Watermarking of structured results and watermark generation

摘要

A way of generating a watermark for a structured result, such as a search result or a machine translation. A hash function is used to generate a bit sequence for each of a plurality of structured results. A ranking score is generated for each resulting bit sequence. The ranking score can be based on the detectability of the bit sequence compared to a randomly-generated bit sequence and the quality of each of the structured results. A structured result is selected as the watermarked structured result based upon the ranking score.

主权项

1. A method, comprising:
receiving a plurality of alternative search results based upon a query from a user; computing a hash of at least part of each alternative search result to produce a bit sequence corresponding to each alternative search result; computing a first ranking score and a second ranking score for each alternative search result based upon its corresponding hash, the first ranking score for a specific alternative search result based upon a deviation of a distribution of binary digits in its corresponding hash from a binary distribution in which a parameter p=0.5 and the second ranking score for a specific alternative search result based upon a quality of the specific alternative search result; and selecting one of the plurality of alternative search results based upon the ranking score to obtain a watermarked search result, the watermarked search result having been determined to include machine-generated content that is not appropriate for machine learning by a language translation model.