CN110163630B

CN110163630B - Product supervision method, device, computer equipment and storage medium

Info

Publication number: CN110163630B
Application number: CN201910300383.7A
Authority: CN
Inventors: 袁佳
Original assignee: Ping An Life Insurance Company of China Ltd
Current assignee: Ping An Life Insurance Company of China Ltd
Priority date: 2019-04-15
Filing date: 2019-04-15
Publication date: 2024-04-05
Anticipated expiration: 2039-04-15
Also published as: CN110163630A

Abstract

The invention discloses a product supervision method, a device, computer equipment and a storage medium, wherein the method comprises the steps of obtaining a product payment request sent by a user side; matching the seat identifier with the blacklist identifier to obtain a list matching result; if the list matching result is that the matching is successful, determining a corresponding product type according to the product identifier, calling corresponding voice broadcasting data according to the product type, and carrying out voice broadcasting on the voice broadcasting data; acquiring voice data to be recognized fed back by a user side based on voice broadcasting data, performing semantic analysis on the voice data to be recognized, and acquiring a semantic analysis result; extracting target voiceprint features from voice data to be recognized, and matching the standard voiceprint features with the target voiceprint features to obtain voiceprint matching results; and carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result so as to solve the problem that the seat has fraudulent conduct in the product selling process.

Description

Product supervision method, device, computer equipment and storage medium

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a method and apparatus for product supervision, a computer device, and a storage medium.

Background

With the development of global economy, large companies are increasingly competing, and there is also a competing relationship among internal teams or sales personnel. Due to insufficient supervision of companies, some sales personnel are for improving performance, users are induced to purchase products through false information and the like in the product selling process, or some team management personnel are for guaranteeing the selling performance of the team, fraud of the sales personnel is hidden, so that users can complain, the reputation of the companies is affected, and therefore, how to avoid fraud of sales agents becomes a problem to be solved urgently.

Disclosure of Invention

The embodiment of the invention provides a product supervision method, a device, computer equipment and a storage medium, which are used for solving the problem that an agent has fraudulent conduct in the product sales process.

A method of product supervision, comprising:

acquiring a product payment request sent by a user side, wherein the product payment request comprises a user identifier, a product identifier and an agent identifier;

matching the agent identifier with a blacklist identifier in a blacklist list to obtain a list matching result;

if the list matching result is that the matching is successful, determining a corresponding product type according to the product identifier, calling corresponding voice broadcasting data according to the product type, and performing voice broadcasting on the voice broadcasting data by adopting a TTS technology;

Acquiring voice data to be recognized fed back by a user side based on the voice broadcasting data, and performing semantic analysis on the voice data to be recognized to acquire a semantic analysis result;

extracting target voiceprint features from the voice data to be identified, and matching the standard voiceprint features corresponding to the user identification with the target voiceprint features to obtain a voiceprint matching result;

and carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result.

A product supervision apparatus comprising:

the payment request acquisition module is used for acquiring a product payment request sent by a user side, wherein the product payment request comprises a user identifier, a product identifier and an agent identifier;

the list matching module is used for matching the agent identifier with a blacklist identifier in a blacklist list to obtain a list matching result;

the voice broadcasting module is used for determining a corresponding product type according to the product identifier if the list matching result is that the matching is successful, calling corresponding voice broadcasting data according to the product type, and performing voice broadcasting on the voice broadcasting data by adopting a TTS technology;

The semantic analysis result acquisition module is used for acquiring voice data to be identified, which is fed back by the user side based on the voice broadcast data, carrying out semantic analysis on the voice data to be identified, and acquiring a semantic analysis result;

the voiceprint matching result acquisition module is used for extracting target voiceprint features from the voice data to be identified, matching the standard voiceprint features corresponding to the user identification with the target voiceprint features, and acquiring a voiceprint matching result;

and the response operation module is used for carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result.

A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the product supervision method described above when executing the computer program.

A computer readable storage medium storing a computer program which when executed by a processor implements the product supervision method described above.

The above provides a method, apparatus, computer device and storage medium for product supervision, when obtaining a product payment request sent by a user, matching an agent identifier with a blacklist identifier in a blacklist to determine whether to perform voice verification on the product payment request. If the list matching result is that the matching is successful, corresponding voice broadcasting data are determined according to the product identification, and the voice broadcasting data are subjected to voice broadcasting to determine whether a user clearly knows the product purchase information of the purchased product. The voice data to be recognized fed back by the user side based on the voice broadcasting data is acquired, semantic analysis is carried out on the voice data to be recognized, whether the user knows about the purchasing of the product clearly or not is acquired, and follow-up user complaints are reduced. And extracting target voiceprint features from the voice data to be identified, matching the standard voiceprint features corresponding to the user identification with the target voiceprint features, and obtaining a voiceprint matching result to realize whether the user replies. And carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result, and realizing supervision of product sales.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic view of an application environment of a product monitoring method according to an embodiment of the present invention;

FIG. 2 is a flow chart of a method of product supervision in an embodiment of the invention;

FIG. 3 is a flow chart of a method of product supervision in an embodiment of the invention;

FIG. 4 is a flow chart of a method of product supervision in an embodiment of the invention;

FIG. 5 is a flow chart of a method of product supervision in an embodiment of the invention;

FIG. 6 is a flow chart of a method of product supervision in an embodiment of the invention;

FIG. 7 is a flow chart of a method of product supervision in an embodiment of the invention;

FIG. 8 is a functional block diagram of a product monitoring device in an embodiment of the invention;

FIG. 9 is a schematic diagram of a computer device in accordance with an embodiment of the invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The product supervision method provided by the embodiment of the invention can be applied to an application environment as shown in fig. 1, and the product supervision method is applied to a product sales system, wherein the product sales system comprises a user side and a server side, and the user side communicates with the server side through a network. The product supervision method is specifically applied to a server side of a product recommendation APP, when a product payment request sent by a user side is received, whether an agent identifier selling the product is matched with a blacklist identifier is judged, if so, corresponding voice broadcast data are broadcast to the user side, so that a user knows about product purchase information according to the voice broadcast data, and the agent illegal operation is avoided. And carrying out corresponding response operation according to the voice data to be recognized fed back by the user side, and reducing follow-up user complaints. And extracting target voiceprint characteristics corresponding to the voice data to be recognized, and determining whether the user replies to the user according to the target voiceprint characteristics. The user terminal may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server may be implemented by an independent server or a server cluster formed by a plurality of servers.

