User number processing method, apparatus, device, and storage medium

Abstract

The application provides a user number processing method, device and equipment and a storage medium, wherein the method comprises: obtaining number basic information of a user number to be analyzed; judging whether the number basic information meets a blacklist condition or a whitelist condition; if not, obtaining call signaling data corresponding to the user number; and judging whether the user number needs to be monitored based on a neural network model according to the call signaling data; if it is determined that the user number needs to be monitored, adding the user number to the blacklist and determining a corresponding label for the user number, so as to intercept or pass the call of the user number according to the corresponding strategy of the label. The user number processing method, device and equipment and the storage medium provided by the application can quickly and accurately find the user number that needs to be monitored, and improve the efficiency and accuracy of identifying harassment calls.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication technology, and in particular to a user number processing method, device, equipment and storage medium. BACKGROUND

[0002] With the extensive application of Internet big data, the contact between various industries in society is increasingly close, and the continuous leakage of personal information has led to an increasing number of scam calls and nuisance calls, which has seriously affected the work and life of the people and become a hot issue that the people strongly reflect.

[0003] The existing abnormal call monitoring technology mainly filters out abnormal and suspicious users according to the basic information of the user number, which can filter out the callers of nuisance calls to a certain extent, but the accuracy is poor. SUMMARY

[0004] The present application provides a user number processing method, device, equipment and storage medium to improve the accuracy of identifying nuisance calls.

[0005] In a first aspect, the present application provides a user number processing method, comprising:

[0006] Obtaining the number basic information of the user number to be analyzed;

[0007] Determining whether the number basic information meets the blacklist condition or the whitelist condition, if not, obtaining the call signaling data corresponding to the user number, and determining whether the user number needs to be monitored based on the neural network model according to the call signaling data;

[0008] If it is determined that the user number needs to be monitored, the user number is added to the blacklist, and a corresponding label is determined for the user number, so that the call of the user number is intercepted or passed according to the strategy corresponding to the label.

[0009] Optionally, determining whether the user number needs to be monitored based on the neural network model according to the call signaling data comprises:

[0010] According to the call signaling data, a corresponding feature vector is generated by a bag-of-words model and feature weight calculation;

[0011] The feature vector is input into the neural network model to obtain the result of whether the user number needs to be monitored.

[0012] Optionally, according to the call signaling data, a corresponding feature vector is generated by a bag-of-words model and feature weight calculation, comprising:

[0013] According to the call signaling data, a corresponding word set is generated;

[0014] generate a feature matrix corresponding to the call signaling data according to a preset bag-of-words model;

[0015] calculate feature weights corresponding to elements in the feature matrix, and obtain a feature vector corresponding to the call signaling data according to the calculated feature weights and the feature matrix.

[0016] Optionally, the bag-of-words model comprises a plurality of attribute information for representing the call signaling data.

[0017] generate a feature matrix corresponding to the call signaling data according to a preset bag-of-words model, comprising:

[0018] select words corresponding to the attribute information in the bag-of-words model from the word set, and generate the feature matrix corresponding to the call signaling data according to the selected words.

[0019] The plurality of attribute information in the bag-of-words model comprises: longitude and latitude information at the time of the call, whether the call is continuous, and number basic information of a calling number.

[0020] Optionally, the method further comprises:

[0021] determining whether the calling number initiating the call is in a blacklist, and if so, searching for a label corresponding to the calling number;

[0022] searching for a strategy corresponding to the label, and processing the call initiated by the calling number according to the corresponding strategy.

[0023] Optionally, processing the call initiated by the calling number according to the corresponding strategy comprises:

[0024] if the label corresponding to the calling number is a high-risk label, then the calling number is prohibited from making outgoing calls;

[0025] if the label corresponding to the calling number is a low-risk label, then alert information is generated and sent, and the number of outgoing calls of the calling number within a unit time is dynamically adjusted.

[0026] Optionally, the call signaling data comprises at least one of the following: user opening information, user behavior information, and user call information.

[0027] In a second aspect, the present application provides a user number processing device, comprising:

[0028] an acquisition module configured to acquire number basic information of a user number to be analyzed;

[0029] determining whether the user number needs to be monitored, if it is determined that the user number needs to be monitored, adding the user number to a blacklist, and determining a corresponding label for the user number, so as to intercept or pass through a call of the user number according to a policy corresponding to the label.

