Identifying underserved command inputs

• US 9,177,553 B1
• Filed: 06/25/2013
• Issued: 11/03/2015
• Est. Priority Date: 06/25/2013
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Uszkoreit, Jakob D., Liang, Percy, Bikel, Daniel M.
• Primary Examiner(s): Neway, Samuel G
• Application Number:
  US13/926,943
• Time in Patent Office: 861 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 40/205 Parsing
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command

Abstract

A language processing system identifies first command input sentences that do not successfully parse by any parsing rule in a set of parsing rules. Each of the parsing rules is associated with an action, and a user device performs the action associated with a parsing rule in response to an input sentence being successfully parsed by the parsing rule. For each of these identified first sentences, the system determines whether the first input sentence has an underserving signal that is indicative of one or more actions being underserved. If the first sentence has the underserving signal, then the first sentence is selected as a candidate input sentence. Each candidate input sentence is provided to an action analysis processes that determines whether a candidate input sentence is to be associated with one action, and upon a positive determination generates a parsing rule for the candidate input sentence.

Claims

1. A computer-implemented method performed by a data processing apparatus, the method comprising:

accessing command input logs storing data defining received input sentences, each input sentence being one or more n-grams;

identifying from the input sentences first input sentences that do not successfully parse by any parsing rule in a set of parsing rules, each of the parsing rules being associated with an action, and wherein each parsing rule causes a user device to perform the action associated with the parsing rule in response to an input sentence received from the user device being successfully parsed by the parsing rule;

for each first input sentence, determining whether the first input sentence has an underserving signal, the underserving signal being a signal that is indicative of one or more actions being underserved; and

selecting first input sentences determined to have an underserving signal as candidate input sentences;

providing each candidate input sentence to an action analysis process that determines whether a candidate input sentence is to be associated with one of a plurality of actions and upon a positive determination generates a parsing rule for the candidate input sentence, wherein the generated parsing rule is associated with the one of the plurality of actions and a successful parse of an input sentence received at a user device by the generated parsing rule causes the user device to perform the one of the plurality of actions associated with the generated parsing rule; and

persisting each generated parsing rule and data describing the action with which the generated parsing rule is associated to command models data that is accessible by a language processing system that parses command input sentences by the generated parsing rules and determines actions to perform based on a successful parses.

2. The computer-implemented method of claim 1, wherein determining whether the first input sentence has an underserving signal comprises, for each first input sentence:

selecting n-grams included in the first input sentence;

for each selected n-gram, determining whether the n-gram is semantically relevant for an action that has an associated parsing rule; and

determining the first input sentence has the underserving signal when the n-gram is determined to be semantically relevant for an action that has an associated parsing rule.

3. The computer-implemented method of claim 2, wherein determining whether the first input sentence has an underserving signal further comprises, for each first input sentence:

determining that the first input sentence was processed as a search query;

determining a quality metric value for search results provided in response to the search query; and

determining the first input sentence has the underserving signal when the quality metric value does not meet a threshold quality value.

4. The computer-implemented method of claim 3, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of quality of resources identified by the search results, the measure of quality being independent of user interactions at the user device from which the input sentence was received.

5. The computer-implemented method of claim 3, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of user satisfaction in response to the search results, the measure of user satisfaction being dependent on user interactions with the search results at the user device from which the input sentence was received.

6. The computer-implemented method of claim 1, wherein determining whether the first input sentence has an underserving signal further comprises, for each first input sentence:

determining that the first input sentence was processed as a search query;

determining a quality metric value for search results provided in response to the search query; and

determining the first input sentence has the underserving signal when the quality metric value does not meet a threshold quality value.

7. The computer-implemented method of claim 6, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of quality of resources identified by the search results, the measure of quality being independent of user interactions at the user device from which the first input sentence was received.

8. The computer-implemented method of claim 6, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of user satisfaction in response to the search results, the measure of user satisfaction being dependent on user interactions with the search results at the user device from which the first input sentence was received.

