Pub Date : 2025-12-09DOI: 10.1016/j.ipm.2025.104550
Jiawei Chen , Hong Chen
In the context of the artificial intelligence (AI) revolution, public perceptions are complex and diverse regarding whether AI signifies a “jobpocalypse” or ushers in a “techno-utopia”. To decode public sentiment and perception regarding AI’s impact on employment, this study captures related public discussion texts (40,299 in total) from Weibo and Douyin. Word cloud visualization presents key public concerns, Word2Vec reveals semantic associations between keywords, and BERTopic analyzes the cognitive focus and thematic characteristics of public attention. Additionally, social media and geographic information are integrated to reveal regional heterogeneity. The research findings indicate: (1) public perceptions show obvious emotional polarity, yet the overall expression tends to be cautious and rational. (2) Public perceptions are multidimensional (10 topics), focusing on human-machine collaboration, technological unemployment, industry applications, and risk expectations. (3) The primary focuses of the two platforms overlap in some areas but also differ in others. (4) An “AI divide” exists across regions. The eastern region emphasizes technological rationality and international comparison, the central region prioritizes technological empowerment and social harmony, while the western region concentrates on unemployment risk and social impact.
{"title":"Jobpocalypse or techno-utopia? geospatially decoding public concerns through the social media noise in AI’s disruption era","authors":"Jiawei Chen , Hong Chen","doi":"10.1016/j.ipm.2025.104550","DOIUrl":"10.1016/j.ipm.2025.104550","url":null,"abstract":"<div><div>In the context of the artificial intelligence (AI) revolution, public perceptions are complex and diverse regarding whether AI signifies a “jobpocalypse” or ushers in a “techno-utopia”. To decode public sentiment and perception regarding AI’s impact on employment, this study captures related public discussion texts (40,299 in total) from Weibo and Douyin. Word cloud visualization presents key public concerns, Word2Vec reveals semantic associations between keywords, and BERTopic analyzes the cognitive focus and thematic characteristics of public attention. Additionally, social media and geographic information are integrated to reveal regional heterogeneity. The research findings indicate: (1) public perceptions show obvious emotional polarity, yet the overall expression tends to be cautious and rational. (2) Public perceptions are multidimensional (10 topics), focusing on human-machine collaboration, technological unemployment, industry applications, and risk expectations. (3) The primary focuses of the two platforms overlap in some areas but also differ in others. (4) An “AI divide” exists across regions. The eastern region emphasizes technological rationality and international comparison, the central region prioritizes technological empowerment and social harmony, while the western region concentrates on unemployment risk and social impact.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104550"},"PeriodicalIF":6.9,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737437","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-09DOI: 10.1016/j.ipm.2025.104537
Hongwei Ding , Songyu Wang , Xiaoming Yuan , Nana Huang , Xiaohui Cui
Class imbalance remains a major challenge in real-world classification tasks. To address this, we propose Diffusion-Enhanced Ensemble Learning (DEEL), a unified framework that synergistically integrates diffusion-based data augmentation and multi-objective ensemble optimization for binary classification tasks. Specifically, we design a Dynamic Attention Diffusion Model (DADM) to generate diverse and realistic minority class samples through a forward noise and reverse denoising process. By incorporating temporal embeddings, residual connections, and attention mechanisms, DADM enhances the fidelity and distributional alignment of the generated data. Complementing this, an ensemble learning strategy based on the Non-dominated Sorting Genetic Algorithm II (NSGA-II) optimizes the fusion of multiple base classifiers across F1-score, G-mean, and AUC metrics. Extensive experiments on 26 real-world imbalanced datasets demonstrate that DEEL improves average F1-score and G-mean by 21.7 % and 24.8 %, respectively, over competitive baselines. Moreover, visualization and Jensen-Shannon distance analyses quantitatively verify the high diversity and distributional coherence of DADM-generated samples, underscoring their effectiveness for imbalanced learning.
