US 11,727,221 B2
Dynamic correlated topic model
Praveen Chandar Ravichandran, New York, NY (US); Mounia Lalmas-Roelleke, Saffron Walden (GB); Federico Tomasi, London (GB); Zhenwen Dai, London (GB); and Gal Levy-Fix, Ithaca, NY (US)
Assigned to Spotify AB, Stockholm (SE)
Filed by Spotify AB, Stockholm (SE)
Filed on Jul. 17, 2020, as Appl. No. 16/932,323.
Prior Publication US 2022/0019750 A1, Jan. 20, 2022
Int. Cl. G06F 40/44 (2020.01); G06F 40/295 (2020.01); G06F 16/35 (2019.01); G06F 40/20 (2020.01); G06F 40/30 (2020.01); G06F 17/15 (2006.01)
CPC G06F 40/44 (2020.01) [G06F 16/355 (2019.01); G06F 17/15 (2013.01); G06F 40/20 (2020.01); G06F 40/295 (2020.01); G06F 40/30 (2020.01)] 22 Claims
OG exemplary drawing
 
1. A method of using machine learning for dynamically modeling topic correlation over time, the method comprising:
training a machine learning dynamic correlated topic model (DCTM) to jointly model evolution of a popularity of a topic across a period of time, a representation of the topic across the period of time, and a correlation with other topics across the period of time;
receiving a set of documents each comprised of a plurality of words and having associated timestamps, wherein the timestamps of the set of documents span a period of time;
identifying a quantity of topics for modeling;
providing the set of documents as input to the DCTM for modeling based on the quantity of topics identified; and
receiving, as output of the DCTM:
a list of topics, wherein clusters of the plurality of words represent the topics; and
for each topic:
a popularity of the topic across the period of time;
a representation of the topic across the period of time; and
a correlation with other topics across the period of time, wherein the correlation with other topics changes across the period of time.