[0030] a processing module, configured to: if it is determined that the user number needs to be monitored, add the user number to a blacklist, and determine a corresponding label for the user number, so as to intercept or pass through a call of the user number according to a policy corresponding to the label.

[0031] In a third aspect, the present application provides an electronic device, comprising:

[0032] at least one processor and a memory;

[0033] the memory stores computer-executable instructions;

[0034] the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the user number processing method according to any one of the first aspect.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium, the computer-readable storage medium stores computer-executable instructions, when the processor executes the computer-executable instructions, the user number processing method according to any one of the first aspect is realized.

[0036] The present application provides a user number processing method, device, equipment and storage medium, by acquiring the number basic information of the user number to be analyzed, determining whether the number basic information meets the blacklist condition or the whitelist condition, if it does not meet, acquiring the call signaling data corresponding to the user number, and determining whether the user number needs to be monitored based on the neural network model according to the call signaling data, if it is determined that the user number needs to be monitored, adding the user number to a blacklist, and determining a corresponding label for the user number, so as to intercept or pass through a call of the user number according to a policy corresponding to the label, quickly and accurately find the user number that needs to be monitored, improve the efficiency and accuracy of identifying harassment phone, and improve the monitoring strength of the blacklist user. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present application and, together with the specification, serve to explain the principles of the application.

[0038] Figure 1 An application scenario schematic diagram provided for the embodiments of the present application;

[0039] Figure 2 A flowchart of a user number processing method provided by an embodiment of the present application is shown in FIG. 1.

[0040] Figure 3 A flowchart of another user number processing method provided by an embodiment of the present application is shown in FIG. 2.

[0041] Figure 4 A schematic diagram of a principle of data processing by a neural network model provided by an embodiment of the present application is shown in FIG. 3.

[0042] Figure 5 A structural schematic diagram of a user number processing apparatus provided by an embodiment of the present application is shown in FIG. 4.

[0043] Figure 6 A structural schematic diagram of an electronic device provided by an embodiment of the present application is shown in FIG. 5.

[0044] Figure 7 A whole architecture diagram of an abnormal call identification and interception system provided by an embodiment of the present application is shown in FIG. 6.

[0045] The above-described figures have shown specific embodiments of the present application, which will be described in more detail hereinafter. These figures and the following description are not intended to limit the scope of the present inventive concept in any way, but to illustrate the present inventive concept by reference to specific embodiments. DETAILED DESCRIPTION

[0046] The exemplary embodiments will be described in detail herein below with reference to the drawings. When the following description refers to the drawings, identical numbers on different figures represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of apparatuses and methods consistent with some aspects of the present application as detailed in the appended claims.

[0047] The technical solutions of the present application and how the technical solutions solve the above-mentioned technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and some embodiments may not be described again for the same or similar concepts or processes.

[0048] An application scenario provided by an embodiment of the present application is explained below.

[0049] The scheme provided by the embodiment of the present application can be applied to the case where a harassing call or a fraudulent call is made, and the system can identify such a number and limit the calling behavior of the number. When a user's personal information is leaked, the user will receive calls from some unknown numbers, which may be harassing calls, such as real estate marketing, education and training, or loan sales, or may be fraudulent calls, which pose a certain security risk to the user's property. When such unknown numbers are identified by the system, the calling behavior of the numbers is limited, and the user is greatly relieved from the harassment of frequent calls from such harassing or fraudulent calls.

[0050] Figure 1 An application scenario diagram is provided for the embodiment of the present application. As shown in Figure 1 When a calling user makes a call, the call information is first sent to the interception system, which determines whether to continue to transfer the call information to the called user or to intercept the call. Specifically, after receiving the information, the interception system determines the information of the calling user. If it is determined that the user information is a harassing call or a fraudulent call, the call is intercepted and the calling behavior of the calling user is limited, and no call link is established between the calling user and the called user. If it is determined that the user information is not a harassing call or a fraudulent call, the call information is transferred to the called user for normal voice communication.

[0051] It should be noted that Figure 1 This is only a scenario diagram for optional call restriction. In actual application, the call can also be identified and intercepted through other network architectures. Optionally, the interception system can be deployed in a service platform for establishing a communication connection between the calling user and the called user, and it is directly determined whether to intercept or pass through when the calling user initiates a call. Alternatively, the interception system and the service platform can be separately arranged, and the service platform communicates with the interception system to determine whether to intercept when the calling user initiates a call.