In an embodiment, as shown in fig. 2, a product supervision method is provided, and the method is applied to the server in fig. 1 for illustration, and specifically includes the following steps:

s10: and acquiring a product payment request sent by the user side, wherein the product payment request comprises a user identifier, a product identifier and an agent identifier.

The product payment request is a request sent by a user based on a user side to pay for the purchased product. The user identification refers to a unique identification allocated to each user by the server, or can be an identification corresponding to the user of the identity card number of the user, and the unique user can be determined through the user identification. The product identifier is an identifier corresponding to the product purchased by the user, and a unique product is determined according to the product identifier. The server configures a corresponding product identifier for each product in advance, wherein each product identifier corresponds to a product type, the product type refers to a category to which the product identifier belongs, for example, a product corresponding to the product identifier is good luck, and the product type to which the product identifier belongs is unexpected insurance. The agent identification refers to an identification which is allocated to the agent in advance by a server, and a unique agent can be determined through the agent identification.

Specifically, the agent can sell the product through the product selling system, and after the user determines to purchase the product, the user can send a product payment request based on the user side, wherein the product payment request comprises a user identifier, a product identifier and an agent identifier.

S20: and matching the agent identifier with a blacklist identifier in the blacklist list to obtain a list matching result.

The blacklist identification refers to an identification corresponding to an agent performing illegal operation.

Specifically, a blacklist is stored in the database, and each blacklist identifier, a corresponding monitoring duration and a corresponding monitoring period are stored in the blacklist. And the server matches the agent identifier with each blacklist identifier in the blacklist to obtain a matching result. The matching result can comprise two types of successful matching of the agent identifier and the blacklist identifier and failure matching of the agent identifier and the blacklist identifier. If the agent identifier is successfully matched with the blacklist identifier, the server needs to perform IVR voice verification on the payment operation, wherein IVR (Interactive Voice Response) is interactive voice response. And supervision on product sales is realized. If the matching of the agent identification and the blacklist identification fails, the user can directly carry out payment operation without IVR voice verification. Through IVR voice verification, whether the user knows about the product purchase information is determined, and the user replying to the confirmation information is ensured to be the person who purchases the product.

S30: if the list matching result is that the matching is successful, determining the corresponding product type according to the product identifier, calling the corresponding voice broadcasting data according to the product type, and performing voice broadcasting on the voice broadcasting data by adopting a TTS technology.

The voice broadcasting data refers to pre-configured text data. It can be understood that each product type corresponds to one piece of voice broadcasting data, and the user can clearly know the data to be noted by the product purchasing information, namely the product purchasing information, through the voice broadcasting data, so as to avoid the operator's illegal operation. Wherein the voice broadcast data should be simple, clear and clear.

Specifically, if the agent identifier is successfully matched with the blacklist identifier, IVR voice verification is required to be performed on the payment request corresponding to the agent identifier. Firstly, a database is queried according to product identifications, and the product types corresponding to each product identification are stored in the database, wherein the product types comprise underwriting, heddle Jin Dai/loan minor, card activation first dial/turn and other types and the like. For example, the product identifier corresponds to the product as the lucky follower, and the comparison table in the database is searched according to the lucky follower, and the product type corresponding to the lucky follower is the accident insurance. And calling corresponding voice broadcasting data according to the product type. For example, (1) the product type is accident insurance, and the corresponding voice broadcast data is: "do you know that you currently purchase an unexpected insurance, but not a revenue insurance? The clear request is clear, if the doubt request is in question, the user is thanks to the inquiry. "(2) the product type is heddle Jin Dai/loan xiao, and the corresponding voice broadcast data is: "do you know that what is you currently buying insurance is not directly related to other integrated financial services such as loans? The clear request is clear, if the doubt request is in question, the user is thanks to the inquiry. "(3) the product type is credit card activation first dial/turn, and the corresponding voice broadcast data is: "do you know that what insurance you are currently buying is not directly related to other integrated financial services such as credit card line improvement? The clear request is clear, if the doubt request is in question, the user is thanks to the inquiry. And searching a database according to the product type, acquiring voice broadcasting data corresponding to the product type, and performing voice broadcasting on the voice broadcasting data by adopting a TTS technology, namely performing IVR voice verification, avoiding the illegal operation of the seat, and realizing the supervision of product sales.

S40: the method comprises the steps of obtaining voice data to be recognized fed back by a user side based on voice broadcasting data, carrying out semantic analysis on the voice data to be recognized, and obtaining a semantic analysis result.

The voice data to be recognized refers to voice data fed back by the user based on the voice broadcasting data at the user side.

