An intelligent strategy quality inspection method and platform based on real-time ASR voice stream

Abstract

The present invention provides an intelligent strategy quality inspection method and platform based on real-time ASR voice stream, which belongs to the field of data processing technology, and specifically includes: taking historical communication feedback data of an existing communication processing object of an agent as basic data, determining the degree of similarity with different violation processing objects, determining the quality inspection strategy of the agent based on the degree of similarity and the probability of illegal communication, when performing communication processing on the existing communication processing object, performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection test result, using the quality inspection test result to determine the recognition result and change of the non-compliant behavior of the agent, and determining the intervention processing strategy of the agent according to the deviation between the recognition result, the change and the quality inspection requirement of the quality inspection strategy, thereby ensuring the timeliness of the intervention processing of the violation.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to an intelligent strategy quality inspection method and platform based on real-time ASR voice stream. Background Art

[0002] In order to implement quality inspection of agents and avoid illegal statements or behaviors of agents, the existing technical solution in CN202311103890.4 "A communication sensitive word quality inspection method and device based on a large model" uses multiple methods to perform quality inspection on the communication statements of agents, thereby identifying non-compliant behaviors and ensuring the compliance of communication behaviors. However, there are the following problems:

[0003] Existing technical solutions often focus on the non-compliant behaviors of agents in the historical communication process, but ignore the detection of real-time non-compliant behaviors of agents, making it difficult to implement timely intervention in non-compliant behaviors, and further making it difficult to meet the compliance requirements of communication.

[0004] In view of the above technical problems, the present invention provides an intelligent strategy quality inspection method and platform based on real-time ASR voice stream. Summary of the invention

[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0006] According to one aspect of the present invention, an intelligent policy quality inspection method based on real-time ASR voice stream is provided.

[0007] An intelligent strategy quality inspection method based on real-time ASR voice stream, specifically comprising:

[0008] S1 obtains historical communication data of the agent, uses the historical communication data to identify illegal behaviors at different historical communication times to obtain an identification result, and determines the probability of illegal communication of the agent based on the identification result;

[0009] S2: taking the communication processing object in which the agent has violated the rules as the violation processing object, and using the historical communication feedback data of the existing communication processing object of the agent as the basic data, determining the similarity with different violation processing objects, and determining the quality inspection strategy of the agent based on the similarity and the probability of violation communication;

[0010] S3, when performing communication processing on the existing communication processing object, performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection result;

[0011] S4 uses the quality inspection results to determine the identification results and changes of the non-compliant behaviors of the seat personnel, and determines the intervention and processing strategy of the seat personnel according to the deviations between the identification results, changes and the quality inspection requirements of the quality inspection strategy.

[0012] A further technical solution is that the historical communication data of the agent is determined based on the communication records of the agent in processing different communication processing objects.

[0013] A further technical solution is that the identification result includes the number of occurrences of different types of illegal behaviors of the agent.

[0014] A further technical solution is that the value range of the probability of illegal communication of the seat personnel is between 0 and 1, wherein the greater the probability of illegal communication of the seat personnel, the greater the probability of illegal communication of the seat personnel.

[0015] A further technical solution is that the quality inspection results include the number of occurrences of different types of violations.

[0016] A further technical solution is that the method for determining the intervention processing strategy of the seat staff is:

[0017] Determine the number of occurrences of different types of violations based on the identification results, and determine the severity of the violations based on the deviation between the number of occurrences and the quality inspection requirements of the quality inspection strategy;

[0018] Divide the current communication data of the seat staff into multiple time periods, obtain the change in the number of violations between different adjacent time periods, determine the screening of the time period when the violations are getting worse based on the change, and determine the violation change problem value based on the number ratio of the time period when the violations are getting worse;

[0019] The weights of the violation severity and the violation change problem value are used to determine the intervention processing problem value of the seat personnel, and the intervention processing strategy of the seat personnel is determined according to the intervention processing problem value.

[0020] A further technical solution is to determine the intervention processing strategy of the agent according to the intervention processing problem value, which specifically includes:

[0021] The intervention processing problem value is used to determine the violation processing type of the agent, and the intervention processing strategy of the agent is determined using the violation processing type.

