WatME: Towards Lossless Watermarking Through Lexical Redundancy

arXiv:2311.09832v2 Announce Type: replace
Abstract: Text watermarking has emerged as an important technique for detecting machine-generated text. However, existing methods generally use arbitrary vocabulary partitioning during decoding, which results in the absence of appropriate words during the response generation and disrupts the language model's expressiveness, thus severely degrading the quality of text response. To address these issues, we introduce a novel approach, Watermarking with Mutual Exclusion (WatME). Specifically, by leveraging linguistic prior knowledge of inherent lexical redundancy, WatME can dynamically …

abstract arxiv cs.cl decoding generated language language model machine partitioning quality redundancy text through type watermarking words