Specifically, the server side provides a voice acquisition interface, the voice acquisition interface is connected with the user side network, the user can feed back voice data to be identified to the server side based on the user side according to voice broadcast data, the server side acquires the voice data to be identified through the voice acquisition interface, performs semantic analysis on the voice data to be identified, determines key content in the voice data to be identified, and obtains a semantic analysis result, wherein the semantic analysis result comprises confirmation information, denial information and uncertainty information. It can be understood that the voice broadcast data corresponding to each product type in the voice broadcast data includes voice broadcast data of "clear please reply is clear, if the request is in question," that is, the voice data to be recognized returned by the user includes clear, that is, the semantic analysis result is confirmation information, and the voice data to be recognized returned by the user includes questions, then the semantic analysis result is denial information; if the voice data to be recognized, which is replied by the user, does not contain the clear and questionable voice data, the semantic analysis result is uncertain information.

S50: and extracting target voiceprint features from the voice data to be identified, and matching the standard voiceprint features corresponding to the user identification with the target voiceprint features to obtain a voiceprint matching result.

The target voiceprint features refer to features extracted from voice data to be recognized, and specifically, the MFCC algorithm may be used to extract the target voiceprint features from the voice data to be recognized. The standard voiceprint features refer to the voiceprint features corresponding to the user identification which are pre-entered. The MFCC (Mel-scale Frequency Cepstral Coefficients, mel cepstrum coefficient) features are cepstrum parameters extracted in a Mel scale frequency domain, the Mel scale describes nonlinear characteristics of human ear frequencies, and a MFCC algorithm is adopted to perform voiceprint feature extraction on voice data to be identified, and the obtained MFCC features are target voiceprint features.

Specifically, firstly, extracting target voiceprint features from voice data to be recognized, searching a database according to correspondence of user identifiers, obtaining standard voiceprint features corresponding to the user identifiers, performing similarity calculation on the target voiceprint features and the standard voiceprint features by adopting a similarity algorithm, and obtaining voiceprint matching results according to the calculated voiceprint similarity, wherein the voiceprint matching results comprise matching success and matching failure.

Further, the voice broadcasting data, the user identification and the voice data to be identified are stored in a database, namely, voice recording is carried out from the voice broadcasting of the voice broadcasting data by adopting a TTS technology until the recording is finished after the voice data to be identified sent by the user is recorded, and the recorded voice data is stored. If the follow-up user complains about the product corresponding to the product identifier, the database can be queried to call the voice data to be identified to determine whether the product is purchased, so that the fraudulent behavior of the user is avoided, and the effectiveness of the voice data to be identified is ensured.

S60: and carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result.

Specifically, if the semantic analysis result is confirmation information, the user is clearly informed of the currently searched and purchased product, and if the semantic analysis result is negative information, the user is questioned about the purchased product, namely, the semantic analysis result is different, and the corresponding response operation should be different. If the voiceprint matching result is successful, the target voiceprint characteristics and the standard voiceprint characteristics are successfully matched, and the voice data to be recognized is replied by the user; if the voiceprint matching result is a matching failure, the target voiceprint characteristic and the standard voiceprint characteristic are successfully matched, and the voice data to be recognized is not replied by the user, namely the voiceprint matching result is different, and the corresponding response operation is different. And carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result so as to ensure that a user clearly knows that the product is purchased and the product is the voice data to be recognized returned by the user, avoid complaints of subsequent users and realize supervision of product sales.

And step S10-S60, when a product payment request sent by a user side is obtained, matching the seat identifier with a blacklist identifier in a blacklist to determine whether to carry out voice verification on the product payment request. If the list matching result is that the matching is successful, corresponding voice broadcasting data are determined according to the product identification, and the voice broadcasting data are subjected to voice broadcasting to determine whether a user clearly knows the product purchase information of the purchased product. The voice data to be recognized fed back by the user side based on the voice broadcasting data is acquired, semantic analysis is carried out on the voice data to be recognized, whether the user knows about the purchasing of the product clearly or not is acquired, and follow-up user complaints are reduced. And extracting target voiceprint features from the voice data to be identified, matching the standard voiceprint features corresponding to the user identification with the target voiceprint features, and obtaining a voiceprint matching result to realize whether the user replies. And carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result, and realizing supervision of product sales.

In one embodiment, as shown in fig. 3, in step S40, the semantic analysis is performed on the voice data to be recognized to obtain a semantic analysis result, which specifically includes the following steps:

S41: preprocessing the voice data to be recognized to obtain voice information.

The preprocessing refers to framing, windowing, pre-emphasis and the like of the voice data to be recognized. Framing is the grouping of N samples into one observation unit, called a frame. Typically, N has a value of 256 or 512, covering a period of about 20-30 ms. To avoid excessive variation between two adjacent frames, the overlapping area includes M samples, typically M has a value of about 1/2 or 1/3 of N, and this process is called framing.

The windowing is that each frame is multiplied by a Hamming Window (i.e. Hamming Window), and as the amplitude-frequency characteristic of the Hamming Window is that the side lobe attenuation is larger, the service end can increase the continuity of the left end and the right end of the frame by windowing single-frame voice data. The pre-emphasis is to make the single frame voice data after windowing pass through a high-pass filter to promote the high frequency part, make the frequency spectrum of the signal smoother, keep in the whole frequency band from low frequency to high frequency, can use the same signal to noise ratio to calculate the frequency spectrum, highlight the formants of the high frequency, obtain the voice information after pre-emphasis.

S42: and extracting the characteristics of the voice information to obtain the voice characteristics.

Where the speech features include, but are not limited to, the use of filter features. The Filter-Bank (Fbank) feature is a speech feature commonly used in speech recognition. Because the mel feature commonly used in the prior art can perform dimension reduction processing on the information in the process of performing model identification, partial information is lost, and in order to avoid the problem, the filter feature can be used to replace the commonly used mel feature in the embodiment.

S43: and recognizing the voice characteristics by adopting a voice recognition model to obtain target text data.