[0022] On the other hand, the present invention provides an intelligent strategy quality inspection platform based on real-time ASR voice stream, which adopts the above-mentioned intelligent strategy quality inspection method based on real-time ASR voice stream, specifically comprising:

[0023] Violation probability assessment module, quality inspection strategy output module, test result output module, intervention processing module;

[0024] The violation probability assessment module is responsible for obtaining the historical communication data of the seat personnel, using the historical communication data to identify the violation behaviors at different historical communication times to obtain the identification results, and determining the violation communication probability of the seat personnel based on the identification results;

[0025] The quality inspection strategy output module is responsible for taking the communication processing object in which the agent has violated the rules as the violation processing object, and using the historical communication feedback data of the existing communication processing object of the agent as the basic data to determine the similarity with different violation processing objects, and determining the quality inspection strategy of the agent based on the similarity and the probability of violation communication;

[0026] The detection result output module is responsible for performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection result when performing communication processing on the existing communication processing object;

[0027] The intervention processing module is responsible for using the quality inspection results to determine the identification results and changes of the non-compliant behaviors of the seat personnel, and determining the intervention processing strategy of the seat personnel based on the deviations between the identification results, changes and the quality inspection requirements of the quality inspection strategy.

[0028] The beneficial effects of the present invention are:

[0029] 1. In the present invention, the identification results of illegal behaviors in different historical communication times are used to determine the probability of illegal communication of the agent, thereby realizing the evaluation of the probability of illegal communication of the agent based on the identification of illegal behaviors in the historical communication times, realizing the screening of agents with a higher probability of illegal communication, avoiding the technical problem that the efficiency of communication processing cannot meet the requirements due to the same communication strategy for all agents, and improving the efficiency of identification and processing of illegal behaviors and the timeliness of processing.

[0030] 2. In the present invention, the intervention and processing strategy of the agent is determined by the deviation between the identification results, the changes and the quality inspection requirements of the quality inspection strategy. Not only the difference in the agent's demand for the access and processing strategy caused by the number of violations is taken into account, but also the changes in the violations are taken into account. It realizes the screening of agents who are emotionally excited or whose violations increase too quickly, thereby realizing timely intervention in the communication behavior and ensuring the reliability of communication processing.

[0031] Other features and advantages will be described in the following description. The objects and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and drawings.

[0032] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings.

[0034] Figure 1 It is a flow chart of an intelligent strategy quality inspection method based on real-time ASR voice stream;

[0035] Figure 2 It is a flow chart of a method for determining the probability of illegal communication by an agent;

[0036] Figure 3 is a flow chart of a method for determining the degree of similarity with different violation processing objects;

[0037] Figure 4 It is a flow chart of the method for determining the quality inspection strategy of the seat staff;

[0038] Figure 5 It is a framework diagram of an intelligent strategy quality inspection platform based on real-time ASR voice stream. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0040] Speech recognition technology, also known as Automatic Speech Recognition (ASR), aims to convert the vocabulary content in human speech into computer-readable input, such as keystrokes, binary codes, or character sequences. It is different from speaker recognition and speaker verification, which try to identify or verify the speaker of the speech rather than the vocabulary content contained in it.

[0041] Existing technical solutions often focus on the non-compliant behaviors of agents in the historical communication process, but ignore the detection of real-time non-compliant behaviors of agents, making it difficult to implement timely intervention in non-compliant behaviors, and further making it difficult to meet the compliance requirements of communication.

[0042] Probability of illegal communication: It is determined based on the proportion of historical communication times with violations. When the proportion of historical communication times with violations is 0.2, the probability of illegal communication is 0.2.

[0043] The degree of similarity with different violation processing objects: determined based on the similarity between historical communication return data and the violation processing objects, the proportion of communication sentences of different statement types is determined based on the proportion of communication feedback sentences of different statement types, the similarity of communication sentences of different statement types between the existing communication processing object and the violation processing object is determined based on the deviation of the proportion of communication sentences of different statement types, the degree of similarity between the existing communication processing object and the different violation processing objects is determined based on the average value of the similarities of communication sentences of different statement types, specifically, when the average value of the similarities of communication sentences of different statement types is 0.7, the degree of similarity between the existing communication processing object and the different violation processing objects is 0.7.

[0044] Quality inspection strategy for seat personnel: similar processing objects are determined by the degree of similarity with different violation processing objects, and the quality inspection strategy for similar processing objects is used to determine the quality inspection strategy for the seat personnel.

