Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology最新文献_第6页

Assessing the use of attention weights to interpret BERT-based stance classification 评估使用注意权重来解释基于bert的姿态分类

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3486622.3493966

Carlos Abel Córdova Sáenz, Karin Becker

{"title":"Assessing the use of attention weights to interpret BERT-based stance classification","authors":"Carlos Abel Córdova Sáenz, Karin Becker","doi":"10.1145/3486622.3493966","DOIUrl":"https://doi.org/10.1145/3486622.3493966","url":null,"abstract":"BERT models are currently state-of-the-art solutions for various tasks, including stance classification. However, these models are a black box for their users. Some proposals have leveraged the weights assigned by the internal attention mechanisms of these models for interpretability purposes. However, whether the attention weights help the interpretability of the model is still a matter of debate, with positions in favor and against. This work proposes an attention-based interpretability mechanism to identify the most influential words for stances predicted using BERT-based models. We target stances expressed in Twitter using the Portuguese language and assess the proposed mechanism using a case study regarding stances on COVID-19 vaccination in the Brazilian context. The interpretation mechanism traces tokens’ attentions back to words, assigning a newly proposed metric referred to as absolute word attention. Through this metric, we assess several aspects to determine if we can find important words for the classification and with meaning for the domain. We developed a broad experimental setting that involved three datasets with tweets in Brazilian Portuguese and three BERT models with support for this language. Our results are encouraging, as we were able to identify 52-82% of words with high absolute attention contributing positively to stance classification. The interpretability mechanism proved to be helpful to understand the influence of words in the classification, and they revealed intrinsic properties of the domain and representative arguments of the stances.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88477038","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}

引用次数: 1

Does environmental awareness play a role in EV adoption? A value-based adoption model analysis with SEM-ANN approach 环保意识在电动汽车的普及中起作用吗?基于价值的采用模型SEM-ANN分析

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3498851.3498992

Sohaib Mustafa, Wen Zhang, Rui Li

{"title":"Does environmental awareness play a role in EV adoption? A value-based adoption model analysis with SEM-ANN approach","authors":"Sohaib Mustafa, Wen Zhang, Rui Li","doi":"10.1145/3498851.3498992","DOIUrl":"https://doi.org/10.1145/3498851.3498992","url":null,"abstract":"Abstract. Global warming is a serious threat to humanity, and one of the main reasons is that the transport sector contributes to CO2 emissions. The auto industry's Introduction of electric-powered vehicles (EVs) is widely viewed as the primary solution. Although electric vehicles (EVs) have numerous benefits, acceptance rates vary widely from country to country, and the promised reductions in energy shortages and emissions have not materialized. We used the Value-based technology adoption model and added Environmental awareness (EA) as a new factor in this model to assess the role of EA in EV adoption intentions. Because of the complexity in consumer adoption decisions and nonlinearity in the dataset, we proposed the SEM-ANN mixed-method approach to address these issues. We gathered responses from 751 EV users in China and used the SEM-ANN dual-stage mix model to evaluate presented hypotheses and prioritize factors by relevance. According to the study's findings, although EA is a major predictor of EV adoption, it is not the most important. According to the study's findings, perceptions of the benefits and sacrifices associated with EV adoption have a considerable influence on the perceived value of EV, so this worth drives adoption intentions. Government and the electric car manufacturing business stand to gain much from this research, both theoretically and practically.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"83 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88393090","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}

引用次数: 16

How to Churn Deep Contextual Models? 如何搅动深度上下文模型?

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3486622.3493962

Mohammad Rashedul Hasan

{"title":"How to Churn Deep Contextual Models?","authors":"Mohammad Rashedul Hasan","doi":"10.1145/3486622.3493962","DOIUrl":"https://doi.org/10.1145/3486622.3493962","url":null,"abstract":"This paper searches for optimal ways of employing deep contextual models to solve practical natural language processing tasks. It addresses the diversity in the problem space by utilizing a variety of techniques that are based on the deep contextual BERT (Bidirectional Encoder Representation from Transformer) model. A collection of datasets on COVID-19 social media misinformation is used to capture the challenge in the misinformation detection task that arises from small labeled data, noisy labels, out-of-distribution (OOD) data, fine-grained & nuanced categories, and heavily-skewed class distribution. To address this diversity, both domain-agnostic (DA) and domain-specific (DS) BERT pretrained models (PTMs) for transfer learning are examined via two methods, i.e., fine-tuning (FT) and extracted feature-based (FB) learning. The FB is implemented using two approaches: non-hierarchical (features extracted from a single hidden layer) and hierarchical (features extracted from a subset of hidden layers are first aggregated, then passed to a neural network for further extraction). Results obtained from an extensive set of experiments show that FB is more effective than FT and that hierarchical FB is more generalizable. However, on the OOD data, the deep contextual models are less generalizable. It identifies the condition under which DS PTM is beneficial. Finally, bigger models may only add an incremental benefit and sometimes degrade the performance.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"23 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75562554","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}