The voice recognition model comprises a pre-trained acoustic model and a language model. The acoustic model is a model for acquiring a phoneme sequence corresponding to a voice feature. The phonemes are the smallest units in the speech, which can be understood as the pinyin within a chinese character, a word containing at least one phoneme. For example: the Chinese syllable ā (o) has only one phoneme, the a i (ai) has two phonemes, etc. Methods of training acoustic models include, but are not limited to, training using GMM-HMM (gaussian mixture model). The language model is a model for converting a phoneme sequence into natural language text. Specifically, the server inputs the voice characteristics into a pre-trained acoustic model for recognition, acquires a phoneme sequence corresponding to the target voice characteristics, and then inputs the acquired phoneme sequence into a pre-trained language model for conversion, so as to acquire corresponding target text data.

S44: and carrying out semantic analysis on the target text data by adopting an NLP technology to obtain a semantic analysis result corresponding to the target text data.

Among them, NLP (Natural Language Processing ) is a language processing technique in which a computer analyzes, understands, and derives meaning from human language in an efficient manner. By utilizing NLP technology, developers can organize and build knowledge systems to perform tasks such as automatic abstracting, translating, named entity recognition, relation extraction, emotion analysis, voice recognition, topic segmentation and the like. In this embodiment, the semantic analysis interface provided by the open source NLP technology may be used to perform intent analysis on the target text data, so as to obtain an analysis result corresponding to the target text data.

Specifically, the target text data is input into a semantic analysis interface for intention analysis, and a semantic analysis result corresponding to the target text data is obtained, wherein the semantic analysis result comprises confirmation information, denial information and uncertainty information. By means of semantic analysis on the target text data, whether a user clearly knows about the purchasing of the product or not is determined, and corresponding response operation is performed according to the semantic analysis result, so that seat violation operation is avoided.

In step S41-S44, the voice data to be recognized is preprocessed to obtain smoother voice information, then the voice information is subjected to feature extraction to obtain voice features, so that the voice features are recognized by using a voice recognition model to obtain target text data, and semantic analysis is performed on the target text data by using an NLP technology to determine whether the user has clear product purchasing result, so that the corresponding response operation is facilitated.

In an embodiment, as shown in fig. 4, in step S50, a standard voiceprint feature corresponding to a user identifier is matched with a target voiceprint feature to obtain a voiceprint matching result, which specifically includes the following steps:

s51: and carrying out similarity calculation on the standard voiceprint features and the target voiceprint features corresponding to the user identification by adopting a cosine similarity algorithm, and obtaining the voiceprint similarity.

The voiceprint similarity refers to a value of similarity between the target voiceprint feature and the standard voiceprint feature.

Specifically, firstly, standard voiceprint features corresponding to user identifications are obtained, and a cosine similarity formula is adopted to calculate the similarity between the standard voiceprint features and target voiceprint features, so that voiceprint similarity is obtained. Wherein, the cosine similarity calculation formula is S is voiceprint similarity, A _i For target voiceprint features, B _i For standard voiceprint features, i is the i-th dimension feature and n is the number of dimensions.

S52: if the voiceprint similarity is larger than the similarity threshold, the target voiceprint features are successfully matched with the standard voiceprint features, and the voiceprint matching result is obtained as successful matching.

The similarity threshold is preset to determine whether the user corresponding to the target voiceprint feature is a user of a prestored standard voiceprint feature.

Specifically, the server compares the obtained voiceprint similarity with a similarity threshold, if the voiceprint similarity is larger than the similarity threshold, the target voiceprint feature is successfully matched with the standard voiceprint feature, and a voiceprint matching result is obtained as successful matching.

S53: if the voiceprint similarity is not greater than the similarity threshold, the target voiceprint feature fails to match the standard voiceprint feature, and the voiceprint matching result is obtained as the matching failure.

Specifically, if the voiceprint similarity is not greater than the similarity threshold, matching the target voiceprint feature with the standard voiceprint feature fails, and obtaining a voiceprint matching result as a matching failure.

In the steps S51-S53, a cosine similarity algorithm is adopted to calculate the similarity of the target voiceprint feature and the standard voiceprint feature, the voiceprint similarity is obtained, the calculation method is simple and quick, and whether the voiceprint feature is the principal reply is determined according to the voiceprint feature. If the voiceprint similarity is larger than the similarity threshold, the target voiceprint features are successfully matched with the standard voiceprint features, and a voiceprint matching result is obtained as successful matching; if the voiceprint similarity is not greater than the similarity threshold, the target voiceprint feature fails to match the standard voiceprint feature, and the voiceprint matching result is obtained as the matching failure, so that the acquisition accuracy of the voiceprint matching result is improved.

In one embodiment, as shown in fig. 5, in step S60, according to the semantic analysis result and the voiceprint matching result, a corresponding response operation is performed on the product payment request, which specifically includes the following steps:

s61: if the voiceprint matching result is successful in matching and the semantic analysis result is confirmation information, entering a corresponding payment operation interface based on the product payment request.

Specifically, if the voiceprint matching result is that the matching is successful, the voice data to be identified is replied by the user, and the semantic analysis result is the confirmation information, which indicates that the user clearly has a product purchase notice, the user enters a corresponding payment operation interface based on the product payment request, and the user can complete the payment operation based on the payment operation interface.

S62: if the voiceprint matching result is successful in matching and the semantic analysis result is negative information, the answering information is fed back to the user side, and the continuous processing information fed back by the user side based on the answering information is obtained.

The answering information is information which corresponds to the product identifier and is used for removing confusion of the user, and understandably, the answering information is a speaking operation of common questions and answer information which corresponds to the product identifier, and the user is subjected to the answering operation. The continued processing information is information that the user determines whether to confuse based on the answer information. The continuing processing information includes continuing payment information and refusing payment information.