[0045] Intervention strategy for agents: Determine the intervention strategy for agents based on the most severe intervention strategy among the preset intervention strategies corresponding to the number of occurrences of different types of violations and the number of time periods during which the violations deteriorate.

[0046] Embodiment 1 To solve the above problem, according to one aspect of the present invention, Figure 1 As shown, an intelligent strategy quality inspection method based on real-time ASR voice stream is provided, which specifically includes:

[0047] S1 obtains historical communication data of the agent, uses the historical communication data to identify illegal behaviors at different historical communication times to obtain an identification result, and determines the probability of illegal communication of the agent based on the identification result;

[0048] Furthermore, the historical communication data of the agent is determined based on the communication records of the agent in processing different communication processing objects.

[0049] It should be noted that the identification result includes the number of occurrences of different types of violations of the agent.

[0050] Specifically, Figure 2 As shown, the method for determining the probability of illegal communication of the seat personnel is:

[0051] The number of historical communications with violations is used as the number of illegal communications, and the violation weight values ​​of different types of violations are obtained. The communication problem values ​​of different illegal communication numbers are determined based on the number of occurrences of different types of violations.

[0052] Determine the basic probability of illegal communication of the agent by the proportion of illegal communication times in the historical communication times, and determine the compensation amount of illegal communication probability by the average value of communication problem values ​​of different illegal communication times;

[0053] The violation communication probability of the agent is determined by using the sum of the violation probability compensation amount and the basic violation communication probability.

[0054] It can be understood that the violation weight value of the violation behavior is determined according to a preset corresponding relationship between the type of the violation behavior and the violation weight value.

[0055] Furthermore, the value range of the probability of illegal communication of the seat personnel is between 0 and 1, wherein the greater the probability of illegal communication of the seat personnel is, the greater the probability of illegal communication of the seat personnel is.

[0056] In another possible embodiment, the method for determining the probability of illegal communication of the agent is:

[0057] The number of historical communications with violations is used as the number of illegal communications, and whether the number of illegal communications is greater than the preset number of communications is determined. If so, the probability of illegal communication of the agent is determined using the preset communication probability. If not, proceed to the next step.

[0058] Obtain the number of occurrences of different types of violations by the agent, and determine whether the total number of occurrences of the violations by the agent meets the requirement. If so, proceed to the next step; if not, determine the probability of the agent's illegal communication using the preset communication probability;

[0059] Obtaining violation weight values ​​of different types of violation behaviors, and determining the sum of the violation weight values ​​of different types of violation behaviors in combination with the number of occurrences of different types of violation behaviors, normalizing the sum of the violation weight values ​​of different types of violation behaviors to obtain a processing result, using the sum of the processing results of different types of violation behaviors to determine the violation risk, and judging whether the violation risk of the seat personnel meets the requirements, if so, proceeding to the next step, if not, using the preset communication probability to determine the violation communication probability of the seat personnel;

[0060] Obtain violation weight values ​​of different types of violation behaviors, and determine communication problem values ​​of different violation communication times in combination with the occurrence times of different types of violation behaviors, and determine whether the number of violation communication times for which the communication problem value does not meet the requirements is greater than the preset number of violation communication times. If so, determine the violation communication probability of the agent using the preset communication probability, and if not, proceed to the next step;

[0061] The basic illegal communication probability of the agent is determined by the proportion of the number of illegal communications in the historical communication times, the violation probability compensation amount is determined by the average value of the communication problem values ​​of different illegal communication times, and the violation communication probability of the agent is determined by the sum of the violation probability compensation amount and the basic illegal communication probability.

[0062] It should be noted that, when the communication problem value of the number of illegal communications is not within a preset interval, it is determined that the communication problem value of the number of illegal communications does not meet the requirements.

[0063] S2: taking the communication processing object in which the agent has violated the rules as the violation processing object, and using the historical communication feedback data of the existing communication processing object of the agent as the basic data, determining the similarity with different violation processing objects, and determining the quality inspection strategy of the agent based on the similarity and the probability of violation communication;

[0064] Specifically, it should be noted that Figure 3 As shown, the method for determining the degree of similarity with different violation processing objects is:

[0065] Determine communication feedback sentences of different sentence types based on historical communication feedback data, and determine the proportion of communication sentences of different sentence types according to the proportion of communication feedback sentences of different sentence types;

[0066] Determine the similarity of the communication sentences of the existing communication processing object and the violation processing object in different sentence types based on the deviation of the proportion of the communication sentences of different sentence types;

[0067] Based on the average values ​​of the communication statement similarities of different statement types, the similarity between the existing communication processing object and different violation processing objects is determined.