{"title":"DEEL: An imbalanced binary data classification method based on diffusion model data augmentation and multi-objective optimization ensemble","authors":"Hongwei Ding , Songyu Wang , Xiaoming Yuan , Nana Huang , Xiaohui Cui","doi":"10.1016/j.ipm.2025.104537","DOIUrl":"10.1016/j.ipm.2025.104537","url":null,"abstract":"<div><div>Class imbalance remains a major challenge in real-world classification tasks. To address this, we propose Diffusion-Enhanced Ensemble Learning (DEEL), a unified framework that synergistically integrates diffusion-based data augmentation and multi-objective ensemble optimization for binary classification tasks. Specifically, we design a Dynamic Attention Diffusion Model (DADM) to generate diverse and realistic minority class samples through a forward noise and reverse denoising process. By incorporating temporal embeddings, residual connections, and attention mechanisms, DADM enhances the fidelity and distributional alignment of the generated data. Complementing this, an ensemble learning strategy based on the Non-dominated Sorting Genetic Algorithm II (NSGA-II) optimizes the fusion of multiple base classifiers across F1-score, G-mean, and AUC metrics. Extensive experiments on 26 real-world imbalanced datasets demonstrate that DEEL improves average F1-score and G-mean by 21.7 % and 24.8 %, respectively, over competitive baselines. Moreover, visualization and Jensen-Shannon distance analyses quantitatively verify the high diversity and distributional coherence of DADM-generated samples, underscoring their effectiveness for imbalanced learning.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104537"},"PeriodicalIF":6.9,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-09DOI: 10.1016/j.ipm.2025.104546
Weiwei Yan , Yanyan Wang , Jiahui Song , Yin Zhang
This study investigates cross-platform behavior on academic social networking sites (ASNSs), focusing on differences among users from academic, government, and corporate institutions. Users often engage with multiple ASNSs due to differing platform features and contexts, leading to distinct behavioral patterns. Drawing on data from Academia.edu (ACA) and ResearchGate (RG), this study analyzes user profiles from 15 institutions to identify cross-platform users and compare behaviors. It proposes an approach for identifying such users and develops a cross-platform user behavior indicator system to support the analysis. A clustering analysis further explores behavior patterns and provides additional insights into cross-platform engagement. Findings show that cross-platform users tend to disclose more information, maintain broader networks, and engage more actively on RG than on ACA. Government-affiliated users are the most active, with high levels of disclosure, publication, and interaction. Corporate users exhibit varied strengths and weaknesses, while academic users demonstrate moderate activity. Most academic cross-platform users fall into a “civilian-type” category, sharing fewer publications and presenting inconsistent profile information. In contrast, many government and corporate users are ”star-type,” showing greater consistency and visibility across platforms. This study advances understanding of cross-platform ASNS behavior and reveals sector-based differences that may inform platform design and user strategies.
{"title":"Analyzing cross-platform academic networking behavior: Methods and insights on institutional affiliations and user clustering","authors":"Weiwei Yan , Yanyan Wang , Jiahui Song , Yin Zhang","doi":"10.1016/j.ipm.2025.104546","DOIUrl":"10.1016/j.ipm.2025.104546","url":null,"abstract":"<div><div>This study investigates cross-platform behavior on academic social networking sites (ASNSs), focusing on differences among users from academic, government, and corporate institutions. Users often engage with multiple ASNSs due to differing platform features and contexts, leading to distinct behavioral patterns. Drawing on data from Academia.edu (ACA) and ResearchGate (RG), this study analyzes user profiles from 15 institutions to identify cross-platform users and compare behaviors. It proposes an approach for identifying such users and develops a cross-platform user behavior indicator system to support the analysis. A clustering analysis further explores behavior patterns and provides additional insights into cross-platform engagement. Findings show that cross-platform users tend to disclose more information, maintain broader networks, and engage more actively on RG than on ACA. Government-affiliated users are the most active, with high levels of disclosure, publication, and interaction. Corporate users exhibit varied strengths and weaknesses, while academic users demonstrate moderate activity. Most academic cross-platform users fall into a “civilian-type” category, sharing fewer publications and presenting inconsistent profile information. In contrast, many government and corporate users are ”star-type,” showing greater consistency and visibility across platforms. This study advances understanding of cross-platform ASNS behavior and reveals sector-based differences that may inform platform design and user strategies.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104546"},"PeriodicalIF":6.9,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-08DOI: 10.1016/j.ipm.2025.104511
Ganlin Xu , Jianzhou Feng , Qin Wang
Document-level relation extraction (DocRE) aims to infer semantic relations between entity pairs1 in a document. Generation-based methods for DocRE only learn superficial text patterns from plain text instead of logical rule patterns while generating uncontrolled outputs. Therefore, this paper proposes a novel generative paradigm, a rule learning and elements alignment (RLEA) method for DocRE. We build a symmetrical structure using two T5 models (text learner and rule learner), where the text learner learns text patterns from symbolic triplets, and the rule learner learns rule patterns from chain-like logic rules. To better solve the above challenges, we proposed three key techniques: the bidirectional gate function, the rule regularizer, and the alignment mechanism. The experimental results indicate that our method achieves state-of-the-art results in relation extraction and logical consistency, with RLEA obtaining 72.37, 79.44 and 94.52 on DWIE w.r.t Ign F1, F1 and Logic respectively, 61.94 and 63.96 on DocRED w.r.t Ign F1 and F1 respectively, 76.81 and 77.06 on Re-DocRED w.r.t Ign F1 and F1 respectively. Besides, quantitative experiments and qualitative analysis show how logical rules work on black-box generation-based models2 for DocRE.
