portray time-varying service dependencies and a temporal dynamics discovery algorithm based on Granger causal intervals. Our algorithm generates a smoother space of causal curves and designs the concept of causal unimodalization to calibrate the causality infidelities brought by missing metrics. Finally, a path search algorithm on dynamic causality graphs is proposed to pinpoint the root cause. Experiments on commercial system cases show that MicroCU outperforms many state-of-the-art approaches and reflects the superiorities of causal unimodalization to raw metric imputation.
{"title":"Look Deep into the Microservice System Anomaly through Very Sparse Logs","authors":"Xinrui Jiang, Yicheng Pan, Meng Ma, Ping Wang","doi":"10.1145/3543507.3583338","DOIUrl":"https://doi.org/10.1145/3543507.3583338","url":null,"abstract":"Intensive monitoring and anomaly diagnosis have become a knotty problem for modern microservice architecture due to the dynamics of service dependency. While most previous studies rely heavily on ample monitoring metrics, we raise a fundamental but always neglected issue - the diagnostic metric integrity problem. This paper solves the problem by proposing MicroCU – a novel approach to diagnose microservice systems using very sparse API logs. We design a structure named dynamic causal curves to portray time-varying service dependencies and a temporal dynamics discovery algorithm based on Granger causal intervals. Our algorithm generates a smoother space of causal curves and designs the concept of causal unimodalization to calibrate the causality infidelities brought by missing metrics. Finally, a path search algorithm on dynamic causality graphs is proposed to pinpoint the root cause. Experiments on commercial system cases show that MicroCU outperforms many state-of-the-art approaches and reflects the superiorities of causal unimodalization to raw metric imputation.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115468450","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Deep Learning (DL) has been widely adopted in almost all domains, from threat recognition to medical diagnosis. Albeit its supreme model accuracy, DL imposes a heavy burden on devices as it incurs overwhelming system overhead to execute DL models, especially on Internet-of-Things (IoT) and edge devices. Collaborative inference is a promising approach to supporting DL models, by which the data owner (the victim) runs the first layers of the model on her local device and then a cloud provider (the adversary) runs the remaining layers of the model. Compared to offloading the entire model to the cloud, the collaborative inference approach is more data privacy-preserving as the owner’s model input is not exposed to outsiders. However, we show in this paper that the adversary can restore the victim’s model input by exploiting the output of the victim’s local model. Our attack is dubbed Ginver 1: Generative model inversion attacks against collaborative inference. Once trained, Ginver can infer the victim’s unseen model inputs without remaking the inversion attack model and thus has the generative capability. We extensively evaluate Ginver under different settings (e.g., white-box and black-box of the victim’s local model) and applications (e.g., CIFAR10 and FaceScrub datasets). The experimental results show that Ginver recovers high-quality images from the victims.
{"title":"Ginver: Generative Model Inversion Attacks Against Collaborative Inference","authors":"Yupeng Yin, Xianglong Zhang, Huanle Zhang, Feng Li, Yue Yu, Xiuzhen Cheng, Pengfei Hu","doi":"10.1145/3543507.3583306","DOIUrl":"https://doi.org/10.1145/3543507.3583306","url":null,"abstract":"Deep Learning (DL) has been widely adopted in almost all domains, from threat recognition to medical diagnosis. Albeit its supreme model accuracy, DL imposes a heavy burden on devices as it incurs overwhelming system overhead to execute DL models, especially on Internet-of-Things (IoT) and edge devices. Collaborative inference is a promising approach to supporting DL models, by which the data owner (the victim) runs the first layers of the model on her local device and then a cloud provider (the adversary) runs the remaining layers of the model. Compared to offloading the entire model to the cloud, the collaborative inference approach is more data privacy-preserving as the owner’s model input is not exposed to outsiders. However, we show in this paper that the adversary can restore the victim’s model input by exploiting the output of the victim’s local model. Our attack is dubbed Ginver 1: Generative model inversion attacks against collaborative inference. Once trained, Ginver can infer the victim’s unseen model inputs without remaking the inversion attack model and thus has the generative capability. We extensively evaluate Ginver under different settings (e.g., white-box and black-box of the victim’s local model) and applications (e.g., CIFAR10 and FaceScrub datasets). The experimental results show that Ginver recovers high-quality images from the victims.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124425549","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This work initiates the study of property testing in signed graphs, where every edge has either a positive or a negative sign. We show that there exist sublinear query and time algorithms for testing three key properties of signed graphs: balance (or 2-clusterability), clusterability and signed triangle freeness. We consider both the dense graph model, where one queries the adjacency matrix entries of a signed graph, and the bounded-degree model, where one queries for the neighbors of a node and the sign of the connecting edge. Our algorithms use a variety of tools from unsigned graph property testing, as well as reductions from one setting to the other. Our main technical contribution is a sublinear algorithm for testing clusterability in the bounded-degree model. This contrasts with the property of k-clusterability in unsigned graphs, which is not testable with a sublinear number of queries in the bounded-degree model. We experimentally evaluate the complexity and usefulness of several of our testers on real-life and synthetic datasets.