引用次数: 0

Session details: Chapter 7: SMA4H: Social Media Analytics for Health Intelligence (SMA4H): How Artificial Intelligence Transforms Healthcare 会议详情:第7章:SMA4H:健康智能的社交媒体分析(SMA4H):人工智能如何改变医疗保健

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3530280

引用次数: 0

Time-aware Service Recommendation Algorithm Based on K-means and Low-rank Matrix Factorization 基于k均值和低秩矩阵分解的时间感知服务推荐算法

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3498851.3498988

Yimu Ji, F. Tian, Shangdong Liu, Si-si Shao, Kui Li, Qiang Liu, Zhao-yu Meng

{"title":"Time-aware Service Recommendation Algorithm Based on K-means and Low-rank Matrix Factorization","authors":"Yimu Ji, F. Tian, Shangdong Liu, Si-si Shao, Kui Li, Qiang Liu, Zhao-yu Meng","doi":"10.1145/3498851.3498988","DOIUrl":"https://doi.org/10.1145/3498851.3498988","url":null,"abstract":"∗ In recent years, with the emergence of more and more similar Web services, how to recommend high-quality Web services to consumers has become a challenging task. Traditional service recommendation has the problems of cold start and data sparseness, resulting in low accuracy of service recommendation through similar users of the target user. In response to the above problems, this paper proposes a time-aware service recommendation algorithm based on K-means and low-rank matrix factorization (K-LMF). First, this paper uses the K-means clustering algorithm to gather data with similar characteristics into the same cluster to improve the effi-ciency and accuracy of service recommendation. In order to prevent data sparseness from causing unpredictability, this paper divides the data from multiple dimensions, and then uses the low-rank matrix decomposition algorithm of the L1 paradigm to complete the quality of service (QoS) matrix to predict the appropriate service for the target user. Finally, experiments are carried out on the WS-DREAM data set to verify the effectiveness and feasibility of this scheme.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"10 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78892785","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}

引用次数: 0

TTNet: Tabular Transfer Network for Few-samples Prediction 用于少样本预测的表格传输网络

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3486622.3493940

Zhao Li, Donghui Ding, Xuanwu Liu, Peng Zhang, Youxi Wu, Lingzhou Ma

{"title":"TTNet: Tabular Transfer Network for Few-samples Prediction","authors":"Zhao Li, Donghui Ding, Xuanwu Liu, Peng Zhang, Youxi Wu, Lingzhou Ma","doi":"10.1145/3486622.3493940","DOIUrl":"https://doi.org/10.1145/3486622.3493940","url":null,"abstract":"Tabular learning has been widely used in practical scenarios to handle tabular data such as the type of data in a spreadsheet or a CSV file. In many applications, it is necessary to transfer knowledge from the abundant source tabular data to the few target data, since models on the few target data are always easy to overfit. However, existing tabular learning methods have not touched the problem of transfer learning between different tabular datasets based on deep neural networks. To this end, we propose in this paper a new Tabular Transfer Network (TTNet for short) to enable effective and efficient knowledge transferring on tabular data. To enable network structures adaptive for transferring learning, TTNet first designs a BucTab network which integrates multi-bucket network with imitated tree-based network structures. In addition, TTNet uses a meta-transfer learning strategy to enable fast adaption by pre-training in the source domain of tabular data. Experiments on benchmark and real-world datasets show that TTNet significantly outperforms the state-of-the-art approaches in terms of effectiveness and efficiency, e.g., a new-shop prediction task testing on the e-commerce platform Ele.me shows that TTNet brings 9% MAE reduction comparing with the baselines.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"103 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80694556","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}

引用次数: 1

Comparing Decentralized Algorithms for Dynamic Task Sharing among Agents with Limited Resources 有限资源agent间动态任务共享的分散算法比较

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3486622.3493924

Hisashi Hayashi

引用次数: 0

Analyzing Features of Passive Twitter’s Users to Estimate Passive Twitter-User’s Interests 分析被动推特用户特征，估算被动推特用户兴趣

