It is increasingly common in digital environments to use A/B tests to compare the performance of recommendation algorithms. However, such experiments often violate the stable unit treatment value assumption (SUTVA), particularly SUTVA's ''no hidden treatments'' assumption, due to the shared data between algorithms being compared. This results in a novel form of bias, which we term ''symbiosis bias,'' where the performance of each algorithm is influenced by the training data generated by its competitor. In this paper, we investigate three experimental designs--cluster-randomized, data-diverted, and user-corpus co-diverted experiments--aimed at mitigating symbiosis bias. We present a theoretical model of symbiosis bias and simulate the impact of each design in dynamic recommendation environments. Our results show that while each design reduces symbiosis bias to some extent, they also introduce new challenges, such as reduced training data in data-diverted experiments. We further validate the existence of symbiosis bias using data from a large-scale A/B test conducted on a global recommender system, demonstrating that symbiosis bias affects treatment effect estimates in the field. Our findings provide actionable insights for researchers and practitioners seeking to design experiments that accurately capture algorithmic performance without bias in treatment effect estimates introduced by shared data.
论文检索
输入标题、作者或关键词，从 447 篇学术成果中精准定位
Large language models (LLMs), with advanced linguistic capabilities, have been employed in reranking tasks through a sequence-to-sequence approach. In this paradigm, multiple passages are reranked in a listwise manner and a textual reranked permutation is generated. However, due to the limited context window of LLMs, this reranking paradigm requires a sliding window strategy to iteratively handle larger candidate sets. This not only increases computational costs but also restricts the LLM from fully capturing all the comparison information for all candidates. To address these challenges, we propose a novel self-calibrated listwise reranking method, which aims to leverage LLMs to produce global relevance scores for ranking. To achieve it, we first propose the relevance-aware listwise reranking framework, which incorporates explicit list-view relevance scores to improve reranking efficiency and enable global comparison across the entire candidate set. Second, to ensure the comparability of the computed scores, we propose self-calibrated training that uses point-view relevance assessments generated internally by the LLM itself to calibrate the list-view relevance assessments. Extensive experiments and comprehensive analysis on the BEIR benchmark and TREC Deep Learning Tracks demonstrate the effectiveness and efficiency of our proposed method.
CAPTCHAs have long been essential tools for protecting applications from automated bots. Initially designed as simple questions to distinguish humans from bots, they have become increasingly complex to keep pace with the proliferation of CAPTCHA-cracking techniques employed by malicious actors. However, with the advent of advanced large language models (LLMs), the effectiveness of existing CAPTCHAs is now being undermined. To address this issue, we have conducted an empirical study to evaluate the performance of multimodal LLMs in solving CAPTCHAs and to assess how many attempts human users typically need to pass them. Our findings reveal that while LLMs can solve most CAPTCHAs, they struggle with those requiring complex reasoning-a type of CAPTCHA that also presents significant challenges for human users. Interestingly, our user study shows that the majority of human participants require a second attempt to pass these reasoning CAPTCHAs, a finding not reported in previous research. Based on empirical findings, we present IllusionCAPTCHA, a novel security mechanism employing the "Human-Easy but AI-Hard" paradigm. This new CAPTCHA employs visual illusions to create tasks that are intuitive for humans but highly confusing for AI models. Furthermore, we developed a structured, step-by-step method that generates misleading options, which particularly guide LLMs towards making incorrect choices and reduce their chances of successfully solving CAPTCHAs. Our evaluation shows that IllusionCAPTCHA can effectively deceive LLMs 100% of the time. Moreover, our structured design significantly increases the likelihood of AI errors when solving these challenges. Results from our user study indicate that 86.95% of participants successfully passed the CAPTCHA on their first attempt, outperforming other CAPTCHA systems.