Specifically, if the voiceprint matching result is successful and the semantic analysis result is denial information, the voice data to be identified is replied by the user, but the user has confusion on the purchased product, the confusion problem point of the user is determined first, then the answering information is called according to a partner problem point of the user, the answering information is voice broadcast by adopting a TTS technology, the user can be confused according to the answering information, and the continuous processing information is fed back.

S63: and if the continuing processing information is the continuing payment information, executing broadcasting of the voice broadcasting data by adopting a TTS technology.

Specifically, if the obtained continuous processing information fed back by the user side based on the answering information is continuous payment information, the user is confused according to the answering information, and a step of broadcasting voice broadcasting data by adopting a TTS technology is executed, which is the same as the step in step S30, and details are not repeated here.

S64: if the continuous processing information is the refused payment information, the objection processing information is fed back to the user side, and the final determination information fed back by the user side based on the objection processing information is obtained.

Specifically, if the continuous processing information fed back by the user side based on the answering information is refused to pay information, the user feeds back the objection processing information to the server side according to the answering information which is not confused, wherein the objection processing information is information which is confused manually, final determining information fed back by the user based on the objection processing information is obtained, the final determining information is information that the user determines whether to confuse based on the objection processing information, and the final determining information comprises payment stopping information and continuous payment information.

S65: and if the final determined information is the payment stopping information, exiting the current interface.

Specifically, if the final determination information fed back by the user side is the payment stopping information, the user is not confused according to the objection processing information, and if the operator possibly has illegal operation, the payment operation cannot be performed, and the current interface is exited. And storing the payment stopping information and the agent identifier in an associated manner, so that the monitoring duration corresponding to the blacklist identifier can be increased conveniently.

S66: and if the final determined information is the continuing payment information, executing broadcasting of the voice broadcasting data by adopting a TTS technology.

Specifically, if the final determination information fed back by the user side is the continuing payment information, the user is confused according to the objection processing information, and the voice broadcasting data is broadcasted by adopting the TTS technology, and the step is the same as the step in the step S30, and is not described in detail herein. It can be understood that, if the user is confused, step S30 is executed, the voice data to be recognized is obtained again through step S30, and the voice data to be recognized, the user identifier, the final determination information and the like are stored in an associated manner, so that the voice data to be recognized can be conveniently used as evidence in the case of complaint of the subsequent user.

Further, if the voiceprint matching result is that the matching is failed, reminding information is sent to the mobile terminal corresponding to the user identifier. The reminding information is "the current reply is not the operation of the user", and the voice data to be recognized is determined to be the reply of the user through the step. The mobile terminal is terminal equipment of a user.

Further, if the voiceprint matching result is successful and the semantic analysis result is uncertain information, repeating executing the voice broadcasting of the voice broadcasting data by adopting the TTS technology. Specifically, when the semantic analysis result is uncertain information, that is, the voice data to be recognized replied by the user does not correspond to the voice broadcasting data, voice broadcasting is performed on the voice broadcasting data by adopting a TTS technology, so that whether the user clearly knows that the product is purchased or not is determined, and the operator is prevented from violating the rules.

In step S61-S66, if the voiceprint matching result is successful and the semantic analysis result is confirmation information, entering a corresponding payment operation interface according to the product payment request, avoiding the seat violation operation and reducing the follow-up user complaints. If the voiceprint matching result is successful in matching and the semantic analysis result is denial information, users are confused according to answering information, objection processing information and the like, so that the users can clearly know about the purchasing of products and the beard and knowledge, and the user experience is improved.

In one embodiment, as shown in fig. 6, before step S20, that is, before matching the agent identifier with the blacklist identifier in the blacklist, the product supervision method further includes:

S201: and acquiring the number of violations corresponding to each agent identifier, and taking the agent identifier as a blacklist identifier if the number of violations is greater than a number threshold.

The number threshold is a threshold preset to determine whether the agent identifier is a blacklist identifier. The blacklist identification refers to the agent identification that the number of violations is greater than the number threshold.

Specifically, obtaining the number of violations corresponding to each seat identifier, wherein the channels for obtaining the number of violations can be divided into a plurality of types, one type is that a customer complains about the seat, and the obtained number of violations corresponds to the seat identifier; one is the number of violations corresponding to the agent identifier fed back by the supervisory personnel, and the other is the information of stopping payment fed back when the client pays, and the number of violations corresponding to the agent identifier is acquired according to the information of stopping payment. Counting all the violation times corresponding to each agent identifier, comparing all the violation times with a time threshold, and taking the agent identifier as a blacklist identifier if the violation times are larger than the time threshold, so that IVR voice verification is carried out on a payment request corresponding to the blacklist identifier (namely, the steps of S30-S50) when the blacklist identifier sells products subsequently, and supervision on sales behaviors corresponding to the blacklist identifier is realized, so that users are ensured to know purchased products clearly, and follow-up user complaints are avoided. If the number of violations is not greater than the number threshold, IVR voice verification is not carried out on the product corresponding to the seat identifier.

S202: and determining the supervision time length corresponding to the blacklist identification and the supervision time limit corresponding to the supervision time length according to the number of violations corresponding to the blacklist identification.

The supervision duration refers to a time corresponding to the number of violations, for example, the number of violations is 2, and the supervision duration is one month. The supervision period refers to the starting time and the deadline corresponding to the blacklist identifier, for example, the number of violations of a certain agent identifier is 2 times, and the supervision period is one month when the number of violations of a certain agent identifier is greater than the number threshold value in 1 month 1, and the supervision period is 1 month 1 day to 2 months 1 day.

