Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection
The rapid spread of misinformation threatens public safety and social stability. Although deep learning-based detection methods have achieved promising results, their effectiveness heavily relies on large amounts of labeled data, limiting their applicability in low-resource scenarios. Existing appro...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-05-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/13/11/1752 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849722524283174912 |
|---|---|
| author | Luyao Hu Guangpu Han Shichang Liu Yuqing Ren Xu Wang Zhengyi Yang Feng Jiang |
| author_facet | Luyao Hu Guangpu Han Shichang Liu Yuqing Ren Xu Wang Zhengyi Yang Feng Jiang |
| author_sort | Luyao Hu |
| collection | DOAJ |
| description | The rapid spread of misinformation threatens public safety and social stability. Although deep learning-based detection methods have achieved promising results, their effectiveness heavily relies on large amounts of labeled data, limiting their applicability in low-resource scenarios. Existing approaches, such as domain adaptation and metalearning, attempt to transfer knowledge from related source domains but often fail to fully address the challenges of data scarcity and annotation costs. Moreover, traditional active learning strategies typically focus solely on textual uncertainty, overlooking domain-specific discrepancies and the critical role of affective information in misinformation content. To address these challenges, this paper proposes a dual-aspect active learning framework with domain-adversarial training (DDT), tailored for low-resource misinformation detection. The framework integrates a dual-aspect sampling strategy that jointly considers textual and affective features to select samples that are both informative (diverse from labeled data) and uncertain (near decision boundaries). Additionally, a domain-adversarial training module is employed to extract domain-invariant representations, mitigating distribution shifts between source and target domains. Experimental results on multiple benchmark datasets demonstrate that DDT consistently outperforms baseline methods in low-resource settings, enhancing the robustness and generalizability of misinformation detection models. |
| format | Article |
| id | doaj-art-790a2f3538d345e29f6dd419e2b852ab |
| institution | DOAJ |
| issn | 2227-7390 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj-art-790a2f3538d345e29f6dd419e2b852ab2025-08-20T03:11:19ZengMDPI AGMathematics2227-73902025-05-011311175210.3390/math13111752Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation DetectionLuyao Hu0Guangpu Han1Shichang Liu2Yuqing Ren3Xu Wang4Zhengyi Yang5Feng Jiang6Chongging Division, PetroChina Southwest Oil & Gasfield Company, Chongqing 400707, ChinaChongging Division, PetroChina Southwest Oil & Gasfield Company, Chongqing 400707, ChinaChongging Division, PetroChina Southwest Oil & Gasfield Company, Chongqing 400707, ChinaChongging Division, PetroChina Southwest Oil & Gasfield Company, Chongqing 400707, ChinaChongging Division, PetroChina Southwest Oil & Gasfield Company, Chongqing 400707, ChinaSchool of Big Data and Software Engineering, Chongqing University, Chongqing 401331, ChinaSchool of Big Data and Software Engineering, Chongqing University, Chongqing 401331, ChinaThe rapid spread of misinformation threatens public safety and social stability. Although deep learning-based detection methods have achieved promising results, their effectiveness heavily relies on large amounts of labeled data, limiting their applicability in low-resource scenarios. Existing approaches, such as domain adaptation and metalearning, attempt to transfer knowledge from related source domains but often fail to fully address the challenges of data scarcity and annotation costs. Moreover, traditional active learning strategies typically focus solely on textual uncertainty, overlooking domain-specific discrepancies and the critical role of affective information in misinformation content. To address these challenges, this paper proposes a dual-aspect active learning framework with domain-adversarial training (DDT), tailored for low-resource misinformation detection. The framework integrates a dual-aspect sampling strategy that jointly considers textual and affective features to select samples that are both informative (diverse from labeled data) and uncertain (near decision boundaries). Additionally, a domain-adversarial training module is employed to extract domain-invariant representations, mitigating distribution shifts between source and target domains. Experimental results on multiple benchmark datasets demonstrate that DDT consistently outperforms baseline methods in low-resource settings, enhancing the robustness and generalizability of misinformation detection models.https://www.mdpi.com/2227-7390/13/11/1752low-resource misinformation detectiondiscrepancy aspectuncertainty aspectadversarial trainingactive learning |
| spellingShingle | Luyao Hu Guangpu Han Shichang Liu Yuqing Ren Xu Wang Zhengyi Yang Feng Jiang Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection Mathematics low-resource misinformation detection discrepancy aspect uncertainty aspect adversarial training active learning |
| title | Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection |
| title_full | Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection |
| title_fullStr | Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection |
| title_full_unstemmed | Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection |
| title_short | Dual-Aspect Active Learning with Domain-Adversarial Training for Low-Resource Misinformation Detection |
| title_sort | dual aspect active learning with domain adversarial training for low resource misinformation detection |
| topic | low-resource misinformation detection discrepancy aspect uncertainty aspect adversarial training active learning |
| url | https://www.mdpi.com/2227-7390/13/11/1752 |
| work_keys_str_mv | AT luyaohu dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection AT guangpuhan dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection AT shichangliu dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection AT yuqingren dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection AT xuwang dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection AT zhengyiyang dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection AT fengjiang dualaspectactivelearningwithdomainadversarialtrainingforlowresourcemisinformationdetection |