Multi-modal sequential recommendation (SR) leverages multi-modal data to learn more comprehensive item features and user preferences than traditional SR methods, which has become a critical topic in both academia and industry. Existing methods typically focus on enhancing multi-modal information utility through adaptive modality fusion to capture the evolving of user preference from user-item interaction sequences. However, most of them overlook the interference caused by redundant interest-irrelevant information contained in rich multi-modal data. Additionally, they primarily rely on implicit temporal information based solely on chronological ordering, neglecting explicit temporal signals that could more effectively represent dynamic user interest over time. To address these limitations, we propose a Hierarchical time-aware Mixture of experts for multi-modal Sequential Recommendation (HM4SR) with a two-level Mixture of Experts (MoE) and a multi-task learning strategy. Specifically, the first MoE, named Interactive MoE, extracts essential user interest-related information from the multi-modal data of each item. Then, the second MoE, termed Temporal MoE, captures user dynamic interests by introducing explicit temporal embeddings from timestamps in modality encoding. To further address data sparsity, we propose three auxiliary supervision tasks: sequence-level category prediction (CP) for item feature understanding, contrastive learning on ID (IDCL) to align sequence context with user interests, and placeholder contrastive learning (PCL) to integrate temporal information with modalities for dynamic interest modeling. Extensive experiments on four public datasets verify the effectiveness of HM4SR compared to several state-of-the-art approaches.
BoxCD: Leveraging Contrastive Probabilistic Box Embedding for Effective and Efficient Learner Modeling
PDF ↗In digital education, Cognitive Diagnosis (CD) is essential for modeling learners' cognitive states, such as problem-solving ability and knowledge proficiency, by analyzing their response data, like answer correctness. However, traditional CD methods struggle with effectiveness and efficiency. They fail to capture the diversity and uncertainty of learners' cognitive states. Additionally, response prediction can be time-consuming. To address these issues, we propose BoxCD, a contrastive probabilistic box embedding model for cognitive diagnosis. BoxCD utilizes high-dimensional axis-aligned hyper-rectangles (boxes) to represent learners and exercises, with the volume of intersecting boxes used to predict learners' responses. This approach effectively captures semantic diversity and uncertainty while enhancing diagnostic effectiveness. To stabilize box embeddings, we integrate contrastive learning objectives with response prediction goals, optimizing the distance between positive and negative samples of learner and exercise boxes to improve uniformity. Additionally, we develop a rank-based response prediction method that leverages the geometric properties of box embeddings to assess learners' response correctness efficiently. Comprehensive experiments on two real-world datasets demonstrate that BoxCD outperforms traditional CD models in effectiveness and efficiency. This showcases its potential to enhance personalized learning in digital education platforms.
Blockchain-based distributed storage systems offer enhanced security, transparency, and lower costs compared to traditional centralized storage, making them ideal for peer-to-peer collaboration. However, with the trend towards the Web of Things (WoT), lower transaction speeds and higher computational requirements limit their access to high-density data such as IoT. To address this, we propose TimeChain, an efficient off-chain blockchain storage system for IoT time series data. TimeChain batches discrete time series data, storing only the hash value of each batch on-chain while keeping the complete data off-chain. This significantly reduces storage overhead on the blockchain and storage latency by 37.4 times. TimeChain adopts an adaptive packaging mechanism to reduce the additional latency in range queries by converting the batch processing problem into a graph partitioning problem. To reduce the overhead of node selection, TimeChain integrates a node selection mechanism based on consensus protocol, combining node selection and consensus processes together. TimeChain also proposes a Locality-Sensitive Hashing tree-based data integrity verification mechanism to reduce transmission size. Our evaluation shows a reduction in query latency by 64.6% and storage latency by 35.3% compared to existing systems.
Personalized algorithms can inadvertently expose users to discomforting recommendations, potentially triggering negative consequences. The subjectivity of discomfort and the black-box nature of these algorithms make it challenging to effectively identify and filter such content. To address this, we first conducted a formative study to understand users' practices and expectations regarding discomforting recommendation filtering. Then, we designed a Large Language Model (LLM)-based tool named DiscomfortFilter, which constructs an editable preference profile for a user and helps the user express filtering needs through conversation to mask discomforting preferences within the profile. Based on the edited profile, DiscomfortFilter facilitates the discomforting recommendations filtering in a plug-and-play manner, maintaining flexibility and transparency. The constructed preference profile improves LLM reasoning and simplifies user alignment, enabling a 3.8B open-source LLM to rival top commercial models in an offline proxy task. A one-week user study with 24 participants demonstrated the effectiveness of DiscomfortFilter, while also highlighting its potential impact on platform recommendation outcomes. We conclude by discussing the ongoing challenges, highlighting its relevance to broader research, assessing stakeholder impact, and outlining future research directions.