[0052] In some technologies, a harassing call can be identified through a blacklist. Specifically, if the calling number belongs to the number in the blacklist, the calling is limited. The blacklist can be constructed through some basic information of the user number, but there is still a problem of inaccuracy, and many user numbers cannot be determined whether they belong to a harassing call through basic information.

[0053] Therefore, the embodiment of the present application provides a user number processing method, which can process the call signaling data of the user number through a neural network model when it is impossible to determine whether the user number belongs to a harassing call through basic information, so as to intelligently identify whether the user number needs to be monitored, and effectively improve the accuracy and efficiency of identifying a harassing call.

[0054] In addition, the label classification can also be performed according to the basic information of the known user number, mainly by extracting keywords from the call information of the call number and the feedback situation (for example, complaints) of other users to the number, and determining the label information of the user number. Specifically, the user number can be labeled as a fraud class, an education propaganda class, a take-out delivery class, a express delivery class, and a product propaganda class, etc. When the call behavior is performed, the system will automatically match the corresponding label and display it on the called user's mobile phone, and the called user decides whether to continue listening.

[0055] Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. The following embodiments and features in the embodiments can be combined with each other without conflict between the embodiments.

[0056] Figure 2 A flowchart of a user number processing method provided by an embodiment of the present application is shown in FIG. 1. As shown in FIG. 1, the method in the embodiment can include: Figure 2

[0057] Step 201: Obtain the number basic information of a user number to be analyzed.

[0058] The user number to be analyzed can be a mobile phone number of any user in the existing network. The number basic information of the user number can include the user's name, ID number, handling time, handling channel, handling location, handling package, number opening location, reported situation, etc.

[0059] Step 202: Determine whether the number basic information meets the black list condition or the white list condition. If not, obtain the call signaling data corresponding to the user number, and determine whether the user number needs to be monitored based on the neural network model according to the call signaling data.

[0060] Optionally, when it is determined through the number basic information that the number meets one or more white list conditions, the number is identified as a white list user, and the normal user is removed from the monitoring. The white list conditions can include: the handling location belongs to a home city with a low reporting rate, the number is certified as a government and enterprise number, the handling channel has never been reported, the handled package belongs to a package that has never been reported or a high amount package, a dedicated package for the logistics industry is subscribed, etc. For example, when a number has handled a household package exceeding a certain amount, it is considered that the number meets the white list condition, and the number is not monitored.

[0061] ​When it is judged by the number basic information that the number meets one or more black list conditions, the number is identified as a black list user, and the black list user can be monitored. The black list conditions include: other numbers handled by the same user certificate have been reported, the user certificate has other numbers canceled in the recent period, the user handles more than two numbers in a week or five numbers in a month, the number boot location is in a high-risk area of fraud, the same terminal uses other numbers reported, and the like. For example, when a number is reported by multiple other users, it is considered that the number meets the black list conditions, and the number is monitored.

[0062] Optionally, the black list conditions can specifically include multiple conditions, and any condition met is considered to meet the black list conditions, and the user number can be added to the black list; similarly, the white list conditions can specifically include multiple conditions, and any condition met is considered to meet the white list conditions, and the user number can be added to the white list.

[0063] If a user number does not meet any black list condition or any white list condition, a neural network model can be introduced to intelligently identify whether it needs to be monitored.

[0064] Optionally, if a user number meets part of the black list conditions and part of the white list conditions, for example, the user number handling channel has been reported by others, and the area is a high-risk area of fraud, but the number also has certain call packages, and the number has not been reported, a neural network model can be introduced to intelligently identify it.

[0065] In the intelligent identification, the call signaling data corresponding to the user number can be obtained, and the neural network model is used to judge whether the user number needs to be monitored based on the call signaling data.

[0066] The call signaling data corresponding to the user number can include three parts. The first part is the user's opening information, which can be whether it is the first time to handle a card, the package situation, and the like. The second part is the user's behavior information (also called geographic latitude information), which can be whether it is fixed in a certain area or often goes to other provinces, whether it is in a region with more harassment calls, and the like. The third part is the user's call information, which can be the user's call duration, call frequency, call number characteristics, and the like.