9. The computer-implemented method of claim 1, wherein the input sentences are generated from voice command inputs.

10. The computer-implemented method of claim 1, wherein the parsing rules are grammar based rules.

11. A non-transitory computer readable storage medium storing instructions executable by a data processing apparatus and that upon such execution causes the data processing apparatus to perform operations comprising:

identifying from the input sentences first input sentences that do not successfully parse by any parsing rule in a set of parsing rules, each of the parsing rules being associated with an action, and wherein each parsing rule causes a user device to perform the action associated with the parsing rule in response to an input sentence received from the user device being successfully parsed by the parsing rule;

for each first input sentence, determining whether the first input sentence has an underserving signal, the underserving signal being a signal that is indicative of one or more actions being underserved; and

selecting first input sentences determined to have an underserving signal as candidate input sentences;

providing each candidate input sentence to an action analysis process that determines whether a candidate input sentence is to be associated with one of a plurality of actions and upon a positive determination generates a parsing rule for the candidate input sentence, wherein the generated parsing rule is associated with the one of the plurality of actions and a successful parse of an input sentence received at a user device by the generated parsing rule causes the user device to perform the one of the plurality of actions associated with the generated parsing rule; and

persisting each generated parsing rule and data describing the action with which the generated parsing rule is associated to command models data that is accessible by a language processing system that parses command input sentences by the generated parsing rules and determines actions to perform based on a successful parses.

12. The non-transitory computer readable storage medium of claim 11, wherein determining whether the first input sentence has an underserving signal comprises, for each first input sentence:

selecting n-grams included in the first input sentence;

for each selected n-gram, determining whether the n-gram is semantically relevant for an action that has an associated parsing rule; and

determining the first input sentence has the underserving signal when the n-gram is determined to be semantically relevant for an action that has an associated parsing rule.

13. The non-transitory computer readable storage medium of claim 12, wherein determining whether the first input sentence has an underserving signal further comprises, for each first input sentence:

determining that the first input sentence was processed as a search query;

determining a quality metric value for search results provided in response to the search query; and

determining the first input sentence has the underserving signal when the quality metric value does not meet a threshold quality value.

14. The non-transitory computer readable storage medium of claim 13, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of quality of resources identified by the search results, the measure of quality being independent of user interactions at the user device from which the first input sentence was received.

15. The non-transitory computer readable storage medium of claim 13, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of user satisfaction in response to the search results, the measure of user satisfaction being dependent on user interactions with the search results at the user device from which the first input sentence was received.

16. The non-transitory computer readable storage medium of claim 11, wherein determining whether the first input sentence has an underserving signal further comprises, for each first input sentence:

determining that the first input sentence was processed as a search query;

determining a quality metric value for search results provided in response to the search query; and

determining the first input sentence has the underserving signal when the quality metric value does not meet a threshold quality value.

17. The non-transitory computer readable storage medium of claim 16, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of quality of resources identified by the search results, the measure of quality being independent of user interactions at the user device from which the first input sentence was received.

18. The non-transitory computer readable storage medium of claim 16, wherein determining a quality metric value for search results provided in response to the search query comprises determining a quality score that is a measure of user satisfaction in response to the search results, the measure of user satisfaction being dependent on user interactions with the search results at the user device from which the first input sentence was received.

19. The non-transitory computer readable storage medium of claim 11, wherein the input sentences are generated from voice command inputs.

20. A system, comprising:

a data processing apparatus; and

a non-transitory computer readable storage medium in data communication with the data processing apparatus storing instructions executable by the data processing apparatus and that upon such execution causes the data processing apparatus to perform operations comprising;

identifying from the input sentences first input sentences that do not successfully parse by any parsing rule in a set of parsing rules, each of the parsing rules being associated with an action, and wherein each parsing rule causes a user device to perform the action associated with the parsing rule in response to an input sentence received from the user device being successfully parsed by the parsing rule;

for each first input sentence, determining whether the first input sentence has an underserving signal, the underserving signal being a signal that is indicative of one or more actions being underserved; and

selecting first input sentences determined to have an underserving signal as candidate input sentences;

providing each candidate input sentence to an action analysis process that determines whether a candidate input sentence is to be associated with one of a plurality of actions and upon a positive determination generates a parsing rule for the candidate input sentence, wherein the generated parsing rule is associated with the one of the plurality of actions and a successful parse of an input sentence received at a user device by the generated parsing rule causes the user device to perform the one of the plurality of actions associated with the generated parsing rule; and

persisting each generated parsing rule and data describing the action with which the generated parsing rule is associated to command models data that is accessible by a language processing system that parses command input sentences by the generated parsing rules and determines actions to perform based on a successful parses.