Decentralized identity (DID) holds significant potential for applications in Web3, such as digital markets and financial systems. Traditional DID paradigms offer a degree of privacy but struggle to prevent the link analysis on user behaviours and repeated public key usage. Anonymity is not fully achieved, as users' real identities or public keys are exposed to the issuing authority, while introducing high public key management complexity. Besides, existing anonymous credential schemes lack effective mechanisms for threshold traceability, not meeting Web3's distributed governance requirements. In this paper, we propose FADID-TT, a Fully Anonymous DID system supporting Threshold Tracing with practical blockchain, to tackle the above challenges. Firstly, we propose a distributed identity registration scheme based on secret sharing. A committee composed of distributed issuing authorities is responsible for issuing user's secret key shares and no single entity in the system can obtain a user's real identity or public key, achieving anonymity to authority. Moreover, we design a fully anonymous DID system combined with anonymous signatures and decentralized anonymous credentials (DAC). A service provider can only use the committee public key to verify a user identity, eliminating the need for user public keys, fully resisting link attacks, and reducing the user public key management complexity from O(n) to O(1). Furthermore, we design a publicly verifiable threshold tracing mechanism that enables committee members to collaboratively trace the identity of a malicious user without compromising privacy guarantees. FADID-TT realizes publicly verifiable tracing via zero-knowledge proofs. Finally, we give comprehensive security analysis and concrete performance evaluation. In addition to evaluate each part of proposal, we also deploy FADID-TT on two well-known blockchain platforms including Hyperledger Fabric (permissioned) and Ethereum (permissionless) to demonstrate the practical feasibility of FADID-TT.
The rapid spread of rumors on social media has posed significant challenges to maintaining public trust and information integrity. Since an information cascade process is essentially a propagation tree, recent rumor detection models leverage graph neural networks to additionally capture information propagation patterns, thus outperforming text-only solutions. Given the variations in topics and social impact of the root node, different source information naturally has distinct outreach capabilities, resulting in different heights of propagation trees. This variation, however, impedes the data-driven design of existing graph-based rumor detectors. Given a shallow propagation tree with limited interactions, it is unlikely for graph-based approaches to capture sufficient cascading patterns, questioning their ability to handle less popular news or early detection needs. In contrast, a deep propagation tree is prone to noisy user responses, and this can in turn obfuscate the predictions. In this paper, we propose a novel Epidemiology-informed Network (EIN) that integrates epidemiological knowledge to enhance performance by overcoming data-driven methods' sensitivity to data quality. Meanwhile, to adapt epidemiology theory to rumor detection, it is expected that each user's stance toward the source information will be annotated. To bypass the costly and time-consuming human labeling process, we take advantage of large language models to generate stance labels, facilitating optimization objectives for learning epidemiology-informed representations. Our experimental results demonstrate that the proposed EIN not only outperforms state-of-the-art methods on real-world datasets but also exhibits enhanced robustness across varying tree depths.
Dynamic graph link prediction is widely utilized in the complex web of the real world, such as social networks, citation networks, recommendation systems, etc. Recent Transformer-based link prediction methods on dynamic graphs not only fail to model the fine-grained structures such as triangles with the vanilla Transformers in the graph serialization process, but also amplify the imbalanced distribution of graphs because of their over-estimation of high-degree nodes. To tackle these issues, we propose a Topology-aware Transformer on Dynamic Graph (TopDyG) for link prediction, consisting of a topology injected Transformer (Ti-Transformer) and a mutual information learning (Mi-Learning). The Ti-Transformer explores the explicit structure of serialized graphs, capturing the topological features. The Mi-Learning mines the relationship between nodes by modeling the mutual information with a prior knowledge, alleviating the over-estimation of high-degree nodes when applying the Transformer-based models for the dynamic graph link prediction task. Extensive experiments on four public datasets containing both transductive and inductive settings present the superiority of our proposal. In particular, TopDyG presents an improvement of 43.27% and 28.75% against the state-of-the-art baselines in terms of NDCG and Jaccard, respectively. The advantages are especially obvious on the high-density graphs.