[0068] It can be understood that the sentence types include cooperative sentence types, emotionally intense and uncooperative sentence types, and emotionally stable and uncooperative sentence types.

[0069] In another possible embodiment, the method for determining the degree of similarity with different violation processing objects is:

[0070] Determine the communication feedback sentences of different sentence types at different historical communication times based on the historical communication feedback data, and determine the proportion of communication sentences of different sentence types at different historical communication times according to the proportion of communication feedback sentences of different sentence types;

[0071] Based on the deviation of the proportion of communication sentences of different sentence types, determine the similarity of communication sentences of different sentence types between the existing communication processing object and the violation processing object at different historical communication times; based on the average value of the similarity of communication sentences of different sentence types, determine the similarity of sentences between the existing communication processing object and different violation processing objects at different historical communication times;

[0072] The sentence similarity average is calculated based on the sentence similarity with the violation processing object at different historical communication times, and it is determined whether the sentence similarity average is greater than a preset similarity. If so, the preset similarity is used to determine the similarity between the existing communication processing object and the violation processing object. If not, proceed to the next step.

[0073] Using the sentence similarity, the historical communication times are divided into similar communication times and deviation communication times, and it is determined whether the similar communication times are within a preset communication times interval. If so, the preset similarity is used to determine the similarity between the existing communication processing object and the violation processing object. If not, proceed to the next step.

[0074] Determine the similarity impact weight values ​​of different similar communication times according to the occurrence time and the sentence similarity, and judge whether the sum of the similarity impact weight values ​​of different similar communication times meets the requirements. If so, determine the similarity between the existing communication processing object and the violation processing object using the preset similarity. If not, proceed to the next step;

[0075] The similarity impact weight values ​​of different deviation communication times are determined according to the occurrence time and the similarity of the sentences, and the similarity degree between the existing communication processing object and different violation processing objects is determined by using the sum of the similarity impact weight values ​​of the deviation communication times and the similarity times.

[0076] Specifically, Figure 4 As shown, the method for determining the quality inspection strategy of the seat personnel is:

[0077] Determine the matching quality inspection strategy of the agent according to the quality inspection strategy corresponding to the probability of illegal communication, and use it as the matching quality inspection strategy;

[0078] Determine similar processing objects according to the degree of similarity with different violation processing objects, and use the proportion of the similar processing objects to the violation processing objects as the similar object proportion;

[0079] The object matching quality inspection strategy in the quality inspection strategy is determined by using the similar object ratio, and the quality inspection strategy of the agent is determined according to the object matching quality inspection strategy and the matching quality inspection strategy.

[0080] Further, determining the quality inspection strategy of the agent according to the object matching quality inspection strategy and the matching quality inspection strategy specifically includes:

[0081] The object matching quality inspection strategy and the quality inspection strategy with a higher quality inspection ratio and quality inspection requirement in the matching quality inspection strategy are used as the quality inspection strategy of the agent.

[0082] In another possible embodiment, the method for determining the quality inspection strategy of the seat personnel is:

[0083] Determine similar processing objects according to the degree of similarity with different violation processing objects, and use the proportion of the similar processing objects to the violation processing objects as the similar object proportion;

[0084] The proportion of similar objects is used as a compensation amount for the illegal communication probability of the illegal agent, and the illegal communication probability of the existing communication processing object is determined by using the compensation amount of the illegal communication probability and the sum of the illegal communication probability;

[0085] The quality inspection strategy corresponding to the violation probability is used as the quality inspection strategy of the agent when processing the existing communication processing object.

[0086] Furthermore, the quality inspection strategy for the agent includes a quality inspection ratio, quality inspection requirements for different violations, and corresponding intervention and processing strategies.

[0087] It should be noted that the access processing strategy includes suspending communication processing, outputting compliance reminders, and increasing the quality inspection ratio.

[0088] In another possible embodiment, the method for determining the quality inspection strategy of the seat personnel is:

[0089] Obtain the communication violation probability of the agent, and determine whether the communication violation probability of the agent is greater than a preset probability. If so, determine the quality inspection strategy of the agent based on the preset quality inspection strategy. If not, proceed to the next step.