Specifically, the supervision time lengths corresponding to different violation times are different, and it can be understood that the more the violation times are, the longer the corresponding supervision time length is, whereas the fewer the violation times are, the shorter the corresponding supervision time length is. And pre-configuring the corresponding relation between each violation number and the supervision time length, and storing the corresponding relation in a database. Searching a database through the number of violations corresponding to the blacklist identification, acquiring the corresponding supervision time length as the supervision time length corresponding to the blacklist identification, and determining the supervision time length according to the supervision time length. Through the step, the supervision time length and supervision time limit corresponding to each blacklist identifier can be rapidly determined, and the determination method is simple and rapid.

S203: a blacklist is formed based on each blacklist identification and supervision period and stored in a database.

The blacklist is a list storing the corresponding relation between blacklist identification and supervision deadlines.

Specifically, a blacklist is formed based on each blacklist identifier and the supervision deadline, so that follow-up determination of which product payment requests need IVR voice verification is facilitated.

In step S201-S203, a blacklist identifier is determined according to the number of violations corresponding to each agent identifier, so that IVR voice verification can be performed on the product payment emotion corresponding to the blacklist identifier. And determining the supervision time length corresponding to the blacklist identification and the supervision time period corresponding to the supervision time length according to the number of violations corresponding to the blacklist identification, so as to supervise the product payment request corresponding to the blacklist identification within the supervision time period.

In an embodiment, as shown in fig. 7, after step S60, that is, after performing a corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result, the product supervision method further includes:

s601: and when the current time of the system is the expiration time of the supervision period corresponding to each blacklist mark, acquiring at least one semantic analysis result of the blacklist mark in the supervision period.

Specifically, the current_date function may be used to obtain the current time of the system, and when the current time of the system is the deadline of the supervision period corresponding to the blacklist identifier, at least one semantic analysis result of the blacklist identifier in the supervision period is obtained, and whether the agent corresponding to the blacklist identifier has the illegal operation is determined through the semantic analysis result.

S602: and if the semantic analysis result is the confirmation information, removing the blacklist identification from the blacklist.

Specifically, if each semantic analysis result corresponding to the blacklist identifier is confirmation information, it is indicated that the agent corresponding to the blacklist identifier does not perform the illegal operation within the supervision period, the blacklist identifier needs to be extracted from the blacklist, and the product payment request corresponding to the blacklist identifier does not need to be supervised.

Further, whether the blacklist identifier has illegal operation or not can be determined in other modes, if yes, the step of processing the total amount of the assembly according to a preset rule and determining the newly added supervision duration is executed; and if no illegal operation exists, executing the step of eliminating the blacklist identification from the blacklist.

S603: if at least one piece of negative information exists, counting the total amount of the stopped payment information corresponding to the blacklist mark, processing the total amount according to a preset rule, and determining the newly-increased supervision time.

The total amount of the exchange refers to the amount of the payment stopping information fed back by the user side corresponding to the blacklist identifier. The preset rule is a preset rule for determining the supervision time length according to the total amount of the exchanges, for example, the larger the total amount of the exchanges is, the longer the corresponding newly-added supervision time length is. The newly added monitoring duration refers to a duration increased on the basis of the monitoring duration, for example, a certain blacklist identifier corresponds to a monitoring duration of 1 month, and according to a preset rule and the total amount of exchanges, the newly added monitoring duration is determined to be 10 days, and the monitoring duration is increased by 10 days on the basis of the monitoring duration of 1 month.

Specifically, if at least one piece of negative information exists in the semantic analysis result corresponding to the blacklist identifier, the user has a question about the purchased product, the total amount of the payment stopping information corresponding to the blacklist identifier is counted, and the newly added monitoring duration is determined according to a preset rule and the total amount.

S604: and updating the monitoring period corresponding to the blacklist identification in the blacklist based on the newly added monitoring period.

Specifically, after the server side obtains the newly-increased supervision time length corresponding to the blacklist identifier, the monitoring deadline corresponding to the blacklist identifier is redetermined according to the newly-increased supervision time length, and the monitoring deadline corresponding to the blacklist identifier is updated in the blacklist.

In steps S601-S604, when the current time of the system is the deadline of the supervision period corresponding to the blacklist identifier, determining a semantic analysis result corresponding to the blacklist identifier; if the result of each semantic analysis is the confirmation information, the agent corresponding to the blacklist identification does not have illegal operation within the monitoring period, and the blacklist identification does not need to be monitored. If at least one piece of negative information exists, the agent corresponding to the blacklist identifier has illegal operation in the monitoring period, and the total amount of the stopped payment information corresponding to the blacklist identifier is counted to increase the monitoring period corresponding to the blacklist identifier. And based on the newly added supervision time length, updating a supervision time period corresponding to the blacklist identifier in the blacklist, so as to supervise the product payment request corresponding to the blacklist identifier within the supervision time period.

It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.

In an embodiment, a product supervision device is provided, where the product supervision device corresponds to the product supervision method in the above embodiment one by one. As shown in fig. 8, the product supervision apparatus includes a payment request acquisition module 10, a list matching module 20, a voice broadcast module 30, a semantic analysis result acquisition module 40, a voiceprint matching result acquisition module 50, and a response operation module 60. The functional modules are described in detail as follows:

the payment request acquiring module 10 is configured to acquire a product payment request sent by a user, where the product payment request includes a user identifier, a product identifier, and an agent identifier.

The list matching module 20 is configured to match the agent identifier with a blacklist identifier in the blacklist list, and obtain a list matching result.

The voice broadcasting module 30 is configured to determine a corresponding product type according to the product identifier if the list matching result is that the matching is successful, invoke corresponding voice broadcasting data according to the product type, and perform voice broadcasting on the voice broadcasting data by adopting a TTS technology.