[0067] The neural network model can be a pre-trained model. Specifically, the neural network model can be trained using historical call signaling data, each training sample can include call signaling data and a corresponding target variable, the target variable is used to indicate whether the user number corresponding to the call signaling data needs to be monitored, that is, whether it needs to be put into the blacklist, and the target variable can be manually input by artificial or obtained by other means. After obtaining the trained model, whether a user number needs to be monitored can be determined based on the trained model.

[0068] Step 203, if it is determined that the user number needs to be monitored, the user number is added to the blacklist, and a corresponding label is determined for the user number, so that the call of the user number is intercepted or passed according to the policy corresponding to the label.

[0069] In this embodiment, the user number that needs to be monitored is added to the blacklist. If it is detected that the user number in the blacklist initiates a call, the call can be limited.

[0070] Optionally, a corresponding label can be set for the user number in the blacklist, the label can be determined by manual labeling, or the label corresponding to the user number can also be determined by a neural network model or other intelligent learning algorithm, or the label can also be determined by the calling user number being called by the user number.

[0071] Optionally, the label can include but is not limited to: fraud, real estate marketing, education and training, loan sales, etc. Different labels correspond to different strategies.

[0072] After detecting that the user number in the blacklist initiates a call, the call is automatically intercepted or passed according to the policy corresponding to the label, and the policy corresponding to the label can be adjusted according to requirements. For example, the fraud class adopts a strategy of completely prohibiting the calling party to dial out the phone, and the education and training class adopts a strategy of reminding processing and dynamically adjusting the outgoing call volume per hour.

[0073] The method provided in this embodiment can obtain the number basic information of the user number to be analyzed, judge whether the number basic information meets the blacklist condition or the whitelist condition, if not, obtain the call signaling data corresponding to the user number, and determine whether the user number needs to be monitored based on the neural network model according to the call signaling data, if it is determined that the user number needs to be monitored, the user number is added to the blacklist, and a corresponding label is determined for the user number, so that the call of the user number is intercepted or passed according to the policy corresponding to the label. The method can quickly and accurately find the user number that needs to be monitored, improve the efficiency and accuracy of identifying harassment calls, and improve the monitoring strength of the blacklist users.

[0074] In combination with the above embodiments, Figure 3 Another flowchart of a user number processing method provided by the embodiments of the present application is shown in FIG. 6. As shown in FIG. 6, after the execution process is started, first, the number basic information of the existing network user number is synchronized, and it is determined whether the number basic information of the existing network user number meets the whitelist condition. If the whitelist condition is met, the user is determined to be a normal user who does not need to be monitored, an export operation is performed, the user number is removed from the system, and the execution process is ended. Figure 3

[0075] Optionally, if the user number does not meet the whitelist condition in the above execution process, it is further determined whether the user meets the blacklist condition. If the blacklist condition is met, the user number is controlled.

[0076] Optionally, if the user number does not meet the blacklist condition in the further determination, a smart network subscription step is performed, that is, the user number is connected through the interface of the intelligent network management system TCP / IP, and the call signaling data of the user number is acquired. Then, a monitoring signaling is added, which is used to send the call signaling data of the user number to the signaling collection and monitoring module, convert the call signaling data into a computer understandable mode, and establish a user call behavior model. Subsequently, a call monitoring is added, and the call signaling data corresponding to the call is transferred to the artificial intelligence recognition and disposal module through the signaling collection and monitoring module, and it is determined whether the call behavior and content are abnormal. If the determination result is no, the user is determined to be a normal user who does not need to be monitored, an export operation is performed, the user number is removed from the system, and the execution process is ended.

[0077] The intelligent network is developed on the basis of the existing telephone network, and refers to a telephone network or an integrated services digital network with intelligence, which can realize the control of network intelligence through software.

[0078] Optionally, if the call behavior and content of the user number are determined to be abnormal through the artificial intelligence recognition and disposal module, the user number is controlled.

[0079] For the blacklist user number which needs to be controlled, it is transported to the label management module for classified management and generation of various label numbers, the strategy module automatically generates corresponding strategies according to the labels, and the user number is added with control according to the corresponding strategies, and the execution process is ended.

