Back to results
US 11,386,268 B2Granted
Discriminating ambiguous expressions to enhance user experience
AssigneeMicrosoft Technology Licensing, LLC
InventorsJean-Philippe Robichaud, Ruhi Sarikaya
Life of this patent
Estimating the term…
Abstract
Methods and systems are provided for discriminating ambiguous expressions to enhance user experience. For example, a natural language expression may be received by a speech recognition component. The natural language expression may include at least one of words, terms, and phrases of text. A dialog hypothesis set from the natural language expression may be created by using contextual information. In some cases, the dialog hypothesis set has at least two dialog hypotheses. A plurality of dialog responses may be generated for the dialog hypothesis set. The dialog hypothesis set may be ranked based on an analysis of the plurality of the dialog responses. An action may be performed based on ranking the dialog hypothesis set.
Claims18
A system comprising:
- at least one processor;
- and
- memory encoding computer executable instructions that, when executed by at least one processor, perform a method comprising:
- receiving a natural language expression during a multi-turn session, wherein the natural language expression includes at least one of words, terms, and phrases of text;
- generating, by executing a machine learning technique, a first dialog hypothesis and a second dialog hypothesis from the natural language expression by using contextual information, wherein the contextual information is extracted from a previous turn in the multi-turn session;
- sending the first dialog hypothesis to a backend component;
- receiving a dialog response for the first dialog hypothesis from the backend component;
- determining a domain component to send the second dialog hypothesis based on a predicted domain for the second dialog hypothesis;
- receiving a dialog response for the second dialog hypothesis from the determined domain component;
- ranking the first dialog hypothesis and the second dialog hypothesis based on an analysis of the plurality of the dialog responses;
- and
- based on the ranking of the first dialog hypothesis and the second dialog hypothesis, executing an action to provide a response to the natural language expression in the multi-turn session, wherein executing the action to provide a response to the natural language expression includes:
- using a highest ranked dialog hypothesis of the first dialog hypothesis or the second dialog hypothesis to query the backend component for results, and
- sending the results to a computing device.
11 more elements
A method comprising:
- receiving a natural language expression, wherein the natural language expression includes at least one of words, terms, and phrases of text;
- generating, by executing a machine learning technique, a first dialog hypothesis and a second dialog hypothesis from the natural language expression by using contextual information, wherein the first dialog hypothesis and the second dialog hypothesis include different hypotheses of actions to perform based on the natural language expression;
- determining a backend component to send the first dialog hypothesis;
- receiving a dialog response for the first dialog hypothesis from the determined backend component;
- determining a domain component to send the second dialog hypothesis based on a predicted domain associated with the second dialog hypothesis;
- receiving a dialog response for the second dialog hypothesis from the determined domain component;
- extracting first features from the first dialog hypothesis and the dialog response for the first dialog hypothesis;
- extracting second features from the second dialog hypothesis and the dialog response for the second dialog hypothesis;
- ranking the first dialog hypothesis and the second dialog hypothesis based on an analysis of the extracted first features and extracted second features;
- and
- based on the ranking of the first dialog hypothesis and the second dialog hypothesis, executing an action to provide a response to the natural language expression,
- wherein executing the action to provide a response to the natural language expression includes:
- using a highest ranked dialog hypothesis of the first dialog hypothesis or the second dialog hypothesis to query a backend component for results, and
- sending the results to a computing device.
11 more elements
One or more computer-readable storage media, having computer-executable instructions that, when executed by at least one processor, perform a method for discriminating ambiguous requests, the method comprising:
- creating, by executing a machine learning technique, a dialog hypothesis set from a natural language expression, received during a multi-turn session, by using contextual information extracted from a previous turn in the multi-turn session, wherein the dialog hypothesis set has at least two dialog hypotheses;
- generating a plurality of dialog responses for the dialog hypothesis set;
- comparing the plurality of dialog responses with a plurality of logged dialog responses;
- determining whether at least one of the plurality of dialog responses matches at least one of the logged dialog responses;
- when it is determined that at least one of the plurality of dialog responses matches at least one of the logged dialog responses, labeling at least one of the two dialog hypotheses in the dialog hypothesis set corresponding to the at least one dialog response that matches the at least one logged dialog response;
- and
- performing an action based on the labeled at least one of the two dialog hypotheses to provide a response to the natural language expression, wherein performing an action based on the labeled at least one of the two dialog hypotheses includes:
- using the labeled at least one of the two dialog hypotheses to query a backend component for results, and
- sending the results to a computing device.
6 more elements
Dependencies are inferred from the claim text.