US 12,271,680 B2
Text compression with predicted continuations
Ronny Lempel, Redmond, WA (US); and Chenyan Xiong, Bellevue, WA (US)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Oct. 17, 2023, as Appl. No. 18/488,275.
Application 18/488,275 is a continuation of application No. 17/351,531, filed on Jun. 18, 2021, granted, now 11,803,693.
Prior Publication US 2024/0046026 A1, Feb. 8, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. H03M 7/00 (2006.01); G06F 40/126 (2020.01); G06F 40/279 (2020.01); H03M 7/30 (2006.01)
CPC G06F 40/126 (2020.01) [G06F 40/279 (2020.01); H03M 7/3071 (2013.01); H03M 7/6011 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for text compression, comprising:
recognizing a prefix string of one or more text characters, the prefix string preceding a target string of a plurality of text characters to be compressed;
providing the prefix string to a natural language generation (NLG) model configured to output one or more predicted continuations to the prefix string; and
upon determining that the one or more predicted continuations include a matching predicted continuation relative to a next one or more text characters of the target string, compressing the next one or more text characters of the target string as a compressed representation based at least in part on the matching predicted continuation.