Call quality inspection method and device, electronic equipment and storage medium

Abstract

The invention discloses a call quality inspection method and device, electronic equipment and a storage medium. The method relates to the technical field of financial science and technology and artificial intelligence data processing, and comprises the following steps: obtaining a current call to be subjected to quality inspection, and segmenting the current call to be subjected to quality inspection through a preset call time axis segmentation method to obtain a segmentation result of the current call to be subjected to quality inspection; using a preset staging rule cue word detection algorithm to respectively and sequentially detect the first segment of current call to be subjected to quality inspection, the second segment of current call to be subjected to quality inspection and the current call to be subjected to quality inspection, and generating a target call quality inspection result; and feeding back the target call quality inspection result to the user in real time. The problems of low call quality inspection efficiency, low accuracy, poor flexibility and poor comprehensiveness caused by manual sampling inspection or rule engine detection are solved, the call quality inspection efficiency and accuracy are improved, real-time quality inspection is realized, and the call quality inspection comprehensiveness and flexibility are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of financial technology and artificial intelligence data processing, and in particular to a call quality inspection method and device, electronic equipment and storage medium. BACKGROUND

[0002] In the field of financial technology and artificial intelligence, for post-loan management scenarios of banks, consumer finance or Internet finance institutions, it is necessary to accurately inspect the calls between customer service and customers, specifically, the calls for collection, so as to better serve customers or banks and other institutions.

[0003] The inventor found that the prior art has the following defects in the process of implementing the present application: Currently, the calls between customer service and customers are generally inspected by manual sampling methods. The sampled call recordings are checked against each bank call quality inspection rule, and the violation points are recorded to form a violation statistics record table. Or through a rule engine detection, the call recording needs to be converted into text, and then the potential violation content is identified through keyword matching or regular expression. This will result in low call quality inspection efficiency, low accuracy and poor comprehensiveness. SUMMARY

[0004] The present application provides a call quality inspection method, device, electronic equipment and storage medium to improve the efficiency, accuracy, comprehensiveness and flexibility of call quality inspection.

[0005] According to one aspect of the present application, a call quality inspection method is provided, comprising:

[0006] Obtaining a current call to be inspected, and segmenting the current call to be inspected by a preset call time axis segmentation method to obtain a current call to be inspected segmentation result;

[0007] The current call to be inspected segmentation result includes a first segment of the current call to be inspected and a second segment of the current call to be inspected.

[0008] Using a pre-set phased rule prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected and the current call to be inspected, respectively, to generate a target call quality inspection result;

[0009] The phased rule prompt word detection algorithm includes at least one of the following: a first phase rule prompt word detection algorithm, a second phase rule prompt word detection algorithm and a third phase rule prompt word detection algorithm.

[0010] Real-time feedback of the target call quality inspection result to the user.

[0011] According to another aspect of the present application, a call quality inspection device is provided, comprising:

[0012] A current call segment result determination module is configured to acquire a current call to be inspected, and segment the current call to be inspected by using a preset call time axis segmentation method to obtain a current call segment result.

[0013] The current call segment result comprises a first segment of the current call to be inspected and a second segment of the current call to be inspected.

[0014] A target call quality inspection result generation module is configured to use a preset staged rule prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected and the current call to be inspected respectively to generate a target call quality inspection result.

[0015] The staged rule prompt word detection algorithm comprises at least one of a first stage rule prompt word detection algorithm, a second stage rule prompt word detection algorithm and a third stage rule prompt word detection algorithm.

[0016] A target call quality inspection result feedback module is configured to feed back the target call quality inspection result to a user in real time.

[0017] According to another aspect of the present application, an electronic device is provided, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the call quality inspection method of any of the embodiments of the present application when executing the computer program.

[0018] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for causing a processor to implement the call quality inspection method of any of the embodiments of the present application when executed.

[0019] The technical solution of the embodiments of the present application acquires a current call to be inspected, segments the current call to be inspected by using a preset call time axis segmentation method to obtain a current call segment result, uses a preset staged rule prompt word detection algorithm to sequentially detect a first segment of the current call to be inspected, a second segment of the current call to be inspected and the current call to be inspected respectively to generate a target call quality inspection result, and feeds back the target call quality inspection result to a user in real time. The technical solution solves the problems of low call quality inspection efficiency, low accuracy, poor flexibility and poor comprehensiveness caused by manual inspection or rule engine detection, improves the efficiency and accuracy of call quality inspection, realizes real-time inspection, and improves the comprehensiveness and flexibility of call quality inspection.

[0020] It is to be understood that the embodiments described herein are merely exemplary of the application and that a person skilled in the art can devise other embodiments without departing from the scope of the present application. It is also to be understood that not all of the benefits described herein need necessarily be realized in any particular embodiment of the application. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort based on these drawings.