{"title":"Learning rules and aligning elements for document-level relation extraction","authors":"Ganlin Xu , Jianzhou Feng , Qin Wang","doi":"10.1016/j.ipm.2025.104511","DOIUrl":"10.1016/j.ipm.2025.104511","url":null,"abstract":"<div><div>Document-level relation extraction (DocRE) aims to infer semantic relations between entity pairs<span><span><sup>1</sup></span></span> in a document. Generation-based methods for DocRE only learn superficial text patterns from plain text instead of logical rule patterns while generating uncontrolled outputs. Therefore, this paper proposes a novel generative paradigm, a rule learning and elements alignment (RLEA) method for DocRE. We build a symmetrical structure using two T5 models (text learner and rule learner), where the text learner learns text patterns from symbolic triplets, and the rule learner learns rule patterns from chain-like logic rules. To better solve the above challenges, we proposed three key techniques: the bidirectional gate function, the rule regularizer, and the alignment mechanism. The experimental results indicate that our method achieves state-of-the-art results in relation extraction and logical consistency, with RLEA obtaining 72.37, 79.44 and 94.52 on DWIE w.r.t <strong>Ign F1, F1</strong> and <strong>Logic</strong> respectively, 61.94 and 63.96 on DocRED w.r.t <strong>Ign F1</strong> and <strong>F1</strong> respectively, 76.81 and 77.06 on Re-DocRED w.r.t <strong>Ign F1</strong> and <strong>F1</strong> respectively. Besides, quantitative experiments and qualitative analysis show how logical rules work on black-box generation-based models<span><span><sup>2</sup></span></span> for DocRE.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104511"},"PeriodicalIF":6.9,"publicationDate":"2025-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-08DOI: 10.1016/j.ipm.2025.104506
Alessandro Fabris , Clara Rus , Jorge Saldivar , Anna Gatzioura , Asia J. Biega , Carlos Castillo
Personnel recruitment is increasingly mediated by Applicant Tracking Systems (ATS), which rank candidates for job positions, making them a central decision-support tool in modern Human Resources (HR) processes. Often framed as an information retrieval (IR) problem, the ranking of candidates in ATS is typically driven by relevance to the job position, with algorithms sorting applicants according to a set of predefined criteria. In recent years, fairness-aware ranking methods have emerged to mitigate the risk of indirect discrimination, where the ordering of candidates may inadvertently favor one demographic group over another. These approaches are inspired by browsing models developed for web search and aim to balance candidate exposure based on protected characteristics. However, ATS in recruitment introduce unique challenges due to their high-stakes nature and the decision-making context in which they operate. In this paper, we present a series of user studies that explore the disconnect between fair exposure and fair outcomes in candidate shortlisting. We focus on how factors such as task design (e.g., how recruiters interact with candidate lists), individual representations of candidates (e.g., national origin cues), and ranking order influence both position bias and demographic balance. Our findings show that while demographic balance may be achieved in terms of ranking visibility, this does not necessarily translate to fair outcomes in terms of who gets shortlisted. Through a crowdsourced experiment and in-depth interviews with recruiters, we identify key task-level, individual, and ranking factors that mediate these effects. We conclude that fairness in ATS rankings is contingent not only on algorithmic design but also on the shortlisting tasks they support, as well as the interfaces, strategies, and assumptions that recruiters use when interacting with candidate lists. Based on these insights, we provide implications for the design of algorithms, interfaces, and recruitment processes that support fairer and more equitable recruitment outcomes.
