Method and apparatus for determining speech interaction satisfaction

1. A method for determining a speech interaction satisfaction, comprising:

acquiring a speech interaction feature, the speech interaction feature including objective data of speech interaction and subjective data of the speech interaction, and the objective data and the subjective data being data for a given topic, wherein the objective data of the speech interaction includes a user intent, response content, a response delay, and a current playback duration of the response content; and the subjective data of the speech interaction includes: text information corresponding to a voice input of a user after playback of the response content is interrupted, or text information inputted by the user after the playback of the response content is completed;

evaluating the objective data to obtain an objective evaluation, and evaluating the subjective data to obtain a subjective evaluation; and

obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model.

2. The method according to claim 1 , wherein the evaluating the objective data to obtain an objective evaluation comprises:

acquiring a first objective evaluation based on an intent matching degree between the user intent and the response content;

acquiring a second objective evaluation based on the response delay and a standard delay; and

acquiring a third objective evaluation based on the current playback duration of the response content and a standard playback duration of the response content; and

the obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model comprises:

obtaining the speech interaction satisfaction outputted by the satisfaction evaluation model by using the first objective evaluation, the second objective evaluation, the third objective evaluation, and the subjective evaluation as the input of the satisfaction evaluation model.

3. The method according to claim 1 , wherein the evaluating the subjective data to obtain a subjective evaluation comprises:

performing a semantic analysis on the text information, to obtain a content attribute corresponding to the text information, wherein the content attribute is an emotional attribute or a topic attribute; and

obtaining the subjective evaluation based on the content attribute corresponding to the text information.

4. The method according to claim 3 , wherein, if the content attribute is the topic attribute, the obtaining the subjective evaluation based on the content attribute corresponding to the text information comprises:

acquiring a topic type corresponding to the text information; and

determining the subjective evaluation being less than a preset evaluation value, if the topic type corresponding to the text information is consistent with the topic type corresponding to the user intent.

5. The method according to claim 3 , wherein, if the content attribute is the emotional attribute, the obtaining the subjective evaluation based on the content attribute corresponding to the text information comprises:

extracting an emotional keyword in the text information;

acquiring an emotional type according to a corresponding relationship between the emotional keyword and the emotional type, wherein the emotional type includes a positive emotion, a negative emotion, and a neutral emotion; and

acquiring the subjective evaluation according to a corresponding relationship between the emotional type and a preset evaluation.

6. The method according to claim 1 , wherein the acquiring a speech interaction feature comprises:

acquiring first log data, wherein an interval between the first log data and second log data in a last period adjacent to the first log data, and the interval between the first log data and third log data in a next period adjacent to the first log data are greater than a preset threshold;

acquiring, in the first log data, topics respectively corresponding to two adjacent voice inputs of the user; and

acquiring the speech interaction feature based on the topics respectively corresponding to the two adjacent voice inputs.

7. The method according to claim 1 , wherein before the obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model, the method further comprises:

acquiring a training sample set, wherein the training sample set includes a first sample evaluation, a second sample evaluation, and a satisfaction inputted by the user, the first sample evaluation is obtained by evaluating objective sample data, the second sample evaluation is obtained by evaluating subjective sample data, and the objective sample data and the subjective sample data are data for an identical topic; and

obtaining the satisfaction evaluation model by performing iterative training on the basis of the training sample set.

8. An apparatus for determining a speech interaction satisfaction, 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:

acquiring a speech interaction feature, the speech interaction feature including objective data of speech interaction and subjective data of the speech interaction, and the objective data and the subjective data being data for a given topic, wherein the objective data of the speech interaction includes a user intent, response content, a response delay, and a current playback duration of the response content; and the subjective data of the speech interaction includes: text information corresponding to a voice input of a user after playback of the response content is interrupted, or text information inputted by the user after the playback of the response content is completed;

evaluating the objective data to obtain an objective evaluation, and evaluating the subjective data to obtain a subjective evaluation; and

obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model.

9. The apparatus according to claim 8 , wherein the evaluating the objective data to obtain an objective evaluation comprises:

acquiring a first objective evaluation based on an intent matching degree between the user intent and the response content;

acquiring a second objective evaluation based on the response delay and a standard delay; and

acquiring a third objective evaluation based on the current playback duration of the response content and a standard playback duration of the response content; and

the obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model comprises:

obtaining the speech interaction satisfaction outputted by the satisfaction evaluation model by using the first objective evaluation, the second objective evaluation, the third objective evaluation, and the subjective evaluation as the input of the satisfaction evaluation model.

10. The apparatus according to claim 8 , wherein the evaluating the subjective data to obtain a subjective evaluation comprises:

performing a semantic analysis on the text information, to obtain a content attribute corresponding to the text information, wherein the content attribute is an emotional attribute or a topic attribute; and

obtaining the subjective evaluation based on the content attribute corresponding to the text information.

11. The apparatus according to claim 10 , wherein, if the content attribute is the topic attribute, the obtaining the subjective evaluation based on the content attribute corresponding to the text information comprises:

acquiring a topic type corresponding to the text information; and

determining the subjective evaluation being less than a preset evaluation value, if the topic type corresponding to the text information is consistent with the topic type corresponding to the user intent.

12. The apparatus according to claim 10 , wherein, if the content attribute is the emotional attribute, the obtaining the subjective evaluation based on the content attribute corresponding to the text information comprises:

extracting an emotional keyword in the text information;

acquiring an emotional type according to a corresponding relationship between the emotional keyword and the emotional type, wherein the emotional type includes a positive emotion, a negative emotion, and a neutral emotion; and

acquiring the subjective evaluation according to a corresponding relationship between the emotional type and a preset evaluation.

13. The apparatus according to claim 8 , wherein the acquiring a speech interaction feature comprises:

acquiring first log data, wherein an interval between the first log data and second log data in a last period adjacent to the first log data, and the interval between the first log data and third log data in a next period adjacent to the first log data are greater than a preset threshold;

acquiring, in the first log data, topics respectively corresponding to two adjacent voice inputs of the user; and

acquiring the speech interaction feature based on the topics respectively corresponding to the two adjacent voice inputs.

14. The apparatus according to claim 8 , before the obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model, the operations further comprise:

acquiring a training sample set, wherein the training sample set includes a first sample evaluation, a second sample evaluation, and a satisfaction inputted by the user, the first sample evaluation is obtained by evaluating objective sample data, the second sample evaluation is obtained by evaluating subjective sample data, and the objective sample data and the subjective sample data are data for an identical topic; and

obtaining the satisfaction evaluation model by performing iterative training on the basis of the training sample set.

15. A non-transitory computer-readable storage medium storing a computer program, the computer program when executed by one or more processors, causes the one or more processors to perform operations, the operations comprising:

acquiring a speech interaction feature, the speech interaction feature including objective data of speech interaction and subjective data of the speech interaction, and the objective data and the subjective data being data for a given topic, wherein the objective data of the speech interaction includes a user intent, response content, a response delay, and a current playback duration of the response content; and the subjective data of the speech interaction includes: text information corresponding to a voice input of a user after playback of the response content is interrupted, or text information inputted by the user after the playback of the response content is completed;

evaluating the objective data to obtain an objective evaluation, and evaluating the subjective data to obtain a subjective evaluation; and

obtaining a speech interaction satisfaction outputted by a satisfaction evaluation model by using the objective evaluation and the subjective evaluation as an input of the satisfaction evaluation model.