[0090] Determine similar processing objects by the degree of similarity with different violation processing objects, and judge whether the number of similar processing objects is greater than the preset number of objects. If so, determine the quality inspection strategy of the seat personnel based on the preset quality inspection strategy. If not, proceed to the next step.

[0091] The ratio of the similar processing objects to the illegal processing objects is taken as the similar object ratio, and the illegal object similarity is determined according to the number of similar processing objects and the similarity degree of different similar processing objects, and whether the illegal object similarity meets the requirements is judged, if so, proceed to the next step, if not, determine the quality inspection strategy of the seat personnel based on the preset quality inspection strategy;

[0092] The proportion of similar objects is used as a compensation amount for the illegal communication probability of the illegal agent, and the illegal communication probability of the existing communication processing object is determined by using the compensation amount of the illegal communication probability and the sum of the illegal communication probability;

[0093] The quality inspection strategy corresponding to the violation probability is used as the quality inspection strategy of the agent when processing the existing communication processing object.

[0094] S3, when performing communication processing on the existing communication processing object, performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection result;

[0095] Specifically, the quality inspection results include the number of occurrences of different types of violations.

[0096] S4 uses the quality inspection results to determine the identification results and changes of the non-compliant behaviors of the seat personnel, and determines the intervention and processing strategy of the seat personnel according to the deviations between the identification results, changes and the quality inspection requirements of the quality inspection strategy.

[0097] It should be noted that the method for determining the intervention handling strategy of the seat staff is:

[0098] Determine the number of occurrences of different types of violations based on the identification results, and determine the severity of the violations based on the deviation between the number of occurrences and the quality inspection requirements of the quality inspection strategy;

[0099] Divide the current communication data of the seat staff into multiple time periods, obtain the change in the number of violations between different adjacent time periods, determine the screening of the time period when the violations are getting worse based on the change, and determine the violation change problem value based on the number ratio of the time period when the violations are getting worse;

[0100] The weights of the violation severity and the violation change problem value are used to determine the intervention processing problem value of the seat personnel, and the intervention processing strategy of the seat personnel is determined according to the intervention processing problem value.

[0101] Further, determining the intervention handling strategy of the agent according to the intervention handling problem value specifically includes:

[0102] The intervention processing problem value is used to determine the violation processing type of the agent, and the intervention processing strategy of the agent is determined using the violation processing type.

[0103] Example 2 On the other hand, if Figure 5 As shown, the present invention provides an intelligent strategy quality inspection platform based on real-time ASR voice stream, which adopts the above-mentioned intelligent strategy quality inspection method based on real-time ASR voice stream, specifically comprising:

[0104] Violation probability assessment module, quality inspection strategy output module, test result output module, intervention processing module;

[0105] The violation probability assessment module is responsible for obtaining the historical communication data of the seat personnel, using the historical communication data to identify the violation behaviors at different historical communication times to obtain the identification results, and determining the violation communication probability of the seat personnel based on the identification results;

[0106] The quality inspection strategy output module is responsible for taking the communication processing object in which the agent has violated the rules as the violation processing object, and using the historical communication feedback data of the existing communication processing object of the agent as the basic data to determine the similarity with different violation processing objects, and determining the quality inspection strategy of the agent based on the similarity and the probability of violation communication;

[0107] The detection result output module is responsible for performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection result when performing communication processing on the existing communication processing object;

[0108] The intervention processing module is responsible for using the quality inspection results to determine the identification results and changes of the non-compliant behaviors of the seat personnel, and determining the intervention processing strategy of the seat personnel based on the deviations between the identification results, changes and the quality inspection requirements of the quality inspection strategy.