[0022] Figure 1 is a flow chart of a call quality inspection method according to an embodiment of the present application;

[0023] Figure 2 is a structural schematic diagram of a call quality inspection device according to an embodiment of the present application;

[0024] Figure 3 is a structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to make the person skilled in the art better understand the present application, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort should be within the scope of the present application.

[0026] It should be noted that the terms "target", "current" and the like in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0027] It is worth noting that in the technical solution of the present application, the collected information is information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards of relevant countries and regions, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for user to choose authorization or refusal; if the user chooses to refuse, enter the expert decision-making process.

[0028] Embodiment one

[0029] Figure 1 A flowchart of a call quality inspection method is provided for embodiment one of the present application. The present embodiment is suitable for detecting abnormalities in real-time calls between customer service and users. The method can be performed by a call quality inspection device, which can be realized in the form of hardware and / or software.

[0030] Correspondingly, as shown in Figure 1 , the method comprises:

[0031] S110, acquiring a current call to be inspected, and performing segmentation processing on the current call to be inspected by a preset call time axis segmentation method to obtain a current call to be inspected segmentation result.

[0032] The current call to be inspected segmentation result includes a first segment of the current call to be inspected and a second segment of the current call to be inspected.

[0033] The current call to be inspected can be obtained in real time or periodically, and the call time length corresponding to each current call to be inspected is different. The call time axis segmentation method can be a method of dividing the call to be inspected into two calls with different time lengths according to the call time.

[0034] The first segment of the current call to be inspected can be the call between the start time and the time axis segmentation time threshold. The second segment of the current call to be inspected is the call between the time axis segmentation time threshold and the end time.

[0035] In the present embodiment, each call to be inspected obtained in real time is inspected by the call quality inspection system, which can realize full-automatic inspection of the call. This can ensure that each call can be inspected for compliance, reducing the risk of missed inspection and the cost of manual review.

[0036] Specifically, the current call to be inspected is segmented by using the preset call time axis segmentation method to obtain a current call to be inspected segmentation result, including: obtaining a time axis segmentation time threshold, and determining whether the current call time length corresponding to the current call to be inspected is greater than the time axis segmentation time threshold; if yes, the call time axis segmentation method is used to segment the current call to be inspected according to the time axis segmentation time threshold to determine a first segment of the current call to be inspected and a second segment of the current call to be inspected; if no, the current call time length abnormal information is fed back to the user.

[0037] The time axis segmentation time threshold can be a preset time threshold for time division of the call.

[0038] For example, it is assumed that the current call time length corresponding to the current call to be inspected is 2 minutes, and the time axis segmentation time threshold is 30 seconds. Since 2 minutes is greater than 30 seconds, the call time axis segmentation method is used to segment the current call to be inspected. Specifically, the call from 0 second to 30 seconds can be divided into the first segment of the current call to be inspected, and the call from 31 seconds to 2 minutes can be divided into the second segment of the current call to be inspected.

[0039] It is assumed that the current call time length corresponding to the current call to be inspected is 20 seconds. It can be determined that the current call time length corresponding to the current call to be inspected is not greater than the time axis segmentation time threshold, so that it is indicated that the call time length of the current call does not meet the requirement, and the current call time length abnormal information is fed back to the user.

[0040] The advantage of such setting is that the size relationship between the current call time length and the time axis segmentation time threshold is determined, and then the call is segmented by using the call time axis segmentation method to obtain the corresponding call segmentation processing result. In this way, the current call to be inspected whose call time length does not meet the requirement can be directly filtered out, and the analysis by the stage rule prompt word detection algorithm is not needed, so that the efficiency of call processing can be improved.

[0041] S120, using a pre-set stage rule prompt word detection algorithm, sequentially detecting the first segment of the current call to be inspected, the second segment of the current call to be inspected and the current call to be inspected respectively to generate a target call inspection result.

[0042] The stage rule prompt word detection algorithm includes at least one of the following: a first stage rule prompt word detection algorithm, a second stage rule prompt word detection algorithm and a third stage rule prompt word detection algorithm.

[0043] The first period of conversation can be detected by the first-stage rule prompt word detection algorithm; the second period of conversation can be detected by the second-stage rule prompt word detection algorithm; and the third period of conversation can be detected by the third-stage rule prompt word detection algorithm.

[0044] Optionally, the pre-set staged rule prompt word detection algorithm is used to sequentially detect the first period of current conversation to be inspected, the second period of current conversation to be inspected, and the current conversation to be inspected, and generate a target conversation inspection result, including: using the first-stage rule prompt word detection algorithm to detect the first period of current conversation to be inspected, and obtaining a first period of conversation inspection result; using the second-stage rule prompt word detection algorithm to detect the second period of current conversation to be inspected, and obtaining a second period of conversation inspection result; using the third-stage rule prompt word detection algorithm to detect the current conversation to be inspected, and obtaining a third period of conversation inspection result; and integrating the first period of conversation inspection result, the second period of conversation inspection result, and the third period of conversation inspection result to generate the target conversation inspection result.