[0080] ​In the embodiment, the number-based information of the user number is first judged preliminarily, some user numbers that do not need to be monitored are judged, and the user numbers are removed from the system, the black list numbers are accurately judged and intercepted, and the influence on other users is reduced. For the user numbers that cannot be judged, the call signaling data of the user numbers is acquired, and further judgment is performed. This process comprehensively covers all numbers, and each number is screened to determine whether it needs to be added to the black list, thereby reducing the negative influence of the harassment calls or the fraud calls on the users to a certain extent.

[0081] On the basis of the technical solutions provided in the above embodiments, optionally, the above method can be executed by any device with data processing capability, and can be executed by one device or by multiple devices together. The training process and the use process of the model can be implemented by the same device or by different devices. Similarly, the process of black list management and control and the process of intercepting the calling numbers can be implemented by the same device or by different devices. The embodiments of the present application do not limit this.

[0082] On the basis of the technical solutions provided in the above embodiments, optionally, based on the call signaling data, whether the user number needs to be monitored can be judged based on a neural network model, which can include:

[0083] According to the call signaling data, a corresponding feature vector is generated through a bag-of-words model and feature weight calculation; and the feature matrix is input into a neural network model to obtain a result of whether the user number needs to be monitored.

[0084] The bag-of-words model refers to putting all words into a bag without considering the word order and syntax, and each word is independent. If there are NN words in the word set, each sample is represented as an NN-dimensional vector, and the element is 0 / 1, which indicates whether the sample contains the corresponding word. Wherein, NN represents the number of non-repeated words in the word set, and the NN-dimensional vector represents that each sample can be represented as a 1*NN-dimensional vector. For example, when NN is 10, it means that there are 10 non-repeated words in the bag, and each sample can be represented as a 1*10-dimensional vector.

[0085] The feature weight calculation includes Boolean weight, TF-IDF (Term Frequency-Inverse Document Frequency, term frequency-inverse document frequency) type weight, and weight based on the entropy concept. The Boolean weight is 1 if it appears, otherwise it is 0, that is, the bag-of-words model; the TF-IDF is defined based on the word frequency; and the entropy-based weight gives a higher weight to the features appearing in the same document.

[0086] According to the bag-of-words model, a feature matrix corresponding to the user feature call signaling data is obtained, so that the text data is converted into a form that can be understood by a computer, facilitating the processing of a large amount of data and the timely monitoring of the user number to be monitored.

[0087] Optionally, according to the call signaling data, the corresponding feature vector is generated by the bag-of-words model and feature weight calculation, which can include: generating a corresponding word set according to the call signaling data; generating a feature matrix corresponding to the call signaling data according to a preset bag-of-words model; calculating the feature weight corresponding to each element in the feature matrix, and obtaining the feature vector corresponding to the call signaling data according to the calculated feature weight and the feature matrix. Through the bag-of-words model and feature weight calculation, the call signaling data can be quickly converted into a feature vector input into a neural network model, improving the efficiency of prediction.

[0088] Optionally, the bag-of-words model includes a plurality of attribute information for representing the call signaling data. According to the preset bag-of-words model, the feature matrix corresponding to the call signaling data can be generated, which can include: selecting words corresponding to the attribute information in the bag-of-words model from the word set, and generating the feature matrix corresponding to the call signaling data according to the selected words; wherein the plurality of attribute information in the bag-of-words model includes: longitude and latitude information at the time of call, whether continuous call, and number basic information of the calling number.

[0089] Specifically, the call signaling data can contain a lot of information, such as opening information, geographic latitude information, and call information, but some information may not be needed when calculating the feature vector. Therefore, the feature matrix can be constructed only according to the words required by the bag-of-words model.

[0090] In this embodiment, the required information is selected according to the attribute information, such as the longitude and latitude information at the time of call, whether continuous call, and number basic information of the calling number, to construct the feature vector, and the neural network model is used to determine whether the user number needs to be monitored.