{"title":"Does fair ranking lead to fair recruitment outcomes? A study of interventions, interfaces, and interactions","authors":"Alessandro Fabris , Clara Rus , Jorge Saldivar , Anna Gatzioura , Asia J. Biega , Carlos Castillo","doi":"10.1016/j.ipm.2025.104506","DOIUrl":"10.1016/j.ipm.2025.104506","url":null,"abstract":"<div><div>Personnel recruitment is increasingly mediated by Applicant Tracking Systems (ATS), which rank candidates for job positions, making them a central decision-support tool in modern Human Resources (HR) processes. Often framed as an information retrieval (IR) problem, the ranking of candidates in ATS is typically driven by relevance to the job position, with algorithms sorting applicants according to a set of predefined criteria. In recent years, fairness-aware ranking methods have emerged to mitigate the risk of indirect discrimination, where the ordering of candidates may inadvertently favor one demographic group over another. These approaches are inspired by browsing models developed for web search and aim to balance candidate exposure based on protected characteristics. However, ATS in recruitment introduce unique challenges due to their high-stakes nature and the decision-making context in which they operate. In this paper, we present a series of user studies that explore the disconnect between <em>fair exposure</em> and <em>fair outcomes</em> in candidate shortlisting. We focus on how factors such as task design (e.g., how recruiters interact with candidate lists), individual representations of candidates (e.g., national origin cues), and ranking order influence both position bias and demographic balance. Our findings show that while demographic balance may be achieved in terms of ranking visibility, this does not necessarily translate to fair outcomes in terms of who gets shortlisted. Through a crowdsourced experiment and in-depth interviews with recruiters, we identify key task-level, individual, and ranking factors that mediate these effects. We conclude that fairness in ATS rankings is contingent not only on algorithmic design but also on the shortlisting tasks they support, as well as the interfaces, strategies, and assumptions that recruiters use when interacting with candidate lists. Based on these insights, we provide implications for the design of algorithms, interfaces, and recruitment processes that support fairer and more equitable recruitment outcomes.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104506"},"PeriodicalIF":6.9,"publicationDate":"2025-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737423","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-06DOI: 10.1016/j.ipm.2025.104533
Chenhao Li, Wenti Huang, Xi Yu, Tingxuan Chen, Jun Long
Multimodal Emotion Recognition (MER) enhances the understanding and identification of human emotions by combining multiple sensory signals, such as speech, facial expressions, body language, and text. In the task of MER, temporal discrepancies and differences in emotional expression between the audio and video modalities hinder the effective alignment of modality features, thereby affecting the accuracy of emotion recognition. We propose an Adaptive Weighted Temporal Prototype Network (AWTPN) to address this issue. The Temporal Prototype Network learns prototype features for each emotion category from audio and visual modalities. At the same time, the adaptive weight framework automatically optimizes the modal features and temporal information for each emotion category, ensuring effective fusion between modalities. We conducted extensive experiments on widely-used datasets. Experimental results demonstrate that the proposed AWTPN achieves the best overall accuracy and weighted F1 score among all baseline methods on the IEMOCAP dataset, surpassing the baseline model SDT by 3.26 % and 4.17 % in average accuracy and average F1 score, respectively. Similarly, on the MELD dataset, AWTPN outperforms the baseline model CBERL by 5.40 % in average accuracy and 3.53 % in average F1 score. On the CMU-MOSEI dataset, AWTPN achieves a weighted F1 score of 61.29 %, surpassing the previous best method by 0.93 %. Moreover, AWTPN significantly improves accuracy across multiple emotion recognition tasks, consistently maintaining robust performance on all datasets.