[0045] In this embodiment, the first-stage rule prompt word detection algorithm can be used to detect the first period of current conversation to be inspected, the second-stage rule prompt word detection algorithm can be used to detect the second period of current conversation to be inspected, and the third-stage rule prompt word detection algorithm can be used to detect the current conversation to be inspected. The determined first period of conversation inspection result, the second period of conversation inspection result, and the third period of conversation inspection result can be integrated and processed, and then the target conversation inspection result can be generated.

[0046] The advantage of this setting is that the current conversation to be inspected is divided into three independent analyses, different rule prompt word detection algorithms are matched to generate corresponding conversation inspection results, which can improve the inspection accuracy and accuracy of the conversation, and can reduce the load of the inspection processing model.

[0047] Specifically, the step of using the first-stage rule prompt word detection algorithm to detect the first current call to be inspected and obtain the first call quality inspection result includes: using the first-stage rule prompt word detection algorithm to extract keywords from the first current call to be inspected, obtaining at least one current first extracted keyword; obtaining basic information specified keywords corresponding to the first-stage rule prompt word detection algorithm; wherein, the basic information specified keywords include the current organization employee identity, the current organization name, the current user identity, and the current user's required execution items; determining whether each of the current first extracted keywords contains all the basic information specified keywords, if so, then determining the first call quality inspection result as the first call quality inspection passed result; if not, then marking at least one missing keyword item, determining the first call quality inspection result as the first call quality inspection failed result, and feeding back the missing keyword item marking information corresponding to each of the missing keyword items to the user.

[0048] For example, if the current call to be inspected is a collection call to be inspected, this call is obtained with user permission to better inspect collection calls that are not in accordance with regulations, thereby improving the standardization of collection calls. Accordingly, based on the current collection call to be inspected, a first segment and a second segment of the current call to be inspected can be obtained. Specifically, the first segment of the current call to be inspected needs to be converted into text, and then keywords need to be extracted. The text of the first segment of the current call to be inspected mainly checks the compliance of the opening remarks.

[0049] After training the first-stage rule-based keyword detection algorithm, keywords can be assigned to match the corresponding basic information. Specifically, these can include the current employee's identity, the current organization's name, the current user's identity (in a collection call, the current user's identity corresponds to the current borrower's identity), and the user's required actions (in a collection call, the user's required actions correspond to the current borrower's required repayment amount and deadline). In other words, the first call to be inspected must contain all of these keywords to pass the quality inspection. Conversely, if one or more keywords are missing, it indicates that the collector did not provide all the necessary information at the beginning of the communication with the user, resulting in a failed quality inspection for the first call. The missing keyword information needs to be marked and fed back to the user.

[0050] For example, it needs to be determined whether the collection agent identified themselves as an employee of their organization and stated the organization's name within the first 30 seconds, whether they verified the borrower's identity, and whether they explained the required repayment amount and term. If any of these are missing, it needs to be marked as a violation, meaning the first call fails the quality inspection. Conversely, if all of these are present, it is confirmed that the first call currently under quality inspection does not contain any violations.

[0051] The advantage of this setup is that it compares the keywords specified by the basic information corresponding to the first-stage rule-based keyword detection algorithm with the extracted first keyword to generate the corresponding first call quality inspection result. This allows for a comprehensive quality inspection of the first call, resulting in a more accurate first call quality inspection result.

[0052] Specifically, the step of using the second-stage rule prompt word detection algorithm to detect the second segment of the current call to be inspected and obtaining the second segment call quality inspection result includes: using the second-stage rule prompt word detection algorithm to extract keywords from the second segment of the current call to be inspected, obtaining at least one current second extracted keyword; obtaining third-party information specified keywords corresponding to the second-stage rule prompt word detection algorithm; wherein, the third-party information specified keywords include third-party authorization letters and third-party contact information; determining whether each of the current second extracted keywords contains any third-party information specified keywords; if not, determining the second segment call quality inspection result as a second segment call quality inspection pass result; if yes, determining the second segment call quality inspection result as a second segment call quality inspection fail result, and feeding back the violation reference rule clause information corresponding to the third-party information specified keywords to the user.

[0053] In this embodiment, the second segment of the current call to be inspected first needs to be converted into text, and then keyword extraction is performed. The text of the second segment of the current call to be inspected mainly detects rules related to third-party information, such as authorization verification when someone other than the borrower is involved.