[0091] For example, the selected attribute information is whether it is the first time to open an account, package information, latitude information when calling, call frequency information, and call duration information. At this time, there are two users. The first user's call signaling data is as follows: the user is the first time to open an account; the user is a package within 30 yuan; the user's location when calling is always a city in A province; the user's call frequency is 4 times / day; and the user's call duration is 20 minutes / call on average. The second user's call signaling data is as follows: the user is not the first time to open an account; the user is a package of 30-60 yuan; the user's location when calling is always a city in A province; the user's call frequency is 3 times / day; and the user's call duration is 10 minutes / call on average. Then, according to the bag-of-words model, the word set is {“first time to open an account”, “not the first time to open an account”, “30 yuan package”, “30-60 yuan package”, “a city in A province”, “call frequency 4 times / day”, “call frequency 2 times / day”, “call duration 20 minutes / call”, “call duration 10 minutes / call”}, different words correspond to different specific values, and the corresponding features can be obtained according to the bag-of-words model, for example:

[0092]

[0093] Similarly, the second user's call signaling data corresponds to the feature matrix:

[0094]

[0095] In this embodiment, the call signaling data corresponding to the user number is obtained according to the selected multiple attribute information, and the corresponding feature vector is generated according to the bag-of-words model and the calculation of the feature weight. The selection of the weight can be adjusted according to the actual situation. After the fixed weight is selected and calculated, the selected multiple attribute information is combined and arranged, and different values are assigned to different features according to the selected weight on the basis of the obtained feature matrix, to obtain the corresponding feature vector. For example, according to any of the above methods, the weight of each element in the matrix is obtained, then each element is multiplied by the corresponding weight to obtain an updated matrix as the feature vector. Specifically, when a Boolean weight is selected, if the selected word appears, the weight is assigned as 1, and if the selected word does not appear, the weight is assigned as 0. The final feature vector is the two feature matrices represented above.

[0096] Each feature vector obtained is input into a neural network model to determine whether the user number needs to be monitored. For example, if the result obtained based on the neural network model is greater than a set value, it is considered that the user number needs to be monitored, otherwise it is considered that the user number does not need to be monitored.

[0097] Figure 4 A principle diagram for data processing by a neural network model is provided for the embodiments of the present application. As shown inFigure 4 As shown, the leftmost column of rectangles in the diagram is divided into 7×5 grids, representing feature vectors. Each row contains word vectors, with a dimension of 5. Neural network models can have multiple layers. After being input into the neural network model, the feature vectors are processed with corresponding parameters at each layer. The resulting feature data then enters the next layer for further processing. The shaded rectangles in the diagram represent the convolutional kernels of the neural network or the feature data obtained from each layer. The arrows in the diagram indicate the direction of feature data transmission. After multiple layers of processing, the final output result is obtained.

[0098] Specifically, the shaded rectangles in the second column from the left represent convolutional kernels. These kernels perform calculations on the sentence in stages: two kernels with a window size of 2, two kernels with a window size of 3, and two kernels with a window size of 4. Each shaded matrix in the second column outputs a "channel," meaning each kernel can output two channels. After pooling, we obtain six cells that effectively represent the features—that is, numbers. The final layer outputs the probability of each category.

[0099] Specifically, for user numbers that cannot be directly identified as meeting the criteria for blacklists or whitelists, the call signaling of the user number is obtained. This call signaling is then used to generate a corresponding feature vector through a bag-of-words model and feature weights. This feature vector is input into the pre-trained neural network model, and through convolutional calculations in a multi-layer neural network model, the probabilities of needing monitoring and not needing monitoring are obtained. For example, if the set value is 0.5, and the probability of needing monitoring is greater than 0.5 while the probability of not needing monitoring is less than 0.5, then the user number is monitored.

[0100] In one optional implementation, assuming the user's number is the first number they registered with operator A, this user is a key user to monitor. Information such as the location of the number (e.g., if the account was opened in city B and then moved to another city), whether the last few digits of consecutive calls are consecutive, call frequency, and characteristics of the calling numbers are input into the aforementioned neural network model for data processing. After convolutional calculations through multiple layers of neural networks, a comprehensive determination is made to obtain the probability that monitoring is required. If the probability that monitoring is required is higher than a set value, the user's number is added to the monitoring list, and the user's calling behavior is controlled, such as disallowing calls or allowing only 50% of calls.

[0101] Optionally, when a number initiates a call, it can be determined whether the calling number is in a blacklist. If it is in the blacklist, the tag corresponding to the calling number is searched; the policy corresponding to the tag is searched, and the call initiated by the calling number is processed according to the corresponding policy. Processing the call using the policy corresponding to the calling number tag is simple and fast, improving call processing efficiency and enhancing user experience.

