Method, apparatus for evaluating review, device and storage medium

Patent metadata

Field	Value
Publication number	US-11403680-B2
Application number	US-201916421921-A
Country	US
Kind code	B2
Filing date	May 24, 2019
Priority date	Feb 12, 2018
Publication date	Aug 2, 2022
Grant date	Aug 2, 2022

Abstract

According to an exemplary embodiment of the present disclosure, a method, apparatus for evaluating a review, a device and a computer readable storage medium are provided. The method for evaluating a review includes obtaining a first vectorized representation of a set of text items in a review for a target object. The method further includes extracting a semantic feature of at least two consecutive text items of the set of text items based on the first vectorized representation. The method further includes determining a degree of importance of the at least two consecutive text items in a context of the review, and determining a degree of the review helping a user to evaluate the target object, based on the degree of importance and the semantic feature. In this way, an automatic, efficient, and accurate evaluation of the value of a review may be achieved.

First claim

What is claimed is:

1. A method for evaluating a review, the method comprising:
obtaining a first vectorized representation of a set of text items in a review for a target object;
extracting a semantic feature of at least two consecutive text items of the set of text items based on the first vectorized representation;
determining a degree of importance of the at least two consecutive text items in a context of the review;
determining a degree of the review helping a user to evaluate the target object, based on the degree of importance and the semantic feature;
obtaining a training review for a training object and a target degree of the training review helping the user to evaluate the training object; and
determining a parameter set of a learning network based on the training review and the target degree, such that an error between a predicted degree of the training review helping the user to evaluate the training object determined using the learning network and the target degree is within a first error threshold, wherein the learning network is utilized to perform at least one of: the extracting of the semantic feature, the determining of the degree of importance, or the determining of the degree of the review helping the user to evaluate the target object.

2. The method according to claim 1 , wherein extracting the semantic feature comprises:
obtaining a second vectorized representation of subtext items in the set of text items;
combining the first vectorized representation and the second vectorized representation to generate a combined vectorized representation; and
extracting the semantic feature from the combined vectorized representation.

3. The method according to claim 2 , wherein each text item comprises a word in the review, and each subtext item comprises a character that makes up a word.

4. The method according to claim 2 , wherein each text item comprises a set of words in the review, and each subtext item comprises a single word.

5. The method according to claim 1 , wherein the determining a degree of the review helping a user to evaluate the target object comprises:
weighting the semantic feature with the degree of importance to generate a combined feature of the review; and
determining the degree of the review helping the user to evaluate the target object based on the combined feature.

6. The method according to claim 1 , wherein determining the parameter set of the learning network further comprises:
obtaining a target score of the training object by a reviewer of the training review; and
determining the parameter set of the learning network based on the target score, such that an error between a predicted score of the training object by the reviewer determined using the learning network and the target score is within a second error threshold.

7. The method according to claim 1 , further comprising:
obtaining an initial vectorized representation of text items included in the training review; and
updating the initial vectorized representation while determining the parameter set to obtain the first vectorized representation.

8. An apparatus for evaluating a review, the apparatus comprising:
at least one processor; and
a memory storing instructions, the instructions when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
obtaining a first vectorized representation of a set of text items in a review for a target object;
extracting a semantic feature of at least two consecutive text items of the set of text items based on the first vectorized representation;
determining a degree of importance of the at least two consecutive text items in a context of the review;
determining a degree of the review helping a user to evaluate the target object, based on the degree of importance and the semantic feature;
obtaining a training review for a training object and a target degree of the training review helping the user to evaluate the training object; and
determining a parameter set of a learning network based on the training review and the target degree, such that an error between a predicted degree of the training review helping the user to evaluate the training object determined using the learning network and the target degree is within a first error threshold, wherein the learning network is utilized to perform at least one of: the extracting of the semantic feature, the determining of the degree of importance, or the determining of the degree of the review helping the user to evaluate the target object.

9. The apparatus according to claim 8 , wherein the extracting the semantic feature comprises:
obtaining a second vectorized representation of subtext items in the set of text items;
combining the first vectorized representation and the second vectorized representation to generate a combined vectorized representation; and
extracting the semantic feature from the combined vectorized representation.

10. The apparatus according to claim 9 , wherein each text item comprises a word in the review, and each subtext item comprises a character that makes up a word.

11. The apparatus according to claim 9 , wherein each text item comprises a set of words in the review, and each subtext item comprises a single word.

12. The apparatus according to claim 8 , wherein the determining a degree of the review helping a user to evaluate the target object comprises:
weighting the semantic feature with the degree of importance to generate a combined feature of the review; and
determining the degree of the review helping the user to evaluate the target object based on the combined feature.

13. The apparatus according to claim 8 , wherein determining the parameter set of the learning network further comprises:
obtaining a target score of the training object by a reviewer of the training review; and
determining the parameter set of the learning network based on the target score, such that an error between a predicted score of the training object by the reviewer determined using the learning network and the target score is within a second error threshold.

14. The apparatus according to claim 8 , the operations further comprising:
obtaining an initial vectorized representation of text items included in the training review; and
updating the initial vectorized representation while determining the parameter set to obtain the first vectorized representation.

15. A non-transitory computer readable storage medium, storing a computer program thereon, the program, when executed by a processor, causes the processor to perform operations, the operations comprising:
obtaining a first vectorized representation of a set of text items in a review for a target object;
extracting a semantic feature of at least two consecutive text items of the set of text items based on the first vectorized representation;
determining a degree of importance of the at least two consecutive text items in a context of the review;
determining a degree of the review helping a user to evaluate the target object, based on the degree of importance and the semantic feature;
obtaining a training review for a training object and a target degree of the training review helping the user to evaluate the training object; and
determining a parameter set of the learning network based on the training review and the target degree, such that an error between a predicted degree of the training review helping the user to evaluate the training object determined using the learning network and the target degree is within a first error threshold.

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Fan Miao
• Feng Yue
• Sun Mingming
• Li Ping
• Wang Haifeng

Classifications

• G06V30/413

  Classification of content, e.g. text, photographs or tables · CPC title

• G06V10/82

  using neural networks · CPC title

• G06F18/2178

  based on feedback of a supervisor · CPC title

• G06Q30/0282 Primary

  Rating or review of business operators or products · CPC title

• G06F40/30

  Semantic analysis · CPC title