{"title":"Testing Cluster Properties of Signed Graphs","authors":"Florian Adriaens, Simon Apers","doi":"10.1145/3543507.3583213","DOIUrl":"https://doi.org/10.1145/3543507.3583213","url":null,"abstract":"This work initiates the study of property testing in signed graphs, where every edge has either a positive or a negative sign. We show that there exist sublinear query and time algorithms for testing three key properties of signed graphs: balance (or 2-clusterability), clusterability and signed triangle freeness. We consider both the dense graph model, where one queries the adjacency matrix entries of a signed graph, and the bounded-degree model, where one queries for the neighbors of a node and the sign of the connecting edge. Our algorithms use a variety of tools from unsigned graph property testing, as well as reductions from one setting to the other. Our main technical contribution is a sublinear algorithm for testing clusterability in the bounded-degree model. This contrasts with the property of k-clusterability in unsigned graphs, which is not testable with a sublinear number of queries in the bounded-degree model. We experimentally evaluate the complexity and usefulness of several of our testers on real-life and synthetic datasets.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"138 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127381698","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yuchen Zhou, Yanan Cao, Yongchao Liu, Yanmin Shang, P. Zhang, Zheng Lin, Yun Yue, Baokun Wang, Xingbo Fu, Weiqiang Wang
Data augmentation has been widely studied as it can be used to improve the generalizability of graph representation learning models. However, existing works focus only on the data augmentation on homogeneous graphs. Data augmentation for heterogeneous graphs remains under-explored. Considering that heterogeneous graphs contain different types of nodes and links, ignoring the type information and directly applying the data augmentation methods of homogeneous graphs to heterogeneous graphs will lead to suboptimal results. In this paper, we propose a novel Multi-Aspect Heterogeneous Graph Augmentation framework named MAHGA. Specifically, MAHGA consists of two core augmentation strategies: structure-level augmentation and metapath-level augmentation. Structure-level augmentation pays attention to network schema aspect and designs a relation-aware conditional variational auto-encoder that can generate synthetic features of neighbors to augment the nodes and the node types with scarce links. Metapath-level augmentation concentrates on metapath aspect, which constructs metapath reachable graphs for different metapaths and estimates the graphons of them. By sampling and mixing up based on the graphons, MAHGA yields intra-metapath and inter-metapath augmentation. Finally, we conduct extensive experiments on multiple benchmarks to validate the effectiveness of MAHGA. Experimental results demonstrate that our method improves the performances across a set of heterogeneous graph learning models and datasets.