HeatSnap: A Hot Page-Aware Continuous Snapshots System for Virtual Machines in Web Infrastructure
PDF ↗Snapshot technology is crucial for data protection and system recovery in virtualized environments, particularly with the growing need for continuous snapshots to maintain the integrity of long-running web-based and distributed applications. However, traditional snapshot methods often suffer from performance bottlenecks, and inefficient storage usage. These challenges are closely tied to the way memory pages are accessed during VM execution, where memory access patterns show significant disparities between frequently accessed "hot" pages and less-used "cold" pages. In this paper, we introduce HeatSnap, a continuous snapshot system designed to address these issues by leveraging the uneven access frequencies of memory pages. HeatSnap distinguishes between intensive hot pages and dirty pages, applying specialized snapshotting and storage strategies to optimize the handling of both hot and cold memory regions. This approach aims to optimize snapshot efficiency, minimize performance impact on the VM, and decrease storage costs. Our implementation of HeatSnap on QEMU/KVM demonstrates significant improvements in VM performance loss, snapshot duration, and storage efficiency compared to existing methods, as evidenced by evaluations on common web and cloud-based workloads.
Graph prompt tuning has achieved significant success for its ability to effectively adapt pre-trained graph neural networks to various downstream tasks. However, the pre-trained models may learn discriminatory representation due to the inherent prejudice in graph-structured data. Existing graph prompt tuning overlooks such unfairness, leading to biased outputs towards certain demographic groups determined by sensitive attributes such as gender, age, and political ideology. To overcome this limitation, we propose a fairness-aware graph prompt tuning method to promote fairness while enhancing the generality of any pre-trained GNNs (named FPrompt). FPrompt introduces hybrid graph prompts to augment counterfactual data while aligning the pre-training and downstream tasks. It also applies edge modification to increase sensitivity heterophily. We provide a two-fold theoretical analysis: first, we demonstrate that FPrompt possesses universal capabilities in handling pre-trained GNN models across various pre-training strategies, ensuring its adaptability in different scenarios. Second, we show that FPrompt effectively reduces the upper bound of generalized statistical parity, thereby mitigating the bias of pre-trained models. Extensive experiments demonstrate that FPrompt outperforms baseline models in both accuracy and fairness (33%) on benchmark datasets. Additionally, we introduce a new benchmark for transferable evaluation, showing that FPrompt achieves state-of-the-art generalization performance.
Modern online advertising systems often involve a substantial number of advertisers in each auction, which results in scalability issues. To address this challenge, two-stage auctions have been designed and implemented in practice. These auctions enable efficient allocation of ad slots among numerous candidate advertisers in a short response time. This approach employs a fast yet coarse model in the first stage to select a small subset of advertisers, followed by a slow, more refined model to determine the final winners. However, existing two-stage auction mechanisms primarily focus on optimizing welfare, overlooking other critical objectives of the platform, such as revenue. In this paper, we propose ad-wise selection metrics, named Max-Wel and Max-Rev, which optimize the platform's welfare and revenue, respectively. These metrics are based on each ad's contribution to the corresponding objective function. We also provide theoretical guarantees for the proposed metrics. Our method is applicable to both welfare and revenue optimizations and can be easily implemented using neural networks. Through extensive experiments conducted on both synthetic and industrial data, we demonstrate the advantages of our proposed selection metrics compared to existing baselines.
Temporal knowledge graph completion (TKGC) seeks to develop more comprehensive knowledge representations by addressing missing relationships and entities within temporal knowledge graphs (TKGs), thereby enhancing reasoning and predictive capabilities in downstream tasks. Nonetheless, real-world knowledge-such as the progression of social network interactions and the unfolding of news events-is inherently dynamic, resulting in substantial sparsity issues in TKGs that profoundly impair the performance of TKGC models. To overcome this challenge, we introduce the Adaptive Neighborhood Enhancement Layer (ANEL), a novel module that can be effortlessly integrated into existing TKGC models to substantially elevate the representation quality of sparse entities. ANEL first derives initial entity embeddings through a base model and then uncovers concealed semantic relationships between entities via a latent relation module, enriching the explicit relationships within the knowledge graph. Furthermore, ANEL incorporates an adaptive latent information adjustment component, which dynamically calibrates the influence of latent information based on the entity's relational structure: entities with fewer connections derive greater benefit from latent information, while entities with denser connections become less dependent on latent augmentation, ensuring precise and resilient representations. We conducted comprehensive experiments on four prominent benchmark datasets, and the results underscore the effectiveness and superiority of ANEL in TKGC tasks.