{"title":"Adaptive weighted temporal prototype network for multimodal emotion recognition","authors":"Chenhao Li, Wenti Huang, Xi Yu, Tingxuan Chen, Jun Long","doi":"10.1016/j.ipm.2025.104533","DOIUrl":"10.1016/j.ipm.2025.104533","url":null,"abstract":"<div><div>Multimodal Emotion Recognition (MER) enhances the understanding and identification of human emotions by combining multiple sensory signals, such as speech, facial expressions, body language, and text. In the task of MER, temporal discrepancies and differences in emotional expression between the audio and video modalities hinder the effective alignment of modality features, thereby affecting the accuracy of emotion recognition. We propose an Adaptive Weighted Temporal Prototype Network (AWTPN) to address this issue. The Temporal Prototype Network learns prototype features for each emotion category from audio and visual modalities. At the same time, the adaptive weight framework automatically optimizes the modal features and temporal information for each emotion category, ensuring effective fusion between modalities. We conducted extensive experiments on widely-used datasets. Experimental results demonstrate that the proposed AWTPN achieves the best overall accuracy and weighted F1 score among all baseline methods on the IEMOCAP dataset, surpassing the baseline model SDT by 3.26 % and 4.17 % in average accuracy and average F1 score, respectively. Similarly, on the MELD dataset, AWTPN outperforms the baseline model CBERL by 5.40 % in average accuracy and 3.53 % in average F1 score. On the CMU-MOSEI dataset, AWTPN achieves a weighted F1 score of 61.29 %, surpassing the previous best method by 0.93 %. Moreover, AWTPN significantly improves accuracy across multiple emotion recognition tasks, consistently maintaining robust performance on all datasets.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104533"},"PeriodicalIF":6.9,"publicationDate":"2025-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145693775","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-06DOI: 10.1016/j.ipm.2025.104535
Qiangwei Li , Xin Gao , Yuan Li , Xinping Diao , Tianyang Chen , Yukun Lin , Taizhi Wang , Yu Hao
Fully mining the commonality and difference of different classes in overlapping areas is the key and difficult point to improving imbalanced classification performance. Existing data-level and algorithm-level methods heavily depend on distribution information or inter-class difference, limiting their ability to capture commonality information. This paper proposes an imbalanced classification framework with serialized neighbor samples commonality extraction and conditional variational latent space optimization. It achieves the sufficient extraction of commonality and difference information of different class samples in overlapping areas by adjusting the distribution of latent codes during the sample reconstruction process, mainly including two key modules. The inter-class commonality information learning module transforms tabular data into serialized neighbor sample groups, utilizes self-attention to extract inter-class commonality information, and quantifies it using cosine similarity. The conditional variational sample reconstruction module adjusts class distributions by leveraging inter-class commonality, so that the distance between the samples in the overlapping areas is closer in the latent space, thereby extracting more realistic discriminative features. Moreover, the variances consistency of the constrained latent codes is utilized to alleviate the classifier decision offset problem caused by diversity differences. Experiments on 50 imbalanced datasets demonstrate the proposed method outperforms most 25 typical imbalanced classification methods in F1-score and G-mean. In particular, the improvement is most significant on 20 datasets with serious inter-class overlap.
{"title":"An imbalanced classification framework with serialized neighbor samples commonality extraction and conditional variational latent space optimization","authors":"Qiangwei Li , Xin Gao , Yuan Li , Xinping Diao , Tianyang Chen , Yukun Lin , Taizhi Wang , Yu Hao","doi":"10.1016/j.ipm.2025.104535","DOIUrl":"10.1016/j.ipm.2025.104535","url":null,"abstract":"<div><div>Fully mining the commonality and difference of different classes in overlapping areas is the key and difficult point to improving imbalanced classification performance. Existing data-level and algorithm-level methods heavily depend on distribution information or inter-class difference, limiting their ability to capture commonality information. This paper proposes an imbalanced classification framework with serialized neighbor samples commonality extraction and conditional variational latent space optimization. It achieves the sufficient extraction of commonality and difference information of different class samples in overlapping areas by adjusting the distribution of latent codes during the sample reconstruction process, mainly including two key modules. The inter-class commonality information learning module transforms tabular data into serialized neighbor sample groups, utilizes self-attention to extract inter-class commonality information, and quantifies it using cosine similarity. The conditional variational sample reconstruction module adjusts class distributions by leveraging inter-class commonality, so that the distance between the samples in the overlapping areas is closer in the latent space, thereby extracting more realistic discriminative features. Moreover, the variances consistency of the constrained latent codes is utilized to alleviate the classifier decision offset problem caused by diversity differences. Experiments on 50 imbalanced datasets demonstrate the proposed method outperforms most 25 typical imbalanced classification methods in F1-score and G-mean. In particular, the improvement is most significant on 20 datasets with serious inter-class overlap.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104535"},"PeriodicalIF":6.9,"publicationDate":"2025-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-05DOI: 10.1016/j.ipm.2025.104536
Binglu Hu , Ying Su , Xuecong Tian , Chen Chen , Xiaoyi Lv , Cheng Chen
Disease metabolite association prediction is of key value for early diagnosis and treatment of diseases. However, existing computational methods face two main challenges: (1) heterogeneous graph information is not fully exploited; (2) most methods are limited to single view analysis, which makes it difficult to achieve effective information interaction between different types of nodes. To address these issues, we propose a computational method GHTMDA based on self-supervised heterogeneous graph learning and hierarchical contrastive learning. First, we fuse multiple similarity information of metabolites and diseases layer-by-layer by using two-tier bi-random walk, and utilize the Graph Transformer self-attention mechanism to achieve dynamic aggregation of neighbouring nodes and thus capture node representations at the global level. Then, a self-supervised heterogeneous graph learning mechanism is designed to capture both heterogeneity and homogeneity information in heterogeneous graphs and enhance the node representations by cross-view contrastive learning. Finally, effective information interaction and integration between different schemas is facilitated by cross-modal contrastive learning. The experimental results show that GHTMDA achieves 98.85 % and 98.87 % in AUC and AUPR, respectively, obviously outperforming the current state-of-the-art methods, and the prediction results in case validation such as colorectal cancer and Parkinson’s disease are highly consistent with the existing studies, which further confirms the reliability of the method. Code and data are available at: https://github.com/Ice-HL1/GHTMDA.