[0054] After training the second-stage rule-based keyword detection algorithm, it can be matched with corresponding third-party information keywords. Specifically, this can include third-party authorization letters and third-party contact information. In other words, the second call to be inspected needs to be checked to see if it contains any of these third-party information keywords, i.e., at least one of the following: third-party authorization letters or third-party contact information. If not, the second call passes the quality inspection; otherwise, it fails, and the user needs to be informed of the relevant violation rules.

[0055] For example, it's necessary to determine whether the collection agent's call, from the 30-second mark until its end, involved any keywords related to third-party information. For instance, if the caller is not the borrower, were they asked to provide authorization from a third party? Did they disclose third-party contact information? If so, the call is abnormal, and the relevant violation rules need to be reported to the user. Specifically, if the call involves third-party contact information, the relevant violation rules can be matched and fed back to the user. Conversely, if no violation is found, it's confirmed that the current second call under inspection does not violate any rules.

[0056] The advantages of this setup are as follows: By comparing the keywords specified in the third-party information corresponding to the second-stage rule-based keyword detection algorithm with the extracted second keywords, a corresponding second-stage call quality inspection result is generated. This allows for a comprehensive quality inspection of the second-stage call, resulting in more accurate results. Providing users with information on the violation rules and regulations provides a clear and detailed understanding of the reasons for the violation, improving the user experience and enriching the violation feedback information.

[0057] Specifically, the step of using the third-stage rule prompt word detection algorithm to detect the current call to be inspected and obtain the third call quality inspection result includes: using the third-stage rule prompt word detection algorithm to extract keywords from the current call to be inspected and obtain at least one current third extracted keyword; obtaining the comprehensive detection information specified keywords corresponding to the third-stage rule prompt word detection algorithm; wherein, the comprehensive detection information specified keywords include keywords related to false promises and keywords related to abnormal attitude language; determining whether each of the current third extracted keywords contains any of the comprehensive detection information specified keywords; if not, determining the third call quality inspection result as a third call quality inspection pass result; if not, determining the third call quality inspection result as a third call quality inspection fail result, and feeding back at least one abnormal call segment corresponding to the comprehensive detection information specified keywords to the user.

[0058] In this embodiment, the first step is to convert all current calls awaiting quality inspection into text. After conversion, keyword extraction is then performed. The main purpose of the text of the current calls awaiting quality inspection is to detect whether any violations exist in the entire call.

[0059] After training the third-stage rule-based prompt word detection algorithm, the algorithm can be assigned corresponding keywords from the comprehensive detection information. Specifically, these can include keywords related to false promises and keywords related to abnormal attitudes and language. That is, in all current calls awaiting quality inspection, it needs to be determined whether any one of the keywords specified in the comprehensive detection information is present, i.e., at least one of the keywords related to false promises and keywords related to abnormal attitudes and language. If none are present, the third call quality inspection result is determined to be passed; otherwise, it is a failure result, and at least one abnormal call segment corresponding to the specified keywords in the comprehensive detection information needs to be fed back to the user.

[0060] Among them, abnormal call segments can be either timestamp information of the call or text range information corresponding to the call.

[0061] For example, it's necessary to determine whether any keywords specified in the comprehensive inspection information are involved in the current call made by the collection agent. For instance, the presence of false promises and unusual language; specifically, unusual language can include threatening, intimidating, and insulting language. If these are present, it indicates an anomaly in the call, and one or more abnormal call segments need to be reported to the user. Specifically, when the call involves keywords related to false promises, at least one corresponding abnormal call segment can be matched and fed back to the user. Conversely, if no such keywords are found, it confirms that the current call under inspection is not abnormal.

[0062] The advantage of this setup is that it uses the comprehensive detection information specified by the third-stage rule-based keyword detection algorithm to compare the extracted keywords with the current third-stage extracted keywords, generating corresponding third-stage call quality inspection results. This allows for comprehensive quality inspection of all calls, accurately generating third-stage call quality inspection results; and by providing feedback on each abnormal call segment to the user, it facilitates better learning and rectification.

[0063] Optionally, before using a pre-set phased rule-based prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected, and the current call to be inspected, and generating the target call quality inspection result, the method further includes: obtaining each historical standard financial regulatory call; segmenting each historical standard financial regulatory call to obtain the first segment of the historical standard financial regulatory call and the second segment of the historical standard financial regulatory call; and performing initial keyword extraction processing on each of the first segment of the historical standard financial regulatory call, the second segment of the historical standard financial regulatory call, and the historical standard financial regulatory call to obtain the keywords specified in the historical standard basic information and the historical standard... Keywords specified in the three-party information and keywords specified in the historical standard comprehensive detection information; using keywords specified in the first segment of historical standard financial regulatory calls and historical standard basic information, keywords specified in the second segment of historical standard financial regulatory calls and historical standard third-party information, and keywords specified in the historical standard financial regulatory calls and historical standard comprehensive detection information, the models corresponding to the initial first-stage rule prompt word detection algorithm, the initial second-stage rule prompt word detection algorithm, and the initial third-stage rule prompt word detection algorithm are trained respectively until the joint detection index output by the model meets the requirements of the preset joint detection index threshold, thus obtaining the constructed phased rule prompt word detection algorithm.

