Information Processing & Management最新文献_第8页

Does usage scenario matter? Investigating user perceptions, attitude and support for policies towards ChatGPT 使用场景重要吗？调查用户对 ChatGPT 的看法、态度和支持政策

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-31 DOI: 10.1016/j.ipm.2024.103867

Wenjia Yan , Bo Hu , Yu-li Liu , Changyan Li , Chuling Song

{"title":"Does usage scenario matter? Investigating user perceptions, attitude and support for policies towards ChatGPT","authors":"Wenjia Yan , Bo Hu , Yu-li Liu , Changyan Li , Chuling Song","doi":"10.1016/j.ipm.2024.103867","DOIUrl":"10.1016/j.ipm.2024.103867","url":null,"abstract":"<div><p>ChatGPT's impressive performance enables users to increasingly apply it to a variety of scenarios. However, previous studies investigating people's perceptions or attitudes towards ChatGPT have not considered the effects of the usage scenario. This paper aims to extract the representative scenarios of ChatGPT, explore differences in user perceptions for each scenario, and provide a policy support model. We extracted five scenarios by collecting 50 open-ended responses from Mturk, including “Scenario 1: Daily life tasks,” “Scenario 2: Enhance efficiency (work and education purposes),” “Scenario 3: Replace manpower (work and education purposes),” “Scenario 4: Browsing and general information seeking,” “Scenario 5: Enjoyment.” Subsequently, we identified four key variables to be tested (i.e., information quality, perceived risk, attitude, and policy support), and classified usage scenarios into different categories according to the perception variables measured via an online survey (<em>n</em> = 514). Finally, we built a model including the four variables and tested it for each scenario. The results of this study provide deep insights into user perceptions towards ChatGPT in distinct scenarios.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142094770","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}

引用次数: 0

Keywords-enhanced Contrastive Learning Model for travel recommendation 用于旅行推荐的关键词增强型对比学习模型

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-31 DOI: 10.1016/j.ipm.2024.103874

Lei Chen , Guixiang Zhu , Weichao Liang , Jie Cao , Yihan Chen

{"title":"Keywords-enhanced Contrastive Learning Model for travel recommendation","authors":"Lei Chen , Guixiang Zhu , Weichao Liang , Jie Cao , Yihan Chen","doi":"10.1016/j.ipm.2024.103874","DOIUrl":"10.1016/j.ipm.2024.103874","url":null,"abstract":"<div><p>Travel recommendation aims to infer travel intentions of users by analyzing their historical behaviors on Online Travel Agencies (OTAs). However, crucial keywords in clicked travel product titles, such as destination and itinerary duration, indicating tourists’ intentions, are often overlooked. Additionally, most previous studies only consider stable long-term user interests or temporary short-term user preferences, making the recommendation performance unreliable. To mitigate these constraints, this paper proposes a novel <strong>K</strong>eywords-enhanced <strong>C</strong>ontrastive <strong>L</strong>earning <strong>M</strong>odel (KCLM). KCLM simultaneously implements personalized travel recommendation and keywords generation tasks, integrating long-term and short-term user preferences within both tasks. Furthermore, we design two kinds of contrastive learning tasks for better user and travel product representation learning. The preference contrastive learning aims to bridge the gap between long-term and short-term user preferences. The multi-view contrastive learning focuses on modeling the coarse-grained commonality between clicked products and their keywords. Extensive experiments are conducted on two tourism datasets and a large-scale e-commerce dataset. The experimental results demonstrate that KCLM achieves substantial gains in both metrics compared to the best-performing baseline methods. Specifically, HR@20 improved by 5.79%–14.13%, MRR@20 improved by 6.57%–18.50%. Furthermore, to have an intuitive understanding of the keyword generation by the KCLM model, we provide a case study for several randomized examples.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142094771","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}

引用次数: 0

SelfCP: Compressing over-limit prompt via the frozen large language model itself SelfCP：通过冻结的大型语言模型本身压缩超限提示

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-30 DOI: 10.1016/j.ipm.2024.103873

