WaterSeeker: Pioneering Efficient Detection of Watermarked Segments in Large Documents
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917934829404160 |
|---|---|
| author | Pan, Leyi Liu, Aiwei Lu, Yijian Gao, Zitian Di, Yichen Huang, Shiyu Wen, Lijie King, Irwin Yu, Philip S. |
| author_facet | Pan, Leyi Liu, Aiwei Lu, Yijian Gao, Zitian Di, Yichen Huang, Shiyu Wen, Lijie King, Irwin Yu, Philip S. |
| contents | Watermarking algorithms for large language models (LLMs) have attained high accuracy in detecting LLM-generated text. However, existing methods primarily focus on distinguishing fully watermarked text from non-watermarked text, overlooking real-world scenarios where LLMs generate only small sections within large documents. In this scenario, balancing time complexity and detection performance poses significant challenges. This paper presents WaterSeeker, a novel approach to efficiently detect and locate watermarked segments amid extensive natural text. It first applies an efficient anomaly extraction method to preliminarily locate suspicious watermarked regions. Following this, it conducts a local traversal and performs full-text detection for more precise verification. Theoretical analysis and experimental results demonstrate that WaterSeeker achieves a superior balance between detection accuracy and computational efficiency. Moreover, its localization capability lays the foundation for building interpretable AI detection systems. Our code is available at https://github.com/THU-BPM/WaterSeeker. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_05112 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | WaterSeeker: Pioneering Efficient Detection of Watermarked Segments in Large Documents Pan, Leyi Liu, Aiwei Lu, Yijian Gao, Zitian Di, Yichen Huang, Shiyu Wen, Lijie King, Irwin Yu, Philip S. Computation and Language 68T50 I.2.7 Watermarking algorithms for large language models (LLMs) have attained high accuracy in detecting LLM-generated text. However, existing methods primarily focus on distinguishing fully watermarked text from non-watermarked text, overlooking real-world scenarios where LLMs generate only small sections within large documents. In this scenario, balancing time complexity and detection performance poses significant challenges. This paper presents WaterSeeker, a novel approach to efficiently detect and locate watermarked segments amid extensive natural text. It first applies an efficient anomaly extraction method to preliminarily locate suspicious watermarked regions. Following this, it conducts a local traversal and performs full-text detection for more precise verification. Theoretical analysis and experimental results demonstrate that WaterSeeker achieves a superior balance between detection accuracy and computational efficiency. Moreover, its localization capability lays the foundation for building interpretable AI detection systems. Our code is available at https://github.com/THU-BPM/WaterSeeker. |
| title | WaterSeeker: Pioneering Efficient Detection of Watermarked Segments in Large Documents |
| topic | Computation and Language 68T50 I.2.7 |
| url | https://arxiv.org/abs/2409.05112 |