[0064] In this embodiment, the model corresponding to the initial first-stage rule prompt word detection algorithm is trained using keywords specified by each first segment of historical standard financial regulatory calls and historical standard basic information; the model corresponding to the initial second-stage rule prompt word detection algorithm is trained using keywords specified by each second segment of historical standard financial regulatory calls and historical standard third-party information; and the model corresponding to the initial third-stage rule prompt word detection algorithm is trained using keywords specified by each historical standard financial regulatory calls and historical standard comprehensive detection information. This yields the initially trained first-stage rule prompt word detection algorithm, second-stage rule prompt word detection algorithm, and third-stage rule prompt word detection algorithm.

[0065] Furthermore, a test set verification operation is required, which involves randomly acquiring a labeled test set, which may include multiple call samples and manually labeled results. These multiple call samples are then input into the initially trained first-stage rule-prompt word detection algorithm, second-stage rule-prompt word detection algorithm, and third-stage rule-prompt word detection algorithm model, respectively. This generates corresponding results, which are then compared with the manually labeled results to calculate a joint detection metric. This joint detection metric can include accuracy, false negative rate, and false positive rate. Specifically, accuracy is calculated based on the consistency between the judgment results and the manually labeled results; the false negative rate is the proportion of genuine violations not detected by the model; and the false positive rate is the proportion of compliant content incorrectly identified by the model.

[0066] Accordingly, the calculated accuracy, false negative rate, and false positive rate can be compared with the joint detection index threshold. If all requirements are met, the first-stage rule-provided word detection algorithm, the second-stage rule-provided word detection algorithm, and the third-stage rule-provided word detection algorithm are considered successfully trained, thus completing the phased rule-provided word detection algorithm. Conversely, if any one of them is not met, attribution analysis of the erroneous cases is required, and the keywords in the model should be optimized based on the analysis results. The process then returns to the testing stage until the joint detection index output by the model meets the preset joint detection index threshold requirements.

[0067] The advantage of this setup is that by using historical standard financial regulatory calls to train and build the phased rule prompt word detection algorithm, a more accurate phased rule prompt word detection algorithm can be trained. The trained phased rule prompt word detection algorithm can be used to perform automated multi-stage call quality inspection operations, improving the convenience and efficiency of call quality inspection.

[0068] S130. The target call quality inspection results are fed back to the user in real time.

[0069] In this embodiment, the determined first call quality inspection results, second call quality inspection results, and third call quality inspection results can be integrated and processed to generate the target call quality inspection result, and the multi-stage results can be fed back to the user in real time.

[0070] The technical solution of this invention obtains the current call to be inspected and segments it using a preset call timeline segmentation method to obtain segmentation results. A pre-set phased rule-based prompt word detection algorithm is then used to sequentially detect the first segment, the second segment, and the current segment of the call to be inspected, generating a target call quality inspection result. This target call quality inspection result is then fed back to the user in real time. This solves the problems of low efficiency, low accuracy, poor flexibility, and poor comprehensiveness in call quality inspection caused by manual sampling or rule engine detection, improving the efficiency and accuracy of call quality inspection, achieving real-time quality inspection, and enhancing the comprehensiveness and flexibility of call quality inspection.

[0071] Example 2

[0072] Figure 2 This is a schematic diagram of a call quality inspection device provided in Embodiment 2 of the present invention. The call quality inspection device provided in this embodiment can be implemented by software and / or hardware, and can be configured in a terminal device or server to implement a call quality inspection method according to an embodiment of the present invention. Figure 2As shown, the device includes: a current call segmentation result determination module 210, a target call quality inspection result generation module 220, and a target call quality inspection result feedback module 230.

[0073] The current call segmentation result determination module 210 is used to obtain the current call to be inspected and to segment the current call to be inspected using a preset call time axis segmentation method to obtain the current call segmentation result.

[0074] The segmentation result of the current call to be inspected includes the first segment of the current call to be inspected and the second segment of the current call to be inspected.

[0075] The target call quality inspection result generation module 220 is used to use a pre-set phased rule prompt word detection algorithm to sequentially detect the first current call to be inspected, the second current call to be inspected, and the current call to be inspected, and generate the target call quality inspection result.

[0076] The phased rule prompt word detection algorithm includes at least one of the following: a first-stage rule prompt word detection algorithm, a second-stage rule prompt word detection algorithm, and a third-stage rule prompt word detection algorithm;

[0077] The target call quality inspection result feedback module 230 is used to provide real-time feedback of the target call quality inspection result to the user.