Jun Gao , Ziqiang Cao , Wenjie Li

{"title":"SelfCP: Compressing over-limit prompt via the frozen large language model itself","authors":"Jun Gao , Ziqiang Cao , Wenjie Li","doi":"10.1016/j.ipm.2024.103873","DOIUrl":"10.1016/j.ipm.2024.103873","url":null,"abstract":"<div><p>Long prompt leads to huge hardware costs when using transformer-based Large Language Models (LLMs). Unfortunately, many tasks, such as summarization, inevitably introduce long documents, and the wide application of in-context learning easily makes the prompt length explode. This paper proposes a Self-Compressor (SelfCP), which adopts the target LLM itself to compress over-limit prompts into dense vectors on top of a sequence of learnable embeddings (<strong>memory tags</strong>) while keeping the allowed prompts unmodified. Dense vectors are then projected into <strong>memory tokens</strong> via a learnable connector, allowing the same LLM to understand them. The connector and the memory tag are supervised-tuned under the language modeling objective of the LLM on relatively long texts selected from publicly accessed datasets involving an instruction dataset to make SelfCP respond to various prompts, while the target LLM keeps frozen during training. We build the lightweight SelfCP upon 2 different backbones with merely 17M learnable parameters originating from the connector and a learnable embedding. Evaluation on both English and Chinese benchmarks demonstrate that SelfCP effectively substitutes 12<span><math><mo>×</mo></math></span> over-limit prompts with memory tokens to reduce memory costs and booster inference throughputs, yet improving response quality. The outstanding performance brings an efficient solution for LLMs to tackle long prompts without training LLMs from scratch.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142094769","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}

引用次数: 0

IDC-CDR: Cross-domain Recommendation based on Intent Disentanglement and Contrast Learning IDC-CDR：基于意图分离和对比学习的跨域推荐

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-29 DOI: 10.1016/j.ipm.2024.103871

Jing Xu, Mingxin Gan, Hang Zhang, Shuhao Zhang

{"title":"IDC-CDR: Cross-domain Recommendation based on Intent Disentanglement and Contrast Learning","authors":"Jing Xu, Mingxin Gan, Hang Zhang, Shuhao Zhang","doi":"10.1016/j.ipm.2024.103871","DOIUrl":"10.1016/j.ipm.2024.103871","url":null,"abstract":"<div><p>Using the user’s past activity across different domains, the cross-domain recommendation (CDR) predicts the items that users are likely to click. Most recent studies on CDR model user interests at the item level. However because items in other domains are inherently heterogeneous, direct modeling of past interactions from other domains to augment user representation in the target domain may limit the effectiveness of recommendation. Thus, in order to enhance the performance of cross-domain recommendation, we present a model called Cross-domain Recommendation based on Intent Disentanglement and Contrast Learning (IDC-CDR) that performs contrastive learning at the intent level between domains and disentangles user interaction intents in various domains. Initially, user–item interaction graphs were created for both single-domain and cross-domain scenarios. Then, by modeling the intention distribution of each user–item interaction, the interaction intention graph and its representation were updated repeatedly. The comprehensive local intent is then obtained by fusing the local domain intents of the source domain and the target domain using the attention technique. In order to enhance representation learning and knowledge transfer, we ultimately develop a cross-domain intention contrastive learning method. Using three pairs of cross-domain scenarios from Amazon and the KuaiRand dataset, we carry out comprehensive experiments. The experimental findings demonstrate that the recommendation performance can be greatly enhanced by IDC-CDR, with an average improvement of 20.62% and 25.32% for HR and NDCG metrics, respectively.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142088396","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}

引用次数: 0

Patent transformation prediction: When a patent can be transformed 专利转化预测：专利何时可以转化

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-27 DOI: 10.1016/j.ipm.2024.103872

Weidong Liu , Yu Zhang , Xiangfeng Luo , Yan Cao , Keqin Gan , Fuming Ye , Wei Tang , Minglong Zhang