The semantic analysis result obtaining module 40 is configured to obtain voice data to be identified, which is fed back by the user terminal based on the voice broadcast data, and perform semantic analysis on the voice data to be identified to obtain a semantic analysis result.

And the voiceprint matching result obtaining module 50 is configured to extract a target voiceprint feature from the voice data to be identified, match the standard voiceprint feature corresponding to the user identifier with the target voiceprint feature, and obtain a voiceprint matching result.

And the response operation module 60 is used for carrying out corresponding response operation on the product payment request according to the semantic analysis result and the voiceprint matching result.

In an embodiment, the semantic analysis result obtaining module 40 includes a preprocessing unit 41, a feature extraction unit 42, a text recognition unit 43, and a semantic analysis unit 44.

The preprocessing unit 41 is configured to preprocess voice data to be recognized, and acquire voice information.

The feature extraction unit 42 is configured to perform feature extraction on the voice information, and obtain voice features.

The word recognition unit 43 is configured to recognize the voice feature by using the voice recognition model, and obtain the target word data.

The semantic analysis unit 44 is configured to perform semantic analysis on the target text data by using an NLP technology, and obtain a semantic analysis result corresponding to the target text data.

In an embodiment, the voiceprint matching result obtaining module 50 includes a voiceprint similarity obtaining unit, a first voiceprint matching result obtaining unit, and a second voiceprint matching result obtaining unit.

And the voiceprint similarity acquisition unit is used for carrying out similarity calculation on the standard voiceprint characteristics and the target voiceprint characteristics corresponding to the user identification by adopting a cosine similarity algorithm to acquire voiceprint similarity.

The first voiceprint matching result obtaining unit is used for obtaining that if the voiceprint similarity is larger than the similarity threshold, the target voiceprint feature is successfully matched with the standard voiceprint feature, and the voiceprint matching result is successfully matched.

And the second voiceprint matching result obtaining unit is used for obtaining that if the voiceprint similarity is not greater than the similarity threshold, the target voiceprint feature fails to match the standard voiceprint feature, and the voiceprint matching result is the matching failure.

In an embodiment, the response operation module 60 includes a payment operation interface acquisition unit, a continued processing information acquisition unit, a first response unit, a final determination information acquisition unit, a second response unit, and a third response unit.

The payment operation interface acquisition unit is used for entering a corresponding payment operation interface based on the product payment request if the voiceprint matching result is successful in matching and the semantic analysis result is confirmation information.

And the continuous processing information acquisition unit is used for feeding back answering information to the user side if the voiceprint matching result is successful in matching and the semantic analysis result is negative information, and acquiring continuous processing information fed back by the user side based on the answering information.

And the first response unit is used for executing broadcasting of the voice broadcasting data by adopting the TTS technology if the continuous processing information is the continuous payment information.

The final determination information acquisition unit is used for feeding back the objection processing information to the user terminal if the continuous processing information is the refused payment information, and acquiring the final determination information fed back by the user terminal based on the objection processing information.

And the second response unit is used for exiting the current interface if the final determination information is the payment stopping information.

And the third response unit is used for executing broadcasting of the voice broadcasting data by adopting a TTS technology if the final determined information is the continuing payment information.

In an embodiment, the product supervision device further comprises a blacklist identification determination unit, a supervision deadline determination unit and a blacklist formation unit before the list matching module 20.

The blacklist identification determining unit is used for obtaining the number of violations corresponding to each agent identification, and if the number of violations is greater than the number threshold, the agent identification is used as the blacklist identification.

And the supervision period determining unit is used for determining supervision time length corresponding to the blacklist identification and supervision period corresponding to the supervision time length according to the number of violations corresponding to the blacklist identification.

And the blacklist forming unit is used for forming a blacklist based on each blacklist identifier and the supervision period and storing the blacklist in the database.

In an embodiment, after responding to the operation module 60, the product supervision apparatus further includes a semantic analysis result acquisition unit, a blacklist identification rejection unit, a newly added supervision time length determination unit, and a supervision time length update unit.

The semantic analysis result acquisition unit is used for acquiring at least one semantic analysis result of the blacklist identification in the supervision period when the current time of the system is the deadline of the supervision period corresponding to the blacklist identification.

And the blacklist identification eliminating unit is used for eliminating the blacklist identification from the blacklist if each semantic analysis result is the confirmation information.

And the newly added supervision time length determining unit is used for counting the total amount of the stopped payment information corresponding to the blacklist identifier if at least one piece of negative information exists, processing the total amount of the stopped payment information according to a preset rule, and determining the newly added supervision time length.

And the monitoring period updating unit is used for updating the monitoring period corresponding to the blacklist identifier in the blacklist based on the newly-added monitoring period.

For specific limitations of the product monitoring device, reference may be made to the limitations of the product monitoring method hereinabove, and will not be described in detail herein. The various modules in the product supervision device described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 9. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used to store data or the like, e.g., blacklist, generated or obtained during execution of the product administration method. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a product supervision method.

In one embodiment, a computer device is provided that includes a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to implement the steps of the product supervision method in the above embodiment, for example, S10 to S60 shown in fig. 2, or the steps shown in fig. 3 to 7. Alternatively, the processor, when executing the computer program, implements the functions of each module in the product supervision device in the above embodiment, for example, the functions of the modules 10 to 60 shown in fig. 8. To avoid repetition, no further description is provided here.

In an embodiment, a computer readable storage medium is provided, on which a computer program is stored, which when executed by a processor implements the product supervision method in the above method embodiment, for example, steps S10 to S60 shown in fig. 2 or steps shown in fig. 3 to 7. Alternatively, the computer program, when executed by the processor, implements the functions of the modules in the product monitoring apparatus of the above embodiment, for example, the functions of the modules 10 to 60 shown in fig. 8. To avoid repetition, no further description is provided here.