[0078] The technical solution of this invention obtains the current call to be inspected and segments it using a preset call timeline segmentation method to obtain segmentation results. A pre-set phased rule-based prompt word detection algorithm is then used to sequentially detect the first segment, the second segment, and the current segment of the call to be inspected, generating a target call quality inspection result. This target call quality inspection result is then fed back to the user in real time. This solves the problems of low efficiency, low accuracy, poor flexibility, and poor comprehensiveness in call quality inspection caused by manual sampling or rule engine detection, improving the efficiency and accuracy of call quality inspection, achieving real-time quality inspection, and enhancing the comprehensiveness and flexibility of call quality inspection.

[0079] Based on the above embodiments, the current call segmentation result determination module 210 can be specifically used to: obtain the time axis segmentation time threshold, and determine whether the current call duration corresponding to the current call to be inspected is greater than the time axis segmentation time threshold; if it is greater, then use the call time axis segmentation method to segment the current call to be inspected according to the time axis segmentation time threshold to determine the first segment of the current call to be inspected and the second segment of the current call to be inspected; if it is not greater, then feed back abnormal information about the current call duration to the user.

[0080] Based on the above embodiments, the target call quality inspection result generation module 220 may specifically include: a first call quality inspection result determination unit, which may be specifically used to: use the first-stage rule prompt word detection algorithm to detect the first current call to be inspected and obtain the first call quality inspection result; a second call quality inspection result determination unit, which may be specifically used to: use the second-stage rule prompt word detection algorithm to detect the second current call to be inspected and obtain the second call quality inspection result; a third call quality inspection result determination unit, which may be specifically used to: use the third-stage rule prompt word detection algorithm to detect the current call to be inspected and obtain the third call quality inspection result; and a target call quality inspection result generation unit, which may be specifically used to: integrate the first call quality inspection result, the second call quality inspection result, and the third call quality inspection result to generate the target call quality inspection result.

[0081] Based on the above embodiments, the first call quality inspection result determination unit can be specifically used to: use the first-stage rule prompt word detection algorithm to extract keywords from the first current call to be inspected, and obtain at least one current first extracted keyword; obtain basic information specified keywords corresponding to the first-stage rule prompt word detection algorithm; wherein, the basic information specified keywords include the current organization employee identity, the current organization name, the current user identity, and the current user's required execution items; determine whether each of the current first extracted keywords contains all the basic information specified keywords, and if so, determine that the first call quality inspection result is the first call quality inspection passed result; if not, mark at least one missing keyword item, determine that the first call quality inspection result is the first call quality inspection failed result, and feed back the missing keyword item marking information corresponding to each of the missing keyword items to the user.

[0082] Based on the above embodiments, the second call quality inspection result determination unit can be specifically used to: use the second-stage rule prompt word detection algorithm to extract keywords from the second current call to be inspected, and obtain at least one current second extracted keyword; obtain third-party information specified keywords corresponding to the second-stage rule prompt word detection algorithm; wherein, the third-party information specified keywords include third-party authorization letters and third-party contact information; determine whether each of the current second extracted keywords contains any third-party information specified keywords, if not, determine the second call quality inspection result as a second call quality inspection pass result; if yes, determine the second call quality inspection result as a second call quality inspection fail result, and feed back the violation reference rule clause information corresponding to the third-party information specified keywords to the user.

[0083] Based on the above embodiments, the third call quality inspection result determination unit can be specifically used to: use the third-stage rule prompt word detection algorithm to extract keywords from the current call to be inspected, and obtain at least one current third extracted keyword; obtain the comprehensive detection information specified keyword corresponding to the third-stage rule prompt word detection algorithm; wherein, the comprehensive detection information specified keyword includes false promise keywords and abnormal attitude language keywords; determine whether each of the current third extracted keywords contains any of the comprehensive detection information specified keywords, if not, determine the third call quality inspection result as a third call quality inspection pass result; if not, determine the third call quality inspection result as a third call quality inspection fail result, and feed back at least one abnormal call segment corresponding to the comprehensive detection information specified keyword to the user.

[0084] Based on the above embodiments, a phased rule-based prompt word detection algorithm construction module is also included, which can be specifically used to: obtain each historical standard financial regulatory call before using a pre-set phased rule-based prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected, and the current call to be inspected to generate the target call quality inspection result; segment each historical standard financial regulatory call to obtain the first segment of the historical standard financial regulatory call and the second segment of the historical standard financial regulatory call; and perform initial keyword extraction processing on each of the first segment of the historical standard financial regulatory call, the second segment of the historical standard financial regulatory call, and the historical standard financial regulatory call to obtain the historical standard Keywords are specified in the quasi-basic information, keywords in the historical standard third-party information, and keywords in the historical standard comprehensive testing information. The models corresponding to the initial first-stage rule prompt word detection algorithm, the initial second-stage rule prompt word detection algorithm, and the initial third-stage rule prompt word detection algorithm are trained using the keywords specified in the first, second, and third stages of historical standard financial regulatory calls, the second, and third stages of historical standard financial regulatory calls, and the comprehensive testing information, respectively, until the joint detection index output by the model meets the preset joint detection index threshold requirements, thus obtaining the completed phased rule prompt word detection algorithm.