{"title":"GHTMDA: A self-supervised heterogeneous graph hierarchical contrastive learning model for efficient metabolite-disease associations prediction","authors":"Binglu Hu , Ying Su , Xuecong Tian , Chen Chen , Xiaoyi Lv , Cheng Chen","doi":"10.1016/j.ipm.2025.104536","DOIUrl":"10.1016/j.ipm.2025.104536","url":null,"abstract":"<div><div>Disease metabolite association prediction is of key value for early diagnosis and treatment of diseases. However, existing computational methods face two main challenges: (1) heterogeneous graph information is not fully exploited; (2) most methods are limited to single view analysis, which makes it difficult to achieve effective information interaction between different types of nodes. To address these issues, we propose a computational method GHTMDA based on self-supervised heterogeneous graph learning and hierarchical contrastive learning. First, we fuse multiple similarity information of metabolites and diseases layer-by-layer by using two-tier bi-random walk, and utilize the Graph Transformer self-attention mechanism to achieve dynamic aggregation of neighbouring nodes and thus capture node representations at the global level. Then, a self-supervised heterogeneous graph learning mechanism is designed to capture both heterogeneity and homogeneity information in heterogeneous graphs and enhance the node representations by cross-view contrastive learning. Finally, effective information interaction and integration between different schemas is facilitated by cross-modal contrastive learning. The experimental results show that GHTMDA achieves 98.85 % and 98.87 % in AUC and AUPR, respectively, obviously outperforming the current state-of-the-art methods, and the prediction results in case validation such as colorectal cancer and Parkinson’s disease are highly consistent with the existing studies, which further confirms the reliability of the method. Code and data are available at: <span><span>https://github.com/Ice-HL1/GHTMDA</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104536"},"PeriodicalIF":6.9,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145693773","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-05DOI: 10.1016/j.ipm.2025.104502
Like Xin , Wanqi Yang , Lei Wang , Ming Yang
In unpaired multi-view clustering (UMC), views have no paired observations, making direct sample matching difficult. Previous studies have used well-clustered views to guide poorly clustered ones for consistent structures, but discrepancies in view distributions can undermine this guidance. To address this issue, we construct a synthesized view and theoretically prove that the incorporation of a synthesized view could reduce inter-view discrepancies and enhance cluster compactness, both of which are key to improving the multi-view clustering performance. Based on these findings, we propose a new method called Global-and-Local guidANCE with synthesized-view for unpaired multi-view clustering (GLANCE). Specifically, a synthesized-view is initialized by combining the samples of views in the multi-view subspace and is guided by the well-clustered views. At the global level, the synthesized view guides the overall distribution of poorly clustered views. At the local level, it guides the distributions of corresponding clusters in the poorly clustered views. As demonstrated by the experiments conducted on five datasets with all views, GLANCE outperforms the relevant state-of-the-art methods, achieving an average improvement of 4.97 % in clustering accuracy as measured by NMI. The source code is available at: https://anonymous.4open.science/r/GLANCE-CECE.
