US 11,720,590 B2
Personalized visualization recommendation system
Ryan Rossi, Santa Clara, CA (US); Vasanthi Holtcamp, Fremont, CA (US); Tak Yeon Lee, Cupertino, CA (US); Sungchul Kim, San Jose, CA (US); Sana Lee, Brea, CA (US); Nathan Ross, Highland, UT (US); John Anderson, American Fork, UT (US); Fan Du, Milpitas, CA (US); Eunyee Koh, San Jose, CA (US); and Xin Qian, Greenbelt, MD (US)
Assigned to ADOBE INC., San Jose, CA (US)
Filed by ADOBE INC., San Jose, CA (US)
Filed on Nov. 6, 2020, as Appl. No. 17/91,941.
Prior Publication US 2022/0147540 A1, May 12, 2022
Int. Cl. G06F 16/9535 (2019.01); G06F 16/26 (2019.01); G06F 3/0482 (2013.01); G06F 11/34 (2006.01); G06F 11/30 (2006.01); G06F 16/9038 (2019.01)
CPC G06F 16/26 (2019.01) [G06F 3/0482 (2013.01); G06F 11/302 (2013.01); G06F 11/3438 (2013.01); G06F 16/9038 (2019.01)] 19 Claims
OG exemplary drawing
 
1. A method for data visualization, comprising:
generating a user visualization interaction matrix by identifying a plurality of data attributes of a plurality of datasets and user interactions with a plurality of visualizations for the plurality of datasets;
generating a meta-feature matrix by mapping the plurality of data attributes to a plurality of meta-features;
generating a user data interaction matrix by identifying user characteristics of a user and data attributes of at least one dataset;
performing joint factorization of the user visualization interaction matrix, the meta-feature matrix, and the user data interaction matrix based on the user interactions and the mapping to obtain a plurality of joint factorization matrices comprising low-dimensional embeddings of the user characteristics of the user and the data attributes of the at least one dataset, wherein a first joint factorization matrix of the plurality of joint factorization matrices represents a first factor of the user data interaction matrix and a factor of the user visualization interaction matrix, a second joint factorization matrix of the plurality of joint factorization matrices represents a second factor of the user data interaction matrix and a factor of the meta-feature matrix;
predicting visualization preference weights corresponding to a plurality of candidate visualizations of the at least one dataset for the user using a model based on the plurality of joint factorization matrices; and
generating a personalized visualization recommendation for the at least one dataset for the user based on the predicted visualization preference weights.