[0085] The call quality inspection device provided in this embodiment of the invention can execute the call quality inspection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0086] Example 3

[0087] Figure 3A schematic diagram of an electronic device 10, which can be used to implement Embodiment 3 of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0088] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0089] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0090] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as call quality inspection methods.

[0091] In some embodiments, the call quality inspection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the call quality inspection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the call quality inspection method by any other suitable means (e.g., by means of firmware).

[0092] The method includes: acquiring the current call to be inspected, and segmenting the current call to be inspected using a preset call timeline segmentation method to obtain the segmentation result of the current call to be inspected; wherein the segmentation result of the current call to be inspected includes a first segment of the current call to be inspected and a second segment of the current call to be inspected; using a pre-set phased rule prompt word detection algorithm, sequentially detecting the first segment of the current call to be inspected, the second segment of the current call to be inspected, and the current call to be inspected, to generate a target call quality inspection result; wherein the phased rule prompt word detection algorithm includes at least one of the following: a first-stage rule prompt word detection algorithm, a second-stage rule prompt word detection algorithm, and a third-stage rule prompt word detection algorithm; and providing real-time feedback of the target call quality inspection result to the user.

[0093] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0094] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0095] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0096] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0097] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0098] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0099] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0100] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

[0101] Example 4

[0102] Embodiment 4 of the present invention also provides a computer-readable storage medium, wherein the computer-readable instructions, when executed by a computer processor, are used to perform a call quality inspection method. The method includes: acquiring a current call to be inspected, and segmenting the current call to be inspected using a preset call timeline segmentation method to obtain a segmentation result of the current call to be inspected; wherein the segmentation result of the current call to be inspected includes a first segment of the current call to be inspected and a second segment of the current call to be inspected; using a pre-set phased rule prompt word detection algorithm, sequentially detecting the first segment of the current call to be inspected, the second segment of the current call to be inspected, and the current call to be inspected, to generate a target call quality inspection result; wherein the phased rule prompt word detection algorithm includes at least one of the following: a first-stage rule prompt word detection algorithm, a second-stage rule prompt word detection algorithm, and a third-stage rule prompt word detection algorithm; and providing real-time feedback of the target call quality inspection result to the user.

[0103] Of course, the computer-readable storage medium provided in the embodiments of the present invention has computer-executable instructions that are not limited to the method operations described above, but can also perform related operations in call quality inspection provided in any embodiment of the present invention.

[0104] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0105] It is worth noting that in the above-mentioned call quality inspection embodiments, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0106] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for call quality inspection, the method comprising: The method comprises the following steps: acquiring a current call to be inspected, and performing segmentation processing on the current call to be inspected by using a preset call time axis segmentation method to obtain a current call to be inspected segmentation result; wherein the current call to be inspected segmentation result comprises a first segment of the current call to be inspected and a second segment of the current call to be inspected; using a preset staged rule prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected and the current call to be inspected respectively to generate a target call inspection result; wherein the staged rule prompt word detection algorithm comprises at least one of the following: a first stage rule prompt word detection algorithm, a second stage rule prompt word detection algorithm and a third stage rule prompt word detection algorithm; feeding back the target call inspection result to a user in real time.

2. The method of claim 1, wherein, The method of performing segmentation processing on the current call to be inspected by using a preset call time axis segmentation method to obtain a current call to be inspected segmentation result comprises the following steps: acquiring a time axis segmentation time threshold, and determining whether a current call time length corresponding to the current call to be inspected is greater than the time axis segmentation time threshold; if yes, performing segmentation processing on the current call to be inspected according to the time axis segmentation time threshold by using the call time axis segmentation method to determine the first segment of the current call to be inspected and the second segment of the current call to be inspected; if no, feeding back current call time length abnormal information to the user.

3. The method of claim 1, wherein, The method of using a preset staged rule prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected and the current call to be inspected respectively to generate a target call inspection result comprises the following steps: using the first stage rule prompt word detection algorithm to detect the first segment of the current call to be inspected to obtain a first segment call inspection result; using the second stage rule prompt word detection algorithm to detect the second segment of the current call to be inspected to obtain a second segment call inspection result; using the third stage rule prompt word detection algorithm to detect the current call to be inspected to obtain a third segment call inspection result; integrating the first segment call inspection result, the second segment call inspection result and the third segment call inspection result to generate the target call inspection result.