{"title":"Multi-Aspect Heterogeneous Graph Augmentation","authors":"Yuchen Zhou, Yanan Cao, Yongchao Liu, Yanmin Shang, P. Zhang, Zheng Lin, Yun Yue, Baokun Wang, Xingbo Fu, Weiqiang Wang","doi":"10.1145/3543507.3583208","DOIUrl":"https://doi.org/10.1145/3543507.3583208","url":null,"abstract":"Data augmentation has been widely studied as it can be used to improve the generalizability of graph representation learning models. However, existing works focus only on the data augmentation on homogeneous graphs. Data augmentation for heterogeneous graphs remains under-explored. Considering that heterogeneous graphs contain different types of nodes and links, ignoring the type information and directly applying the data augmentation methods of homogeneous graphs to heterogeneous graphs will lead to suboptimal results. In this paper, we propose a novel Multi-Aspect Heterogeneous Graph Augmentation framework named MAHGA. Specifically, MAHGA consists of two core augmentation strategies: structure-level augmentation and metapath-level augmentation. Structure-level augmentation pays attention to network schema aspect and designs a relation-aware conditional variational auto-encoder that can generate synthetic features of neighbors to augment the nodes and the node types with scarce links. Metapath-level augmentation concentrates on metapath aspect, which constructs metapath reachable graphs for different metapaths and estimates the graphons of them. By sampling and mixing up based on the graphons, MAHGA yields intra-metapath and inter-metapath augmentation. Finally, we conduct extensive experiments on multiple benchmarks to validate the effectiveness of MAHGA. Experimental results demonstrate that our method improves the performances across a set of heterogeneous graph learning models and datasets.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125038409","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Web defacement is one of the major promotional channels for online underground economies. It regularly compromises benign websites and injects fraudulent content to promote illicit goods and services. It inflicts significant harm to websites’ reputations and revenues and may lead to legal ramifications. In this paper, we uncover proactive web defacements, where the involved web pages (i.e., landing pages) proactively deface themselves within browsers using JavaScript (i.e., control scripts). Proactive web defacements have not yet received attention from research communities, anti-hacking organizations, or law-enforcement officials. To detect proactive web defacements, we designed a practical tool, PACTOR. It runs in the browser and intercepts JavaScript API calls that manipulate web page content. It takes snapshots of the rendered HTML source code immediately before and after the intercepted API calls and detects proactive web defacements by visually comparing every two consecutive snapshots. Our two-month empirical study, using PACTOR, on 2,454 incidents of proactive web defacements shows that they can evade existing URL safety-checking tools and effectively promote the ranking of their landing pages using legitimate content/keywords. We also investigated the vendor network of proactive web defacements and reported all the involved domains to law-enforcement officials and URL-safety checking tools.
{"title":"The Chameleon on the Web: an Empirical Study of the Insidious Proactive Web Defacements","authors":"Rui Zhao","doi":"10.1145/3543507.3583377","DOIUrl":"https://doi.org/10.1145/3543507.3583377","url":null,"abstract":"Web defacement is one of the major promotional channels for online underground economies. It regularly compromises benign websites and injects fraudulent content to promote illicit goods and services. It inflicts significant harm to websites’ reputations and revenues and may lead to legal ramifications. In this paper, we uncover proactive web defacements, where the involved web pages (i.e., landing pages) proactively deface themselves within browsers using JavaScript (i.e., control scripts). Proactive web defacements have not yet received attention from research communities, anti-hacking organizations, or law-enforcement officials. To detect proactive web defacements, we designed a practical tool, PACTOR. It runs in the browser and intercepts JavaScript API calls that manipulate web page content. It takes snapshots of the rendered HTML source code immediately before and after the intercepted API calls and detects proactive web defacements by visually comparing every two consecutive snapshots. Our two-month empirical study, using PACTOR, on 2,454 incidents of proactive web defacements shows that they can evade existing URL safety-checking tools and effectively promote the ranking of their landing pages using legitimate content/keywords. We also investigated the vendor network of proactive web defacements and reported all the involved domains to law-enforcement officials and URL-safety checking tools.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115325521","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Measuring and assessing the impact and “fairness’’ of recommendation algorithms is central to responsible recommendation efforts. However, the complexity of fairness definitions and the proliferation of fairness metrics in research literature have led to a complex decision-making space. This environment makes it challenging for practitioners to operationalize and pick metrics that work within their unique context. This suggests that practitioners require more decision-making support, but it is not clear what type of support would be beneficial. We conducted a literature review of 24 papers to gather metrics introduced by the research community for measuring fairness in recommendation and ranking systems. We organized these metrics into a ‘decision-tree style’ support framework designed to help practitioners scope fairness objectives and identify fairness metrics relevant to their recommendation domain and application context. To explore the feasibility of this approach, we conducted 15 semi-structured interviews using this framework to assess which challenges practitioners may face when scoping fairness objectives and metrics for their system, and which further support may be needed beyond such tools.