{"title":"Patent transformation prediction: When a patent can be transformed","authors":"Weidong Liu , Yu Zhang , Xiangfeng Luo , Yan Cao , Keqin Gan , Fuming Ye , Wei Tang , Minglong Zhang","doi":"10.1016/j.ipm.2024.103872","DOIUrl":"10.1016/j.ipm.2024.103872","url":null,"abstract":"<div><p>Patent transformation is a pivotal pathway for realizing technological advancements, and patent transformation prediction is a potential strategy for improving the patent transformation rate. Existing automated patent transformation prediction models do not predict the transformation time, causing invalid conclusions for these valid patents. In this study, we propose a patent transformation prediction model to predict patent transformation time. (1) To obtain patent features in different time periods, the years elapsed since the patent application are segmented into multiple time slots; (2) For each patent, we extract static features and dynamic features of each time slot after constructing and embedding a dynamic graph of the patent; (3) The features for each time slot are concatenated as the input of the dynamic model which utilizes a neural network to predict the patent transformation of the time slot. We measure the model in diverse domains, each of which includes 10,000 patent transformation data. The experimental results show that precision, recall, and F1 scores are approximately 80% for predicting patent transformation in the next 3 years. Additionally, our study yields some novel findings: (1) later applied patents have a higher transformation speed; (2) over 90% of patent transformations occur within 13 years since the patent application; (3) dynamic features, especially dynamic structured features, have a significantly greater impact on patent transformation prediction compared to static features; (4) our model performs stably on different experiment data.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142083416","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}

引用次数: 0

Enhancing protection in high-dimensional data: Distributed differential privacy with feature selection 加强对高维数据的保护：带有特征选择的分布式差分隐私

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-27 DOI: 10.1016/j.ipm.2024.103870

I Made Putrama , Péter Martinek

{"title":"Enhancing protection in high-dimensional data: Distributed differential privacy with feature selection","authors":"I Made Putrama , Péter Martinek","doi":"10.1016/j.ipm.2024.103870","DOIUrl":"10.1016/j.ipm.2024.103870","url":null,"abstract":"<div><p>The computational cost for implementing data privacy protection tends to rise as the dimensions increase, especially on correlated datasets. For this reason, a faster data protection mechanism is needed to handle high-dimensional data while balancing utility and privacy. This study introduces an innovative framework to improve the performance by leveraging distributed computing strategies. The framework integrates specific feature selection algorithms and distributed mutual information computation, which is crucial for sensitivity assessment. Additionally, it is optimized using a hyperparameter tuning technique based on Bayesian optimization, which focuses on minimizing either a combined score of the Bayesian information criterion (BIC) and Akaike’s Information Criterion (AIC) or by minimizing the Maximal Information Coefficient (MIC) score individually. Extensive testing on 12 datasets with tens to thousands of features was conducted for classification and regression tasks. With our method, the sensitivity of the resulting data is lower than alternative approaches, requiring less perturbation for an equivalent level of privacy. Using a novel Privacy Deviation Coefficient (PDC) metric, we assess the performance disparity between original and perturbed data. Overall, there is a significant execution time improvement of 64.30% on the computation, providing valuable insights for practical applications.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142083417","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}

引用次数: 0

Multi-granularity attribute similarity model for user alignment across social platforms under pre-aligned data sparsity 预对齐数据稀疏性下跨社交平台用户对齐的多粒度属性相似性模型

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-23 DOI: 10.1016/j.ipm.2024.103866

