Granted Patent B2

US 11,983,207 · App. 17/146,558 · Granted May 14, 2024

Inventors: Zijia Wang (Shanghai, CN); Jiacheng Ni (Shanghai, CN); Zhen Jia (Shanghai, CN); Bo Wei (Shanghai, CN); Chun Xi Chen (Shanghai, CN)

Assignee: EMC IP Holding Company LLC

G06F16/3346G06F16/3344G06F16/353G06N5/04G06N20/00

Abstract

Claims (103)

1. An information processing method, including:

receiving text comprising a service request from a user of a storage system;

determining, utilizing a first machine learning model of a processor-implemented machine learning system, and based on multiple weights corresponding to multiple words in the text, a target object associated with the text among predetermined multiple objects, and a set of key words among the multiple words and with respect to the determination of the target object;

determining, utilizing a second machine learning model of the processor-implemented machine learning system, and among the set of key words, a set of target words related to a text topic of the text;

outputting, by the processor-implemented machine learning system, the set of target words and an identifier of the target object in an associated manner, wherein outputting the set of target words and the identifier of the target object comprises generating at least one visualization in which the set of target words is associated with the identifier of the target object in a visually perceptible manner;

processing the outputted set of target words and the identifier of the target object to generate a corresponding recommendation;

sending the recommendation to the user as at least a portion of a service response configured to address the service request in the text; and

implementing one or more actions relating to the storage system to address the service request in the text based on the recommendation;

wherein the first machine learning model of the processor-implemented machine learning system comprises a trained text classification model for natural language processing;

wherein the second machine learning model of the processor-implemented machine learning system comprises a topic model;

wherein determining the set of target words utilizing the second machine learning model includes:

determining a first number of text topic correlations between a first number of words in the set of key words and the text topic, respectively;

determining, among the first number of text topic correlations, a second number of text topic correlations according to an order of their magnitudes; and

determining, from the first number of words, a second number of words corresponding to the second number of text topic correlations;

wherein the multiple weights include a first weight of a first word in the multiple words, and the first weight includes multiple first weight components that are determined based on multiple word correlations between the multiple words and the first word, respectively; and

wherein generating the at least one visualization further comprises generating visual representations of respective ones of the multiple word correlations.

2. The method according to claim 1 , wherein determining the set of key words includes:

determining multiple weight averages respectively corresponding to the multiple weights, each weight average being determined based on multiple weight components of a corresponding weight;

determining, among the multiple weight averages, a first number of weight averages according to an order of their magnitudes; and

determining, from the multiple words, a first number of words corresponding to the first number of weight averages.

3. The method according to claim 1 , wherein determining the first number of text topic correlations includes:

determining multiple probabilities that the text topic is multiple predetermined topics, respectively;

determining multiple predetermined topic correlations between a first word in the first number of words and the multiple predetermined topics, respectively; and

determining a first text topic correlation corresponding to the first word based on the multiple predetermined topic correlations and the multiple probabilities.

4. The method according to claim 3 , wherein determining the multiple predetermined topic correlations includes:

determining a first set of words included in a first predetermined topic in the multiple predetermined topics and a corresponding first set of probabilities;

determining a first set of word vectors of the first set of words;

determining a first predetermined topic vector of the first predetermined topic based on the first set of probabilities and the first set of word vectors; and

determining a first predetermined topic correlation corresponding to the first predetermined topic based on a word vector of the first word and the first predetermined topic vector.

5. The method according to claim 3 , wherein determining the multiple probabilities includes:

determining the multiple probabilities based on the topic model for determining the text topic, wherein the topic model is trained using a set of text associated with the multiple predetermined objects.

6. The method according to claim 1 , wherein outputting the set of target words includes:

outputting the set of target words as explanation information about the association between the text and the target object.

7. The method according to claim 1 , wherein the text is a description text about a failure of the storage system, and the multiple predetermined objects are predetermined multiple knowledge bases, the multiple knowledge bases including different sets of text associated with solving different types of failures of the storage system.

8. An electronic device, including:

at least one processor; and

at least one memory storing computer program instructions, wherein the at least one memory and the computer program instructions are configured to cause, along with the at least one processor, the electronic device to:

receive text comprising a service request from a user of a storage system;

determine, utilizing a first machine learning model of a processor-implemented machine learning system, and based on multiple weights corresponding to multiple words in the text, a target object associated with the text among predetermined multiple objects, and a set of key words among the multiple words and with respect to the determination of the target object;

determine, utilizing a second machine learning model of the processor-implemented machine learning system, and among the set of key words, a set of target words related to a text topic of the text;

output, by the processor-implemented machine learning system, the set of target words and an identifier of the target object in an associated manner, wherein outputting the set of target words and the identifier of the target object comprises generating at least one visualization in which the set of target words is associated with the identifier of the target object in a visually perceptible manner;

process the outputted set of target words and the identifier of the target object to generate a corresponding recommendation;

send the recommendation to the user as at least a portion of a service response configured to address the service request in the text; and

implement one or more actions relating to the storage system to address the service request in the text based on the recommendation;

wherein the first machine learning model of the processor-implemented machine learning system comprises a trained text classification model for natural language processing;

wherein the second machine learning model of the processor-implemented machine learning system comprises a topic model;

wherein determining the set of target words includes:

determining a first number of text topic correlations between a first number of words in the set of key words and the text topic, respectively;

determining, among the first number of text topic correlations, a second number of text topic correlations according to an order of their magnitudes; and

determining, from the first number of words, a second number of words corresponding to the second number of text topic correlations;

wherein the multiple weights include a first weight of a first word in the multiple words, and the first weight includes multiple first weight components that are determined based on multiple word correlations between the multiple words and the first word, respectively; and

wherein generating the at least one visualization further comprises generating visual representations of respective ones of the multiple word correlations.