{"title":"Scoping Fairness Objectives and Identifying Fairness Metrics for Recommender Systems: The Practitioners’ Perspective","authors":"Jessie J. Smith, Lex Beattie, H. Cramer","doi":"10.1145/3543507.3583204","DOIUrl":"https://doi.org/10.1145/3543507.3583204","url":null,"abstract":"Measuring and assessing the impact and “fairness’’ of recommendation algorithms is central to responsible recommendation efforts. However, the complexity of fairness definitions and the proliferation of fairness metrics in research literature have led to a complex decision-making space. This environment makes it challenging for practitioners to operationalize and pick metrics that work within their unique context. This suggests that practitioners require more decision-making support, but it is not clear what type of support would be beneficial. We conducted a literature review of 24 papers to gather metrics introduced by the research community for measuring fairness in recommendation and ranking systems. We organized these metrics into a ‘decision-tree style’ support framework designed to help practitioners scope fairness objectives and identify fairness metrics relevant to their recommendation domain and application context. To explore the feasibility of this approach, we conducted 15 semi-structured interviews using this framework to assess which challenges practitioners may face when scoping fairness objectives and metrics for their system, and which further support may be needed beyond such tools.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122444318","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recognizing emotions from multi-modal data is an emotion recognition task that requires strong multi-modal representation ability. The general approach to this task is to naturally train the representation model on training data without intervention. However, such natural training scheme is prone to modality bias of representation (i.e., tending to over-encode some informative modalities while neglecting other modalities) and data bias of training (i.e., tending to overfit training data). These biases may lead to instability (e.g., performing poorly when the neglected modality is dominant for recognition) and weak generalization (e.g., performing poorly when unseen data is inconsistent with overfitted data) of the model on unseen data. To address these problems, this paper presents two adversarial training strategies to learn more robust multi-modal representation for multi-label emotion recognition. Firstly, we propose an adversarial temporal masking strategy, which can enhance the encoding of other modalities by masking the most emotion-related temporal units (e.g., words for text or frames for video) of the informative modality. Secondly, we propose an adversarial parameter perturbation strategy, which can enhance the generalization of the model by adding the adversarial perturbation to the parameters of model. Both strategies boost model performance on the benchmark MMER datasets CMU-MOSEI and NEMu. Experimental results demonstrate the effectiveness of the proposed method compared with the previous state-of-the-art method. Code will be released at https://github.com/ShipingGe/MMER.
{"title":"Learning Robust Multi-Modal Representation for Multi-Label Emotion Recognition via Adversarial Masking and Perturbation","authors":"Shiping Ge, Zhiwei Jiang, Zifeng Cheng, Cong Wang, Yafeng Yin, Qing Gu","doi":"10.1145/3543507.3583258","DOIUrl":"https://doi.org/10.1145/3543507.3583258","url":null,"abstract":"Recognizing emotions from multi-modal data is an emotion recognition task that requires strong multi-modal representation ability. The general approach to this task is to naturally train the representation model on training data without intervention. However, such natural training scheme is prone to modality bias of representation (i.e., tending to over-encode some informative modalities while neglecting other modalities) and data bias of training (i.e., tending to overfit training data). These biases may lead to instability (e.g., performing poorly when the neglected modality is dominant for recognition) and weak generalization (e.g., performing poorly when unseen data is inconsistent with overfitted data) of the model on unseen data. To address these problems, this paper presents two adversarial training strategies to learn more robust multi-modal representation for multi-label emotion recognition. Firstly, we propose an adversarial temporal masking strategy, which can enhance the encoding of other modalities by masking the most emotion-related temporal units (e.g., words for text or frames for video) of the informative modality. Secondly, we propose an adversarial parameter perturbation strategy, which can enhance the generalization of the model by adding the adversarial perturbation to the parameters of model. Both strategies boost model performance on the benchmark MMER datasets CMU-MOSEI and NEMu. Experimental results demonstrate the effectiveness of the proposed method compared with the previous state-of-the-art method. Code will be released at https://github.com/ShipingGe/MMER.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122030249","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Dongqi Fu, Dawei Zhou, Ross Maciejewski, A. Croitoru, Marcus Boyd, Jingrui He
With the widespread development of algorithmic fairness, there has been a surge of research interest that aims to generalize the fairness notions from the attributed data to the relational data (graphs). The vast majority of existing work considers the fairness measure in terms of the low-order connectivity patterns (e.g., edges), while overlooking the higher-order patterns (e.g., k-cliques) and the dynamic nature of real-world graphs. For example, preserving triangles from graph cuts during clustering is the key to detecting compact communities; however, if the clustering algorithm only pays attention to triangle-based compactness, then the returned communities lose the fairness guarantee for each group in the graph. Furthermore, in practice, when the graph (e.g., social networks) topology constantly changes over time, one natural question is how can we ensure the compactness and demographic parity at each timestamp efficiently. To address these problems, we start from the static setting and propose a spectral method that preserves clique connections and incorporates demographic fairness constraints in returned clusters at the same time. To make this static method fit for the dynamic setting, we propose two core techniques, Laplacian Update via Edge Filtering and Searching and Eigen-Pairs Update with Singularity Avoided. Finally, all proposed components are combined into an end-to-end clustering framework named F-SEGA, and we conduct extensive experiments to demonstrate the effectiveness, efficiency, and robustness of F-SEGA.