4. The method of claim 3, wherein, The method of using the first stage rule prompt word detection algorithm to detect the first segment of the current call to be inspected to obtain a first segment call inspection result comprises the following steps: using the first stage rule prompt word detection algorithm to extract keywords from the first segment of the current call to be inspected to obtain at least one current first extracted keyword; acquiring a basic information specified keyword corresponding to the first stage rule prompt word detection algorithm; wherein the basic information specified keyword comprises a current institution employee identity, a current institution name, a current user identity and a current user required to perform an affair; determining whether all the basic information specified keywords are contained in each of the current first extracted keywords, and if yes, determining that the first segment call inspection result is a first segment call inspection pass result; If not, at least one missing keyword item is marked, the first segment call quality inspection result is determined as a first segment call quality inspection failure result, and missing keyword item marking information corresponding to each missing keyword item is fed back to the user.

5. The method of claim 4, wherein, The second segment current call to be inspected is detected using the second-stage rule prompt keyword detection algorithm to obtain a second segment call quality inspection result, including: The second segment current call to be inspected is detected using the second-stage rule prompt keyword detection algorithm to obtain a second segment current keyword extraction result; Third-party information regulation keywords corresponding to the second-stage rule prompt keyword detection algorithm are obtained, wherein the third-party information regulation keywords include third-party provided authorization and third-party contact information; It is determined whether each of the second segment current keyword extraction result contains any one of the third-party information regulation keywords, and if not, the second segment call quality inspection result is determined as a second segment call quality inspection pass result; If yes, the second segment call quality inspection result is determined as a second segment call quality inspection failure result, and rule violation reference rule clause information corresponding to the third-party information regulation keywords is fed back to the user.

6. The method of claim 5, wherein, The third segment current call to be inspected is detected using the third-stage rule prompt keyword detection algorithm to obtain a third segment call quality inspection result, including: The third segment current call to be inspected is detected using the third-stage rule prompt keyword detection algorithm to obtain at least one third segment current keyword extraction result; Comprehensive detection information regulation keywords corresponding to the third-stage rule prompt keyword detection algorithm are obtained, wherein the comprehensive detection information regulation keywords include false promise type keywords and abnormal attitude language keywords; It is determined whether each of the third segment current keyword extraction result contains any one of the comprehensive detection information regulation keywords, and if not, the third segment call quality inspection result is determined as a third segment call quality inspection pass result; If not, the third segment call quality inspection result is determined as a third segment call quality inspection failure result, and at least one abnormal call segment corresponding to the comprehensive detection information regulation keywords is fed back to the user.

7. The method of claim 6, wherein, Before the first segment current call to be inspected, the second segment current call to be inspected, and the current call to be inspected are sequentially detected using the pre-set staged rule prompt keyword detection algorithm to generate a target call quality inspection result, it further includes: Each historical standard financial supervision call is obtained; Each of the historical standard financial supervision calls is segmented to obtain a first segment historical standard financial supervision call and a second segment historical standard financial supervision call; Each of the first segment historical standard financial supervision call, the second segment historical standard financial supervision call, and the historical standard financial supervision call is subjected to initial keyword extraction processing to obtain historical standard basic information regulation keywords, historical standard third-party information regulation keywords, and historical standard comprehensive detection information regulation keywords; The first segment of historical standard financial regulation conversation and historical standard basic information regulation keywords, the second segment of historical standard financial regulation conversation and historical standard third-party information regulation keywords, and the historical standard financial regulation conversation and historical standard comprehensive detection information regulation keywords are used to train the models corresponding to the initial first-stage rule prompt word detection algorithm, the initial second-stage rule prompt word detection algorithm, and the initial third-stage rule prompt word detection algorithm, respectively, until the joint detection index output by the model meets the requirements of the preset joint detection index threshold, and a constructed phased rule prompt word detection algorithm is obtained.

8. A call quality inspection apparatus characterized by comprising: Comprise: A current call to be inspected segment result determination module is configured to obtain a current call to be inspected, and segment the current call to be inspected by using a preset call time axis segmentation method to obtain a current call to be inspected segment result. The current call to be inspected segment result comprises a first segment of the current call to be inspected and a second segment of the current call to be inspected. A target call quality inspection result generation module is configured to use a pre-set phased rule prompt word detection algorithm to sequentially detect the first segment of the current call to be inspected, the second segment of the current call to be inspected, and the current call to be inspected, respectively, to generate a target call quality inspection result. The phased rule prompt word detection algorithm comprises at least one of the following: a first-stage rule prompt word detection algorithm, a second-stage rule prompt word detection algorithm, and a third-stage rule prompt word detection algorithm. A target call quality inspection result feedback module is configured to feed back the target call quality inspection result to a user in real time.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the call quality inspection method in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing the processor to execute when implementing the call quality inspection method in any one of claims 1-7.

Images