Graph neural networks (GNNs) with differential privacy (DP) offer a reliable solution for safeguarding sensitive information within graph data. Nonetheless, existing DP-based privacy-preserving GNN learning frameworks generally overlook the local topological heterogeneity of graph nodes and tailor the same privacy budget for all nodes, which may lead to either overprotection or underprotection of some nodes, potentially diminishing model utility or posing privacy leakage risks. To address this issue, we propose a Topology-aware Differential Privacy Graph Neural Network learning framework, termed TDP-GNN, which can achieve personalized privacy protection for each node with improved privacy-utility guarantees. Specifically, TDP-GNN first identifies the topological importance of each node via an adjacency information entropy method. Then, the personalized topology-aware privacy budget is designed to quantify the privacy sensitivity of each node and adaptively allocate the privacy protection strength. Besides, a weighted neighborhood aggregation mechanism is proposed during the message-passing process of GNN training, which can eliminate the impact of the introduced differentiated DP noise on the utility of the GNN model. Since TDP-GNN is based on node-level local DP, it can be seamlessly integrated into any GNN architecture in a plug-and-play manner while ensuring formal privacy guarantees. Theoretical analysis indicates that TDP-GNN achieves ε-differential privacy over the entire graph nodes while providing personalized privacy protection. Extensive experiments demonstrate that TDP-GNN consistently yields better utilities when applied to various GNN architectures (e.g., GCN and GraphSAGE) across a diverse set of benchmarks.
MSDZip: Universal Lossless Compression for Multi-source Data via Stepwise-parallel and Learning-based Prediction
PDF ↗With the rapid development of the Internet, the huge amount of Multi-Source Data (MSD) brings challenges in data sharing and storing. Lossless data compression is the major way to solve those problems. Nowadays, neural-network technologies bring significant advantage in data modeling, making learning-based lossless compressors (LLCs) for multi-source data have emerged continuously. Compared with traditional compressors, the LLCs are more useful to catch complex redundancy patterns in MSD, and thus have great potential in enhancing compression ratio. However, existing LLCs still suffer from unsatisfactory compression ratios and lower throughput. To solve those problems, we propose a novel universal MSD lossless compressor called MSDZip via Stepwise-parallel and learning-based prediction technologies, it introduces two major designs: 1) We propose a Local-Global-Deep Mixing block in the learning-based prediction module to establish dependencies for MSD symbols, where designed Deep Mixing block solves the problem of unstable weights in the perceptual layers caused by cold-start problem to enhance the compression ratio significantly. 2) We design a Stepwise-parallel multi-GPU-accelerated compression strategy to address the compression speed and graphics memory constraints of single GPU in the face of large-scale data. The Stepwise-parallel module passes the source MSD to learning-based prediction model through the data chunking strategy, where the model of the previous chunk is used to guide the compression of the next chunk in parallel. We compare MSDZip with 5 classical learning-based and 6 traditional compressors on 12 well-studied real-world datasets. The experimental results demonstrate that MSDZip optimizes 3.418%-69.874% in terms of compression ratio and 31.171%-495.649% in terms of throughput compared to advanced LLCs. The source code of MSDZip and the linkages of the experimental datasets are available at https://github.com/mhuidong/MSDZip.
Many ethical issues in machine learning are connected to the training data. Online data markets are an important source of training data, facilitating both production and distribution. Recently, a trend has emerged of for-profit ''ethical'' participants in online data markets. This trend raises a fascinating question: Can online data markets sustainably and efficiently address ethical issues in the broader machine-learning economy? In this work, we study this question in a stylized model of an online data market. We investigate the effects of intervening in the data market to achieve balanced training-data production. The model reveals the crucial role of market conditions. In small and emerging markets, an intervention can drive the data producers out of the market, so that the cost of fairness is maximal. Yet, in large and established markets, the cost of fairness can vanish (as a fraction of overall welfare) as the market grows. Our results suggest that ''ethical'' online data markets can be economically feasible under favorable market conditions, and motivate more models to consider the role of data production and distribution in mediating the impacts of ethical interventions.
