US 12,248,523 B1
Frustration scoring system
Mengzhu Liu, Paris (FR); Mohammad Reza Loghmani, Paris (FR); and Philipe Moura, Paris (FR)
Assigned to Content Square SAS, Paris (FR)
Filed by Content Square SAS, Paris (FR)
Filed on Jan. 30, 2024, as Appl. No. 18/427,170.
Int. Cl. G06F 16/00 (2019.01); G06F 16/95 (2019.01); H04L 67/50 (2022.01)
CPC G06F 16/95 (2019.01) [H04L 67/535 (2022.05); G06F 2216/03 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
accessing a first plurality of sessions within a predefined time period, each session comprising one or more actions performed by a given user on one or more webpages of a website;
extracting a subset of sessions from the first plurality of sessions that each comprise a user feedback score at a predefined point within a session;
generating a label for each session of the subset of sessions by determining where the user feedback score for each session falls within a score range, the label indicating that a user associated with each session was frustrated or not frustrated;
generating training data comprising the subset of sessions and the generated label for each session indicating that the user associated with each session was frustrated or not frustrated;
training, using the training data comprising the subset of sessions and the generated label for each session indicating that the user associated with each session was frustrated or not frustrated, a first machine learning model to generate an initial frustration score based on features derived for each session of a given set of sessions;
for each session of a second plurality of sessions that each comprise one or more actions performed by a given user on one or more webpages of a website, deriving values for a set of features, each value indicating whether or not a respective feature occurred in the session;
generating an initial frustration score for each session of the second plurality of sessions by analyzing the derived values for the set of features for each session of the second plurality of sessions using a first machine learning model trained to generate an initial frustration score based on values derived for a set of features for each session of a given set of sessions;
adding the initial frustration score to the set of features as an additional feature to generate an updated set of features for each session including the initial frustration score; and
generating a final frustration score for each session of the second plurality of sessions by analyzing the updated set of features using a second machine learning model trained to generate a final frustration score based on the initial frustration score and features derived for each session of a given set of sessions.