US 11,983,638 B2
Example-driven machine learning scheme for dialog system engines
Ilya Gennadyevich Gelfenbeyn, Sunnyvale, CA (US); Artem Goncharuk, Arlington, VA (US); and Pavel Aleksandrovich Sirotin, Sunnyvale, CA (US)
Assigned to GOOGLE LLC, Mountain View, CA (US)
Filed by GOOGLE LLC, Mountain View, CA (US)
Filed on Oct. 12, 2021, as Appl. No. 17/499,616.
Application 17/499,616 is a continuation of application No. 15/062,151, filed on Mar. 6, 2016, abandoned.
Claims priority of provisional application 62/129,759, filed on Mar. 6, 2015.
Prior Publication US 2022/0036015 A1, Feb. 3, 2022
Int. Cl. G06N 5/022 (2023.01); G06F 40/30 (2020.01)
CPC G06N 5/022 (2013.01) [G06F 40/30 (2020.01)] 16 Claims
OG exemplary drawing
 
1. A method implemented by a server, comprising:
receiving, from a developer via a developer interface, and during creation of a given dialog system rule of a dialog system engine, at least a portion of the given dialog system rule,
wherein the given dialog system rule is one of a plurality of dialog system rules that indicate instructions for response of the dialog system engine to future user requests;
receiving, from the developer via the developer interface, multiple exemplary phrases defining an intent for the given dialog system rule,
wherein each exemplary phrase, of the multiple exemplary phrases defining the intent for the given dialog system rule, include at least one or more entities;
in response to receiving a given exemplary phrase of the multiple exemplary phrases defining the intent for the given dialog system rule:
causing the given exemplary phrase to be provided for review by the developer via the developer interface, including:
causing the given exemplary phrase to be presented in the developer interface, and
causing a given linguistic element, that is present in the given exemplary phrase and that is a subset of the given exemplary phrase, to be presented along with a remainder of the given exemplary phrase and with an indication that the given linguistic element is assigned to a first entity type and to be presented with an indication of a plurality of alternative entity types;
receiving, in response to causing the given linguistic element to be presented with the indication that the given linguistic element is assigned to the first entity type and to be presented with the indication of alternative entity types, a selection of a second entity type, the second entity type being one of the alternative entity types; and
in response to receiving the selection, causing the assignment of the given linguistic element to the first entity type to be changed to an alternative assignment of the given linguistic element to the second entity type;
associating, in one or more databases, the multiple exemplary phrases, the one or more entities, and the intent with the given dialog system rule to create the given dialog system rule, including associating the given linguistic element with the second entity type based on the alternative assignment; and
subsequent to creation of the given dialog system rule:
receiving, from the developer via the developer interface, a test phrase to test the given dialog system rule;
processing, using the given dialog system rule, the test phrase to identify a particular intent and one or more particular entities associated with the test phrase;
causing the particular intent and the one or more particular entities associated with the test phrase to be provided for review by the developer via the developer interface;
receiving, from the developer via the developer interface, a modification to the intent for the given dialog system rule, wherein the modification to the intent is received responsive to the particular intent and the one or more particular entities associated with the test phrase being provided for review by the developer; and
modifying, in one or more of the databases, and based on the received modification to the intent, the given dialog system rule.