Yongqiang Peng , Xiaoliang Chen , Duoqian Miao , Xiaolin Qin , Xu Gu , Peng Lu

{"title":"Multi-granularity attribute similarity model for user alignment across social platforms under pre-aligned data sparsity","authors":"Yongqiang Peng , Xiaoliang Chen , Duoqian Miao , Xiaolin Qin , Xu Gu , Peng Lu","doi":"10.1016/j.ipm.2024.103866","DOIUrl":"10.1016/j.ipm.2024.103866","url":null,"abstract":"<div><p>Cross-platform User Alignment (UA) aims to identify accounts belonging to the same individual across multiple social network platforms. This study seeks to enhance the performance of UA tasks while reducing the required sample data. Previous research has focused excessively on model design, lacking optimization throughout the entire process, making it challenging to achieve performance without heavy reliance on labeled data. This paper proposes a semi-supervised Multi-Granularity Attribute Similarity Model (MGASM). First, MGASM optimizes the embedding process through multi-granularity modeling at the levels of characters, words, articles, structures, and labels, and enhances missing data by leveraging adjacent text attributes. Next, MGASM quantifies the correlation between attributes of the same granularity by constructing Multi-Granularity Attribute Cosine Distance Distribution Vectors (MA-CDDVs). These vectors form the basis for a binary classification similarity model trained to calculate similarity scores for user pairs. Additionally, an attribute reappearance score correction (ARSC) mechanism is introduced to further refine the ranking of candidate users. Extensive experiments on the Weibo-Douban and DBLP17-DBLP19 datasets demonstrate that compared to state-of-the-art methods, The hit-precision of the MGASM series has significantly improved by 68.15% and 27.02%, almost reaching 100% precision. The F1 score has increased by 37.6% and 21.4%.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142044545","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}

引用次数: 0

FairColor: An efficient algorithm for the Balanced and Fair Reviewer Assignment Problem FairColor：平衡与公平审稿人分配问题的高效算法

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-22 DOI: 10.1016/j.ipm.2024.103865

Khadra Bouanane , Abdeldjaouad Nusayr Medakene , Abdellah Benbelghit , Samir Brahim Belhaouari

{"title":"FairColor: An efficient algorithm for the Balanced and Fair Reviewer Assignment Problem","authors":"Khadra Bouanane , Abdeldjaouad Nusayr Medakene , Abdellah Benbelghit , Samir Brahim Belhaouari","doi":"10.1016/j.ipm.2024.103865","DOIUrl":"10.1016/j.ipm.2024.103865","url":null,"abstract":"<div><p>As the volume of submitted papers continues to rise, ensuring a fair and accurate assignment of manuscripts to reviewers has become increasingly critical for academic conference organizers. Given the paper-reviewer similarity scores, this study introduces the Balanced and Fair Reviewer Assignment Problem (BFRAP), which aims to maximize the overall similarity score (efficiency) and the minimum paper score (fairness) subject to coverage, load balance, and fairness constraints. Addressing the challenges posed by these constraints, we conduct a theoretical investigation into the threshold conditions for the problem’s feasibility and optimality. To facilitate this investigation, we establish a connection between BFRAP, defined over <span><math><mi>m</mi></math></span> reviewers, and the Equitable m-Coloring Problem. Building on this theoretical foundation, we propose FairColor, an algorithm designed to retrieve fair and efficient assignments. We compare FairColor to Fairflow and FairIR, two state-of-the-art algorithms designed to find fair assignments under similar constraints. Empirical experiments were conducted on four real and two synthetic datasets involving (paper, reviewer) matching scores ranging from (100,100) to (10124,5880). Results demonstrate that FairColor is able to find efficient and fair assignments quickly compared to Fairflow and FairIR. Notably, in the largest instance involving 10,124 manuscripts and 5680 reviewers, FairColor retrieves fair and efficient assignments in just 67.64 s. This starkly contrasts both other methods, which require significantly longer computation times (45 min for Fairflow and 3 h 24 min for FairIR), even on more powerful machines. These results underscore FairColor as a promising alternative to current state-of-the-art assignment techniques.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142040406","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}

引用次数: 0

An adaptive approach to noisy annotations in scientific information extraction 科学信息提取中噪声注释的自适应方法

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-12 DOI: 10.1016/j.ipm.2024.103857