{"title":"Global-and-local guidance with synthesized view for unpaired multi-view clustering","authors":"Like Xin , Wanqi Yang , Lei Wang , Ming Yang","doi":"10.1016/j.ipm.2025.104502","DOIUrl":"10.1016/j.ipm.2025.104502","url":null,"abstract":"<div><div>In unpaired multi-view clustering (UMC), views have no paired observations, making direct sample matching difficult. Previous studies have used well-clustered views to guide poorly clustered ones for consistent structures, but discrepancies in view distributions can undermine this guidance. To address this issue, we construct a synthesized view and theoretically prove that the incorporation of a synthesized view could reduce inter-view discrepancies and enhance cluster compactness, both of which are key to improving the multi-view clustering performance. Based on these findings, we propose a new method called Global-and-Local guidANCE with synthesized-view for unpaired multi-view clustering (GLANCE). Specifically, a synthesized-view is initialized by combining the samples of views in the multi-view subspace and is guided by the well-clustered views. At the global level, the synthesized view guides the overall distribution of poorly clustered views. At the local level, it guides the distributions of corresponding clusters in the poorly clustered views. As demonstrated by the experiments conducted on five datasets with all views, GLANCE outperforms the relevant state-of-the-art methods, achieving an average improvement of 4.97 % in clustering accuracy as measured by NMI. The source code is available at: <span><span>https://anonymous.4open.science/r/GLANCE-CECE</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104502"},"PeriodicalIF":6.9,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145693772","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-05DOI: 10.1016/j.ipm.2025.104508
Jing Zhang , HuiHui Yu , Rui Ye , Qun Dai
In Irregular Multivariate Time Series (IMTS) prediction, most methods address intra-series and inter-series irregularities through techniques like imputation and neural ordinary differential equations. However, they often overlook the multi-granularity nature of IMTS, limiting their ability to capture the dynamic, multi-granularity spatial and temporal interdependencies present in data. To overcome this, we propose HiP-IMTS, a novel Hierarchical Prediction model that enables accurate forecasting of IMTS across granularities, from coarse to fine, within a hierarchical framework. HiP-IMTS first extracts hierarchical patch embeddings at varying granularity levels. It then employs a finite difference-based attention mechanism for effectively addressing baseline drift, and a frequency convolution network for comprehensive temporal modeling from frequency domains. Next, a multi-granularity adaptive graph learning is introduced to model dynamic spatial correlations across different temporal granularities. Finally, a hierarchical prediction mechanism is designed to integrate complementary forecasting signals across multiple granularity levels, enabling effective fusion from coarse to fine scales. We perform a thorough assessment using four authentic datasets spanning various fields such as healthcare, biomechanics, and climate science, benchmarking HiP-IMTS against sixteen competitive baselines. HiP-IMTS achieves the best average critical difference ranks, with 1.1667 for MSE and 1.2083 for MAE, significantly outperforming existing state-of-the-art IMTS models.
{"title":"Hierarchical prediction of irregular multivariate time series from a multi-granularity perspective","authors":"Jing Zhang , HuiHui Yu , Rui Ye , Qun Dai","doi":"10.1016/j.ipm.2025.104508","DOIUrl":"10.1016/j.ipm.2025.104508","url":null,"abstract":"<div><div>In Irregular Multivariate Time Series (IMTS) prediction, most methods address intra-series and inter-series irregularities through techniques like imputation and neural ordinary differential equations. However, they often overlook the multi-granularity nature of IMTS, limiting their ability to capture the dynamic, multi-granularity spatial and temporal interdependencies present in data. To overcome this, we propose <strong>HiP-IMTS</strong>, a novel <strong>Hi</strong>erarchical <strong>P</strong>rediction model that enables accurate forecasting of <strong>IMTS</strong> across granularities, from coarse to fine, within a hierarchical framework. HiP-IMTS first extracts hierarchical patch embeddings at varying granularity levels. It then employs a finite difference-based attention mechanism for effectively addressing baseline drift, and a frequency convolution network for comprehensive temporal modeling from frequency domains. Next, a multi-granularity adaptive graph learning is introduced to model dynamic spatial correlations across different temporal granularities. Finally, a hierarchical prediction mechanism is designed to integrate complementary forecasting signals across multiple granularity levels, enabling effective fusion from coarse to fine scales. We perform a thorough assessment using four authentic datasets spanning various fields such as healthcare, biomechanics, and climate science, benchmarking HiP-IMTS against sixteen competitive baselines. HiP-IMTS achieves the best average critical difference ranks, with 1.1667 for MSE and 1.2083 for MAE, significantly outperforming existing state-of-the-art IMTS models.</div></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":"63 3","pages":"Article 104508"},"PeriodicalIF":6.9,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145693774","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号