[0109] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0110] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. An intelligent strategy quality inspection method based on real-time ASR voice stream, characterized in that: Specifically include: Acquire historical communication data of the agent, use the historical communication data to identify illegal behaviors at different historical communication times to obtain an identification result, and determine the probability of illegal communication of the agent based on the identification result; The communication processing object in which the agent has violated the rules is taken as the violation processing object, and the historical communication feedback data of the existing communication processing object of the agent is used as the basic data to determine the similarity with different violation processing objects, and the quality inspection strategy of the agent is determined based on the similarity and the probability of violation communication; When performing communication processing on the existing communication processing object, performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection result; Determine the identification result and change of the non-compliant behavior of the agent by using the quality inspection test result, and determine the intervention strategy of the agent by the deviation between the identification result, the change and the quality inspection requirement of the quality inspection strategy; The method for determining the probability of illegal communication of the seat personnel is: The number of historical communications with violations is used as the number of illegal communications, the violation weight values ​​of different types of violations are obtained, and the communication problem values ​​of different illegal communication numbers are determined based on the number of occurrences of different types of violations; Determine the basic probability of illegal communication of the agent by the proportion of illegal communication times in the historical communication times, and determine the compensation amount of illegal communication probability by the average value of communication problem values ​​of different illegal communication times; Determining the communication violation probability of the agent by using the sum of the violation probability compensation amount and the basic communication violation probability; The method for determining the quality inspection strategy of the seat personnel is: Determine the matching quality inspection strategy of the agent according to the quality inspection strategy corresponding to the probability of illegal communication, and use it as the matching quality inspection strategy; Determine similar processing objects according to the degree of similarity with different violation processing objects, and use the proportion of the similar processing objects to the violation processing objects as the similar object proportion; Determine the object matching quality inspection strategy in the quality inspection strategy by using the similar object ratio, and determine the quality inspection strategy of the agent according to the object matching quality inspection strategy and the matching quality inspection strategy; Determining the quality inspection strategy of the agent according to the object matching quality inspection strategy and the matching quality inspection strategy specifically includes: The object matching quality inspection strategy and the quality inspection strategy with a higher quality inspection ratio and quality inspection requirement in the matching quality inspection strategy are used as the quality inspection strategy of the agent.

2. The intelligent strategy quality inspection method based on real-time ASR voice stream according to claim 1, characterized in that: The historical communication data of the agent is determined according to the communication records of the agent in processing different communication processing objects.

3. The intelligent strategy quality inspection method based on real-time ASR voice stream according to claim 1, characterized in that: The identification result includes the number of occurrences of different types of violations of the agent.

4. The intelligent strategy quality inspection method based on real-time ASR voice stream according to claim 1, characterized in that: The violation weight value of the violation behavior is determined according to a preset corresponding relationship between the type of the violation behavior and the violation weight value.

5. The intelligent strategy quality inspection method based on real-time ASR voice stream according to claim 1, characterized in that: The method for determining the degree of similarity with different violation processing objects is: Determine communication feedback sentences of different sentence types based on historical communication feedback data, and determine the proportion of communication sentences of different sentence types according to the proportion of communication feedback sentences of different sentence types; Determine the similarity of the communication sentences of the existing communication processing object and the violation processing object in different sentence types based on the deviation of the proportion of the communication sentences of different sentence types; Based on the average values ​​of the communication statement similarities of different statement types, the similarity between the existing communication processing object and different violation processing objects is determined.

6. The intelligent strategy quality inspection method based on real-time ASR voice stream according to claim 5, characterized in that: The sentence types include cooperative sentence types, emotionally intense and uncooperative sentence types, and emotionally stable and uncooperative sentence types.

7. An intelligent policy quality inspection platform based on real-time ASR voice stream, using an intelligent policy quality inspection method based on real-time ASR voice stream according to any one of claims 1 to 6, characterized in that: Specifically include: Violation probability assessment module, quality inspection strategy output module, test result output module, intervention processing module; The violation probability assessment module is responsible for obtaining the historical communication data of the seat personnel, using the historical communication data to identify the violation behaviors at different historical communication times to obtain the identification results, and determining the violation communication probability of the seat personnel based on the identification results; The quality inspection strategy output module is responsible for taking the communication processing object in which the agent has violated the rules as the violation processing object, and using the historical communication feedback data of the existing communication processing object of the agent as the basic data to determine the similarity with different violation processing objects, and determining the quality inspection strategy of the agent based on the similarity and the probability of violation communication; The detection result output module is responsible for performing quality inspection processing on the voice stream of the agent according to the quality inspection ratio and ASR corresponding to the quality inspection strategy to obtain a quality inspection result when performing communication processing on the existing communication processing object; The intervention processing module is responsible for using the quality inspection results to determine the identification results and changes of the non-compliant behaviors of the seat personnel, and determining the intervention processing strategy of the seat personnel based on the deviations between the identification results, changes and the quality inspection requirements of the quality inspection strategy.

Images