Those skilled in the art will appreciate that implementing all or part of the above-described methods may be accomplished by way of a computer program, which may be stored on a non-transitory computer readable storage medium and which, when executed, may comprise the steps of the above-described embodiments of the methods. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.

The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.

Claims

1. A method of product supervision, comprising:

the voice information processing method comprises the steps of obtaining voice data to be recognized fed back by a user side based on voice broadcasting data, preprocessing the voice data to be recognized, and obtaining voice information; extracting the characteristics of the voice information to obtain voice characteristics; identifying the voice characteristics by adopting a voice identification model to obtain target text data; performing semantic analysis on the target text data by using an NLP technology to obtain a semantic analysis result corresponding to the target text data;

extracting target voiceprint features from the voice data to be identified, and performing similarity calculation on standard voiceprint features corresponding to user identifiers and the target voiceprint features by adopting a cosine similarity algorithm to obtain voiceprint similarity; if the voiceprint similarity is greater than a similarity threshold, the target voiceprint features are successfully matched with the standard voiceprint features, and a voiceprint matching result is obtained as a successful matching result; if the voiceprint similarity is not greater than a similarity threshold, the target voiceprint feature fails to match the standard voiceprint feature, and a voiceprint matching result is obtained as a matching failure;

If the voiceprint matching result is successful in matching and the semantic analysis result is confirmation information, entering a corresponding payment operation interface based on the product payment request; if the voiceprint matching result is successful and the semantic analysis result is negative information, the answering information is fed back to the user side, and the continuous processing information fed back by the user side based on the answering information is obtained; if the continuing processing information is continuing payment information, executing the voice broadcasting data broadcasting by adopting a TTS technology; if the continuing processing information is refused payment information, the objection processing information is fed back to the user side, and final determining information fed back by the user side based on the objection processing information is obtained; if the final determination information is the payment stopping information, exiting the current interface; and if the final determination information is the continuing payment information, executing the voice broadcasting data broadcasting by adopting the TTS technology.

2. The product supervision method according to claim 1, wherein prior to the matching the agent identification with a blacklist identification in a blacklist, the product supervision method further comprises:

Acquiring the number of violations corresponding to each agent identifier, and taking the agent identifier as a blacklist identifier if the number of violations is greater than a number threshold;

determining supervision time length corresponding to the blacklist identification and supervision time limit corresponding to the supervision time length according to the number of violations corresponding to the blacklist identification;

and forming a blacklist based on each blacklist identifier and the supervision period.

3. The product supervision method according to claim 1, wherein after the corresponding response operation to the product payment request is performed according to the semantic analysis result and the voiceprint matching result, the product supervision method further comprises:

when the current time of the system is the deadline of the supervision period corresponding to the blacklist identification, acquiring at least one semantic analysis result of the blacklist identification in the supervision period;

if each semantic analysis result is the confirmation information, the blacklist identification is removed from the blacklist;

if at least one piece of negative information exists, counting the total amount of the stopped payment information corresponding to the blacklist mark, processing the total amount according to a preset rule, and determining a newly-increased supervision time;

And based on the newly added supervision time length, updating the supervision time period corresponding to the blacklist identification in the blacklist.

4. A product supervision apparatus, comprising:

the semantic analysis result acquisition module is used for acquiring voice data to be identified, which is fed back by the user side based on the voice broadcast data, preprocessing the voice data to be identified and acquiring voice information; extracting the characteristics of the voice information to obtain voice characteristics; identifying the voice characteristics by adopting a voice identification model to obtain target text data; performing semantic analysis on the target text data by using an NLP technology to obtain a semantic analysis result corresponding to the target text data;

The voiceprint matching result acquisition module is used for extracting target voiceprint features from the voice data to be identified, and carrying out similarity calculation on standard voiceprint features corresponding to the user identification and the target voiceprint features by adopting a cosine similarity algorithm to acquire voiceprint similarity; if the voiceprint similarity is greater than a similarity threshold, the target voiceprint features are successfully matched with the standard voiceprint features, and a voiceprint matching result is obtained as a successful matching result; if the voiceprint similarity is not greater than a similarity threshold, the target voiceprint feature fails to match the standard voiceprint feature, and a voiceprint matching result is obtained as a matching failure;

the response operation module is used for entering a corresponding payment operation interface based on the product payment request if the voiceprint matching result is successful in matching and the semantic analysis result is confirmation information; if the voiceprint matching result is successful and the semantic analysis result is negative information, the answering information is fed back to the user side, and the continuous processing information fed back by the user side based on the answering information is obtained; if the continuing processing information is continuing payment information, executing the voice broadcasting data broadcasting by adopting a TTS technology; if the continuing processing information is refused payment information, the objection processing information is fed back to the user side, and final determining information fed back by the user side based on the objection processing information is obtained; if the final determination information is the payment stopping information, exiting the current interface; and if the final determination information is the continuing payment information, executing the voice broadcasting data broadcasting by adopting the TTS technology.

5. The product supervision device of claim 4, wherein the semantic analysis result acquisition module comprises:

the preprocessing unit is used for preprocessing the voice data to be recognized to acquire voice information;

the feature extraction unit is used for extracting features of the voice information to obtain voice features;

the character recognition unit is used for recognizing the voice characteristics by adopting a voice recognition model to acquire target character data;

the semantic analysis unit is used for carrying out semantic analysis on the target text data by adopting an NLP technology, and obtaining a semantic analysis result corresponding to the target text data.

6. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the product supervision method according to any one of claims 1 to 3 when the computer program is executed.

7. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the product supervision method according to any one of claims 1 to 3.