9. The electronic device according to claim 8 , wherein the at least one memory and the computer program instructions are configured to cause, along with the at least one processor, the electronic device to determine the set of key words by:

determining multiple weight averages respectively corresponding to the multiple weights, each weight average being determined based on multiple weight components of a corresponding weight;

determining, among the multiple weight averages, a first number of weight averages according to an order of their magnitudes; and

determining, from the multiple words, a first number of words corresponding to the first number of weight averages.

10. The electronic device according to claim 8 , wherein the at least one memory and the computer program instructions are configured to cause, along with the at least one processor, the electronic device to determine the first number of text topic correlations by:

determining multiple probabilities that the text topic is multiple predetermined topics, respectively;

determining multiple predetermined topic correlations between a first word in the first number of words and the multiple predetermined topics, respectively; and

determining a first text topic correlation corresponding to the first word based on the multiple predetermined topic correlations and the multiple probabilities.

11. The electronic device according to claim 10 , wherein the at least one memory and the computer program instructions are configured to cause, along with the at least one processor, the electronic device to determine the multiple predetermined topic correlations by:

determining a first set of words included in a first predetermined topic in the multiple predetermined topics and a corresponding first set of probabilities;

determining a first set of word vectors of the first set of words;

determining a first predetermined topic vector of the first predetermined topic based on the first set of probabilities and the first set of word vectors; and

determining a first predetermined topic correlation corresponding to the first predetermined topic based on a word vector of the first word and the first predetermined topic vector.

12. The electronic device according to claim 10 , wherein the at least one memory and the computer program instructions are configured to cause, along with the at least one processor, the electronic device to determine the multiple probabilities by:

determining the multiple probabilities based on the topic model for determining the text topic, wherein the topic model is trained using a set of text associated with the multiple predetermined objects.

13. The electronic device according to claim 8 , wherein the at least one memory and the computer program instructions are configured to cause, along with the at least one processor, the electronic device to output the set of target words by:

outputting the set of target words as explanation information about the association between the text and the target object.

14. A computer program product tangibly stored on a non-transitory computer-readable medium and including machine-executable instructions, wherein the machine-executable instructions, when executed, cause a machine to perform steps of an information processing method, the method including:

receiving text comprising a service request from a user of a storage system;

determining, utilizing a first machine learning model of a processor-implemented machine learning system, and based on multiple weights corresponding to multiple words in the text, a target object associated with the text among predetermined multiple objects, and a set of key words among the multiple words and with respect to the determination of the target object;

determining, utilizing a second machine learning model of the processor-implemented machine learning system, and among the set of key words, a set of target words related to a text topic of the text;

outputting, by the processor-implemented machine learning system, the set of target words and an identifier of the target object in an associated manner, wherein outputting the set of target words and the identifier of the target object comprises generating at least one visualization in which the set of target words is associated with the identifier of the target object in a visually perceptible manner;

processing the outputted set of target words and the identifier of the target object to generate a corresponding recommendation;

sending the recommendation to the user as at least a portion of a service response configured to address the service request in the text; and

implementing one or more actions relating to the storage system to address the service request in the text based on the recommendation;

wherein the first machine learning model of the processor-implemented machine learning system comprises a trained text classification model for natural language processing;

wherein the second machine learning model of the processor-implemented machine learning system comprises a topic model;

wherein determining the set of target words includes:

determining a first number of text topic correlations between a first number of words in the set of key words and the text topic, respectively;

determining, among the first number of text topic correlations, a second number of text topic correlations according to an order of their magnitudes; and

determining, from the first number of words, a second number of words corresponding to the second number of text topic correlations;

wherein the multiple weights include a first weight of a first word in the multiple words, and the first weight includes multiple first weight components that are determined based on multiple word correlations between the multiple words and the first word, respectively; and

wherein generating the at least one visualization further comprises generating visual representations of respective ones of the multiple word correlations.

15. The computer program product according to claim 14 , wherein determining the first number of text topic correlations includes:

determining multiple probabilities that the text topic is multiple predetermined topics, respectively;

determining multiple predetermined topic correlations between a first word in the first number of words and the multiple predetermined topics, respectively; and

determining a first text topic correlation corresponding to the first word based on the multiple predetermined topic correlations and the multiple probabilities.

16. The computer program product according to claim 15 , wherein determining the multiple predetermined topic correlations includes:

determining a first set of words included in a first predetermined topic in the multiple predetermined topics and a corresponding first set of probabilities;

determining a first set of word vectors of the first set of words;

determining a first predetermined topic vector of the first predetermined topic based on the first set of probabilities and the first set of word vectors; and

determining a first predetermined topic correlation corresponding to the first predetermined topic based on a word vector of the first word and the first predetermined topic vector.

17. The computer program product according to claim 15 , wherein determining the multiple probabilities includes:

determining the multiple probabilities based on the topic model for determining the text topic, wherein the topic model is trained using a set of text associated with the multiple predetermined objects.

18. The computer program product according to claim 14 , wherein outputting the set of target words includes:

outputting the set of target words as explanation information about the association between the text and the target object.

19. The computer program product according to claim 14 , wherein determining the set of key words includes:

determining multiple weight averages respectively corresponding to the multiple weights, each weight average being determined based on multiple weight components of a corresponding weight;

determining, among the multiple weight averages, a first number of weight averages according to an order of their magnitudes; and

determining, from the multiple words, a first number of words corresponding to the first number of weight averages.

20. The computer program product according to claim 14 , wherein the text is a description text about a failure of the storage system, and the multiple predetermined objects are predetermined multiple knowledge bases, the multiple knowledge bases including different sets of text associated with solving different types of failures of the storage system.

Assignments (9)