US 11,687,539 B2
Automatic neutral point of view content generation
Aaron K. Baughman, Cary, NC (US); Gray Franklin Cannon, Atlanta, GA (US); Stephen C Hammer, Marietta, GA (US); and Shikhar Kwatra, San Jose, CA (US)
Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed by International Business Machines Corporation, Armonk, NY (US)
Filed on Mar. 17, 2021, as Appl. No. 17/204,373.
Prior Publication US 2022/0300517 A1, Sep. 22, 2022
Int. Cl. G06F 17/00 (2019.01); G06F 7/00 (2006.01); G06F 16/2457 (2019.01); G06F 16/22 (2019.01); G06F 16/248 (2019.01); G06F 16/242 (2019.01)
CPC G06F 16/24578 (2019.01) [G06F 16/2246 (2019.01); G06F 16/243 (2019.01); G06F 16/248 (2019.01)] 25 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
constructing, from a set of natural language text documents, a concept tree, a node of the concept tree representing a subject of a subset of the set of natural language text documents (subset), a level of the concept tree corresponding to a level of specificity of a subject of a node in the level;
scoring, for a node in the concept tree (node) using a trained polarity scoring model, a polarity of the subset represented by the node;
adding a second set of natural language text documents to the subset, the adding resulting in a modified subset of natural language text documents, the modified subset having a polarity score within a predefined neutral polarity score range;
selecting, from the modified subset according to a sentence selection parameter, a bin of sentences, a sentence in the bin of sentences extracted from a selected document in the modified subset;
removing, from the bin of sentences, the removing resulting in a filtered bin of sentences, a sentence having a factuality score below a threshold factuality score; and
generating, from the filtered bin of sentences using a transformer deep learning narration generation model, a new natural language text document corresponding to the filtered bin of sentences.