Necva Bölücü, Maciej Rybinski, Xiang Dai, Stephen Wan

{"title":"An adaptive approach to noisy annotations in scientific information extraction","authors":"Necva Bölücü, Maciej Rybinski, Xiang Dai, Stephen Wan","doi":"10.1016/j.ipm.2024.103857","DOIUrl":"10.1016/j.ipm.2024.103857","url":null,"abstract":"<div><p>Despite recent advances in large language models (LLMs), the best effectiveness in information extraction (IE) is still achieved by fine-tuned models, hence the need for manually annotated datasets to train them. However, collecting human annotations for IE, especially for scientific IE, where expert annotators are often required, is expensive and time-consuming. Another issue widely discussed in the IE community is noisy annotations. Mislabelled training samples can hamper the effectiveness of trained models. In this paper, we propose a solution to alleviate problems originating from the high cost and difficulty of the annotation process. Our method distinguishes <em>clean</em> training samples from <em>noisy</em> samples and then employs weighted weakly supervised learning (WWSL) to leverage noisy annotations. Evaluation of Named Entity Recognition (NER) and Relation Classification (RC) tasks in Scientific IE demonstrates the substantial impact of detecting clean samples. Experimental results highlight that our method, utilising clean and noisy samples with WWSL, outperforms the baseline RoBERTa on NER (＋4.28, ＋4.59, ＋29.27, and ＋5.21 gain for the ADE, SciERC, STEM-ECR, and WLPC datasets, respectively) and the RC (＋6.09 and ＋4.39 gain for the SciERC and WLPC datasets, respectively) tasks. Comprehensive analyses of our method reveal its advantages over state-of-the-art denoising baseline models in scientific NER. Moreover, the framework is general enough to be adapted to different NLP tasks or domains, which means it could be useful in the broader NLP community.</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S0306457324002164/pdfft?md5=fff788405d49af01c42a5d5a7a592f76&pid=1-s2.0-S0306457324002164-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141979785","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Robust and resource-efficient table-based fact verification through multi-aspect adversarial contrastive learning 通过多视角对抗性对比学习，实现基于表格的稳健且资源节约型事实验证

IF 7.4 1区管理学

Information Processing & Management Pub Date : 2024-08-12 DOI: 10.1016/j.ipm.2024.103853

Ruiheng Liu , Yu Zhang , Bailong Yang , Qi Shi , Luogeng Tian

{"title":"Robust and resource-efficient table-based fact verification through multi-aspect adversarial contrastive learning","authors":"Ruiheng Liu , Yu Zhang , Bailong Yang , Qi Shi , Luogeng Tian","doi":"10.1016/j.ipm.2024.103853","DOIUrl":"10.1016/j.ipm.2024.103853","url":null,"abstract":"<div><p>Table-based fact verification focuses on determining the truthfulness of statements by cross-referencing data in tables. This task is challenging due to the complex interactions inherent in table structures. To address this challenge, existing methods have devised a range of specialized models. Although these models demonstrate good performance, they still exhibit limited sensitivity to essential variations in information relevant to reasoning within both statements and tables, thus learning spurious patterns and leading to potentially unreliable outcomes. In this work, we propose a novel approach named <strong>M</strong>ulti-Aspect <strong>A</strong>dversarial <strong>Co</strong>ntrastive <strong>L</strong>earning (<span>Macol</span>), aimed at enhancing the accuracy and robustness of table-based fact verification systems under the premise of resource efficiency. Specifically, we first extract pivotal logical reasoning clues to construct positive and adversarial negative instances for contrastive learning. We then propose a new training paradigm that introduces a contrastive learning objective, encouraging the model to recognize noise invariance between original and positive instances while also distinguishing logical differences between original and negative instances. Extensive experimental results on three widely studied datasets TABFACT, INFOTABS and SEM-TAB-FACTS demonstrate that <span>Macol</span> achieves state-of-the-art performance on benchmarks across various backbone architectures, with accuracy improvements reaching up to 5.4%. Furthermore, <span>Macol</span> exhibits significant advantages in robustness and low-data resource scenarios, with improvements of up to 8.2% and 9.1%. It is worth noting that our method achieves comparable or even superior performance while being more resource-efficient compared to approaches that employ specific additional pre-training or utilize large language models (LLMs).</p></div>","PeriodicalId":50365,"journal":{"name":"Information Processing & Management","volume":null,"pages":null},"PeriodicalIF":7.4,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141979786","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}

引用次数: 0