Large Language Models (LLMs) have achieved impressive results in various tasks but struggle with hallucination problems and lack of relevant knowledge, especially in deep complex reasoning and knowledge-intensive tasks. Knowledge Graphs (KGs), which capture vast amounts of facts in a structured format, offer a reliable source of knowledge for reasoning. However, existing KG-based LLM reasoning methods face challenges like handling multi-hop reasoning, multi-entity questions, and effectively utilizing graph structures. To address these issues, we propose Paths-over-Graph (PoG), a novel method that enhances LLM reasoning by integrating knowledge reasoning paths from KGs, improving the interpretability and faithfulness of LLM outputs. PoG tackles multi-hop and multi-entity questions through a three-phase dynamic multi-hop path exploration, which combines the inherent knowledge of LLMs with factual knowledge from KGs. In order to improve the efficiency, PoG prunes irrelevant information from the graph exploration first and introduces efficient three-step pruning techniques that incorporate graph structures, LLM prompting, and a pre-trained language model (e.g., SBERT) to effectively narrow down the explored candidate paths. This ensures all reasoning paths contain highly relevant information captured from KGs, making the reasoning faithful and interpretable in problem-solving. PoG innovatively utilizes graph structure to prune the irrelevant noise and represents the first method to implement multi-entity deep path detection on KGs for LLM reasoning tasks. Comprehensive experiments on five benchmark KGQA datasets demonstrate PoG outperforms the state-of-the-art method ToG across GPT-3.5-Turbo and GPT-4, achieving an average accuracy improvement of 18.9%. Notably, PoG with GPT-3.5-Turbo surpasses ToG with GPT-4 by up to 23.9%.
Mitigating Forgetting in Adapting Pre-trained Language Models to Text Processing Tasks via Consistency Alignment
PDF ↗There are a large number of text processing tasks in web applications, such as sentiment classification, summary extraction, and question answering. Recently, fine-tuning pre-trained language models (PLMs) to adapt to downstream text-processing tasks has attracted much attention. However, due to the differences in data, model, and tasks between the pre-training and fine-tuning processes, the fine-tuning process may suffer from catastrophic forgetting of pre-training knowledge, which may implicitly limit the model's performance and generalization ability. To address these challenges, we propose a novel dual-model framework, termed as consistency alignment (CoAi). The insight of CoAi lies in building an auxiliary model that simulates the distribution of pre-training knowledge in real-time according to the current task, and co-training the task-specific model and the auxiliary model to balance the pre-training knowledge and task-specific knowledge during fine-tuning. Specifically, the auxiliary model is constructed on-the-fly to maintain the pre-training knowledge. Subsequently, CoAi simulates the pre-training process by performing distributional exploration in the parameter space, which is built upon our novel insight into the transformation between data and model parameter space. However, the objectives leveraged to construct the auxiliary model lead to the misalignment between the pre-training and task-specific knowledge. To alleviate the inconsistency, we employ an auxiliary variable to align the prediction distribution of the task-specific and the auxiliary models, inspired by constrastive clustering. We validate the effectiveness of CoAi on nine classic classification tasks and three generation tasks, showing consistent and significant improvements compared with state-of-the-art methods.
Multimodal Knowledge Graph Error Detection with Disentanglement VAE and Multi-Grained Triplet Confidence
PDF ↗Multimodal knowledge graphs inevitably contain numerous errors due to the lack of human supervision in their automated construction and updating processes. These errors can significantly degrade the performance of downstream applications that rely on them. Existing researches on knowledge graph error detection primarily focus on leveraging graph structural and textual information to identify triplet errors in unimodal knowledge graphs. However, unlike unimodal knowledge graphs, multimodal knowledge graphs also suffer from mismatches between images and their corresponding entities, referred to as modality errors. These modality errors not only hinder the performance of downstream applications but also impede our effective utilization of the abundant complementary information provided by the visual modality for detecting triplet errors. To this end, we introduce a novel task of multimodal knowledge graph error detection (MKGED) in this paper, aiming at simultaneously identifying both modality errors and triplet errors. Given the lack of datasets for evaluating this task, we first establish two comprehensive MKGED datasets. Furthermore, we propose a novel framework, KGDMC, to address the MKGED task. Within KGDMC, we devise a disentanglement modality reconstruction (DMR) module for modality error detection. This module disentangles each original modality representation into two disjoint components: modality-specific representations and modality-invariant representations, leveraging the cross-modality reconstruction process to detect mismatched visual modalities. Additionally, for the triplet error detection, we propose a multi-grained triplet confidence (MTC) module, incorporating local triplet confidence, global structure confidence, and global path confidence, to collaboratively detect mismatched triplets. Extensive experiments on our constructed two datasets demonstrate the superiority of our proposed framework.