US 11,657,076 B2
System for uniform structured summarization of customer chats
Priya Radhakrishnan, Bangalore (IN); and Shourya Roy, Bangalore (IN)
Assigned to American Express Travel Related Services Company, INC., New York, NY (US)
Filed by American Express Travel Related Services Company, Inc., New York, NY (US)
Filed on Mar. 18, 2021, as Appl. No. 17/205,415.
Claims priority of application No. 202011015288 (IN), filed on Apr. 7, 2020.
Prior Publication US 2021/0311973 A1, Oct. 7, 2021
Int. Cl. G06F 16/00 (2019.01); G06F 16/31 (2019.01); G06F 40/30 (2020.01); G06F 16/33 (2019.01); G06F 16/35 (2019.01); G06F 16/36 (2019.01); H04L 51/21 (2022.01)
CPC G06F 16/313 (2019.01) [G06F 16/3335 (2019.01); G06F 16/35 (2019.01); G06F 16/36 (2019.01); G06F 40/30 (2020.01); H04L 51/21 (2022.05)] 19 Claims
OG exemplary drawing
 
1. A method comprising;
receiving, by a computing device, a corpus of digital chat records;
grouping, by the computing device, the corpus of digital chat records into subgroups of tasks of a plurality of tasks, wherein each subgroup corresponds to one task of the plurality of tasks;
generating, by the computing device via a keyword chat module, for each subgroup of tasks, a plurality of task keywords and a plurality of task related words from the corpus of digital chat records based at least in part on a tokenization analysis of the corpus of digital chat records;
generating, by the computing device, for each subgroup of digital chat records, an expandable template data structure based at least in part on the plurality of task keywords and the plurality of task related words;
receiving, by the computing device, a digital chat record;
extracting, by the computing device, a plurality of chat utterances from chat content data of the received digital chat record based at least in part on the plurality of task keywords associated with the task and the plurality of task related words associated with the task;
extracting, by the computing device, a plurality of chat snippets from the plurality of chat utterances based at least in part on a data proximity window in relation to a task keyword;
ranking, by the computing device via a template scoring module, the plurality of chat utterances and the plurality of chat snippets based at least in part on the expandable template data structure for a respective subgroup of digital task records;
modifying, by the computing device, the expandable template data structure for the respective subgroup of digital task records to form an adjusted template data structure based at least in part on the plurality of ranked chat utterances and the plurality of ranked chat snippets;
populating, by the computing device, the adjusted template data structure with a subset of the ranked chat utterances from the plurality of ranked chat utterances and a subset of the ranked chat snippets from the plurality of the ranked chat snippets based at least in part on a ranking score to form a chat summary data structure for the received digital chat record; and
storing, by the computing device, the chat summary data structure for the received digital chat record in a database.