US 12,326,972 B2
Saccade driven text comprehension artificial intelligence engine
Tyler Waite, Lake Washington, WA (US); Jason Andrew Telner, Lackawanna, NY (US); Jon Gerard Temple, Southbury, CT (US); Peter Guzewich, Vestal, NY (US); Donna Thelander, Kailua Kona, HI (US); and Mark Willard Malatesta, Ortonville, MI (US)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed on Mar. 9, 2023, as Appl. No. 18/181,299.
Prior Publication US 2024/0302895 A1, Sep. 12, 2024
Int. Cl. G06F 3/01 (2006.01); G06F 3/0481 (2022.01); G06F 3/04842 (2022.01); G06F 9/451 (2018.01); G06N 20/00 (2019.01)
CPC G06F 3/013 (2013.01) [G06F 3/0481 (2013.01); G06F 9/453 (2018.02); G06N 20/00 (2019.01); G06F 3/04842 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method for providing additional information to a user relating to a content, using user's eye movement data, comprising:
collecting, from at least one sensor, eye gazing data relating to an eye movement of a user looking at a content;
using an Artificial Intelligence (AI) module to calculate a saccade rate associated with said eye gazing data; wherein said Artificial Intelligence (AI) module uses one or more machine learning models;
using said AI module to determine a saccade pattern using said saccade rate and an associated saccade pattern data;
determining when said saccade rate falls below a first value with respect to viewing a first area of said content and flagging a location for a first subset of said content;
determining when said saccade rate shifts backwards over a second subset and second flagged area of said content one or more times based on said saccade pattern data, and flagging said location as a second area of the content; wherein a shift backward is a movement that gazes at least a second time over a set of words or images;
emphasizing a portion of said content located between said first and said second flagged areas;
determining one or more tooltips that can be provided to said user corresponding to information in said content that falls in said emphasized portion;
determining when a saccade delay is detected and moving across a portion location where the delay is detected a plurality of times using the one or more machine learning models to highlight the portion that is slowing the saccade using the one or more tooltips; and
determining a difficulty of the one or more sentences based on the saccade pattern data and providing alternatives using said one or more tooltips.