{"title":"Fairness-Aware Clique-Preserving Spectral Clustering of Temporal Graphs","authors":"Dongqi Fu, Dawei Zhou, Ross Maciejewski, A. Croitoru, Marcus Boyd, Jingrui He","doi":"10.1145/3543507.3583423","DOIUrl":"https://doi.org/10.1145/3543507.3583423","url":null,"abstract":"With the widespread development of algorithmic fairness, there has been a surge of research interest that aims to generalize the fairness notions from the attributed data to the relational data (graphs). The vast majority of existing work considers the fairness measure in terms of the low-order connectivity patterns (e.g., edges), while overlooking the higher-order patterns (e.g., k-cliques) and the dynamic nature of real-world graphs. For example, preserving triangles from graph cuts during clustering is the key to detecting compact communities; however, if the clustering algorithm only pays attention to triangle-based compactness, then the returned communities lose the fairness guarantee for each group in the graph. Furthermore, in practice, when the graph (e.g., social networks) topology constantly changes over time, one natural question is how can we ensure the compactness and demographic parity at each timestamp efficiently. To address these problems, we start from the static setting and propose a spectral method that preserves clique connections and incorporates demographic fairness constraints in returned clusters at the same time. To make this static method fit for the dynamic setting, we propose two core techniques, Laplacian Update via Edge Filtering and Searching and Eigen-Pairs Update with Singularity Avoided. Finally, all proposed components are combined into an end-to-end clustering framework named F-SEGA, and we conduct extensive experiments to demonstrate the effectiveness, efficiency, and robustness of F-SEGA.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122158599","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Extracting structured information from all manner of webpages is an important problem with the potential to automate many real-world applications. Recent work has shown the effectiveness of leveraging DOM trees and pre-trained language models to describe and encode webpages. However, they typically optimize the model to learn the semantic co-occurrence of elements and labels in the same webpage, thus their effectiveness depends on sufficient labeled data, which is labor-intensive. In this paper, we further observe structural co-occurrences in different webpages of the same website: the same position in the DOM tree usually plays the same semantic role, and the DOM nodes in this position also share similar surface forms. Motivated by this, we propose a novel method, Structor, to effectively incorporate the structural co-occurrences over DOM tree and surface form into pre-trained language models. Such structural co-occurrences help the model learn the task better under low-resource settings, and we study two challenging experimental scenarios: website-level low-resource setting and webpage-level low-resource setting, to evaluate our approach. Extensive experiments on the public SWDE dataset show that Structor significantly outperforms the state-of-the-art models in both settings, and even achieves three times the performance of the strong baseline model in the case of extreme lack of training data.
{"title":"Learning Structural Co-occurrences for Structured Web Data Extraction in Low-Resource Settings","authors":"Zhenyu Zhang, Yu Bowen, Tingwen Liu, Tianyun Liu, Yubin Wang, Li Guo","doi":"10.1145/3543507.3583387","DOIUrl":"https://doi.org/10.1145/3543507.3583387","url":null,"abstract":"Extracting structured information from all manner of webpages is an important problem with the potential to automate many real-world applications. Recent work has shown the effectiveness of leveraging DOM trees and pre-trained language models to describe and encode webpages. However, they typically optimize the model to learn the semantic co-occurrence of elements and labels in the same webpage, thus their effectiveness depends on sufficient labeled data, which is labor-intensive. In this paper, we further observe structural co-occurrences in different webpages of the same website: the same position in the DOM tree usually plays the same semantic role, and the DOM nodes in this position also share similar surface forms. Motivated by this, we propose a novel method, Structor, to effectively incorporate the structural co-occurrences over DOM tree and surface form into pre-trained language models. Such structural co-occurrences help the model learn the task better under low-resource settings, and we study two challenging experimental scenarios: website-level low-resource setting and webpage-level low-resource setting, to evaluate our approach. Extensive experiments on the public SWDE dataset show that Structor significantly outperforms the state-of-the-art models in both settings, and even achieves three times the performance of the strong baseline model in the case of extreme lack of training data.","PeriodicalId":296351,"journal":{"name":"Proceedings of the ACM Web Conference 2023","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129050180","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：