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3486622.3493979

Tessai Hayama

{"title":"Analyzing Features of Passive Twitter’s Users to Estimate Passive Twitter-User’s Interests","authors":"Tessai Hayama","doi":"10.1145/3486622.3493979","DOIUrl":"https://doi.org/10.1145/3486622.3493979","url":null,"abstract":"User modeling based on the contents of social network services has been developed to recommend information related to each user’s preference. Most of the previous studies have analyzed active users’ tweets and estimated their interests. Meanwhile, although more than a certain number of passive users do not tweet but only gather information, little research has been conducted on interest estimation due to the lack of clues for estimating their interests. These studies have achieved the estimation method using cues other than users’ tweets without understanding the behavior of passive Twitter users. Therefore, in this study, I analyzed the Twitter data with the user features used in the previous studies by using statistical methods to clarify the clue for extracting the interest of the passive user. To do so, a dataset including features of Twitter passive user and the active user was generated. The features of the passive user were clarified by statistical methods, such as Support Vector Machine, Principal Component Analysis, and Decision Tree Analysis. The results showed that it was possible to identify the passive user with an accuracy of 0.93 using features regarding user profiles, followers, and followed users. It was also found that most passive users had fewer than 8 followers and tended to be friendly connected to celebrities without self-disclosure. The results of this study identified types of Twitter passive users using the features. It contributes to the development of an interest estimation for the targeted types of a passive user.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"6 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84739887","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}

引用次数: 0

STANN: Spatio-Temporal Attention-based Neural Network for Epidemic Prediction 基于时空注意力的流行病预测神经网络

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3498851.3498972

Zhuanghu Lv, Jing Li, Dafeng Liu, Yue Peng, B. Shi

{"title":"STANN: Spatio-Temporal Attention-based Neural Network for Epidemic Prediction","authors":"Zhuanghu Lv, Jing Li, Dafeng Liu, Yue Peng, B. Shi","doi":"10.1145/3498851.3498972","DOIUrl":"https://doi.org/10.1145/3498851.3498972","url":null,"abstract":"Accurately forecasting the future trend of an epidemic plays an essential role in making effective and efficient public health policies for disease prevention and control. In reality, the local transmission of an epidemic depends not only on the cumulative number of infections at the same location, but also on the geographical spread of the disease from nearby locations. Therefore, the epidemic data usually show high nonlinearity and certain spatio-temporal patterns. Most existing methods lack the ability to simultaneously characterize the dynamic spatio-temporal patterns, thus cannot make satisfactory prediction results. In this paper, we propose a spatio-temporal attention-based neural network (STANN) to solve the epidemic prediction problem, where attention mechanisms are adopted to effectively capture the dynamic correlations of epidemic data in both spatial and temporal dimensions. The architecture of the network consists of three modules: a temporal attention module, a spatial attention module, and a temporal convolution module. Experimental results on the epidemic prediction of malaria cases in Yunnan Province, China, demonstrate that the STANN model outperforms the state-of-the-art baselines.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"67 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86892273","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}

引用次数: 0

Applying Benford's Law as an Efficient and Low-cost Solution for Verifying the Authenticity of Users’ Video Streams in Learning Management Systems 应用本福德定律作为学习管理系统中用户视频流真实性验证的高效低成本解决方案

Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology Pub Date : 2021-12-14 DOI: 10.1145/3486622.3493993

Argyris Constantinides, Christodoulos Constantinides, Marios Belk, C. Fidas, A. Pitsillides

{"title":"Applying Benford's Law as an Efficient and Low-cost Solution for Verifying the Authenticity of Users’ Video Streams in Learning Management Systems","authors":"Argyris Constantinides, Christodoulos Constantinides, Marios Belk, C. Fidas, A. Pitsillides","doi":"10.1145/3486622.3493993","DOIUrl":"https://doi.org/10.1145/3486622.3493993","url":null,"abstract":"An important challenge of online learning management systems (LMS) relates to continuously verifying the identity of students even after they have successfully authenticated. Although various continuous user identification solutions exist, they are rather focused on complex examination proctoring systems. Challenges further increase within large-scale online courses, which require a strong infrastructure to support numerous real-time video streams for verifying the identity of students. Considering that the students’ input video stream is an important factor for verifying their identity, and given that naturally generated data streams have been found to adhere to a pre-defined behavior as indicated by the Benford's law, in this work we investigate whether Benford's law can be applied as a reliable, efficient and cost-effective method for the detection of authentic vs. pre-recorded input video streams during continuous students’ identity verification within online LMS. In doing so, we suggest a prediction model based on the distribution of the first digits of image Discrete Cosine Transform (DCT) coefficients from the students’ input video stream. We found that the input video stream type (authentic vs. pre-recorded) can be inferred within a few seconds in real-time. A system performance evaluation indicates that the suggested model can support up to 1000 concurrent online students using a conventional and low-cost server setup and architecture.","PeriodicalId":89230,"journal":{"name":"Proceedings. IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology","volume":"27 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82515078","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}

引用次数: 2