[0102] Optionally, processing the call initiated by the calling number according to the corresponding strategy can include: if the label corresponding to the calling number is a high-risk label, prohibiting the calling number from making calls; and if the label corresponding to the calling number is a low-risk label, generating and sending a reminder message and dynamically adjusting the number of calls made by the calling number within a unit time.

[0103] The high-risk label and the low-risk label can be set according to actual needs. Optionally, the high-risk label can include scam calls and the like, and such user numbers have a great impact on other users, especially causing some property losses. Therefore, the label corresponding to the scam calls is a high-risk label, and the calling of the user number corresponding to such label is prohibited. The low-risk label can include real estate marketing, education and training, loan sales, and the like. Such user numbers can have some impact on other users, but not all of the impact is negative. Therefore, the label corresponding to such numbers is a low-risk label, and the calling number of such calling user can be reminded that the label is a low-risk label and the number of calls has been limited, so that the calling user can make calls according to the situation.

[0104] When the number of calls made by the user number corresponding to the low-risk label within a unit time is limited, the number of calls can be dynamically adjusted. If the calling frequency of the user number in the recent period (which can be one day or two days as a cycle) does not exceed the limited number of calls, the number of calls within a unit time of the user number is increased. If the calling frequency of the user number exceeds the limited number of calls by more than a threshold value, the number of calls is further limited. That is, the number of calls within a unit time of the low-risk label number can be adjusted in real time.

[0105] In this embodiment, when a user number is determined to be a blacklisted number, the user number is labeled. When the user number makes a call, the system automatically finds the label corresponding to the number and processes the call according to the strategy corresponding to the label.

[0106] For example, if a user number has been labeled as a scam call, when the user number makes a call, the system automatically identifies the label of the user number as a scam, and the corresponding strategy of the scam label is to prohibit the call. Therefore, the system blocks the call of the user number. If the label of the user number is an advertisement, when the number makes a call, the system identifies the label of the advertisement, and the corresponding strategy of the label is to first remind the processing and then dynamically adjust the calling frequency per hour. Therefore, the user is reminded that the number has been listed as an advertisement, and the number of calls within a unit time is limited to reduce the calling frequency of the number.

[0107] According to different labels, a label corresponding strategy is adopted to perform certain behavior restriction on the black list number, so as to flexibly adjust the management and control of the black list user, and meet the number monitoring demand of different industries. Through intelligent analysis of the call behavior and content of the user, an interception strategy is automatically generated, manual intervention is avoided, user privacy is prevented from being leaked, and legal risks are eliminated, so that the monitoring of the interception system can be more comprehensive, the management and control can be more accurate, and the disposal can be faster.

[0108] Figure 5 A structural schematic diagram of a user number processing device provided by the embodiment of the application is shown in the figure. Figure 5 As shown in the figure, the user number processing device provided by the embodiment can include:

[0109] The acquisition module 501 is configured to acquire number basic information of a user number to be analyzed.

[0110] The judgment module 502 is configured to judge whether the number basic information meets a black list condition or a white list condition, and if not, acquire call signaling data corresponding to the user number, and judge whether the user number needs to be monitored based on a neural network model according to the call signaling data.

[0111] The processing module 503 is configured to add the user number to a black list and determine a corresponding label for the user number when it is determined that the user number needs to be monitored, so as to intercept or pass through the call of the user number according to the strategy corresponding to the label.

[0112] Optionally, when judging whether the user number needs to be monitored based on the neural network model according to the call signaling data, the judgment module 502 is specifically configured to:

[0113] generate a corresponding feature matrix through a bag-of-words model and feature weight calculation according to the call signaling data;

[0114] input the feature matrix into the neural network model to obtain the result of whether the user number needs to be monitored.

[0115] Optionally, when generating a corresponding feature matrix through a bag-of-words model and feature weight calculation according to the call signaling data, the judgment module 502 is specifically configured to:

[0116] generate a corresponding word set according to the call signaling data;

[0117] generate a feature matrix corresponding to the call signaling data according to a preset bag-of-words model;

[0118] calculate the feature weight corresponding to each element in the feature matrix, and obtain a feature vector corresponding to the call signaling data according to the calculated feature weight and the feature matrix.

[0119] Optionally, the bag-of-words model comprises a plurality of attribute information for representing the call signaling data.

[0120] The judgment module 502 is specifically configured to, when generating the feature matrix corresponding to the call signaling data according to the preset bag-of-words model, generate the feature matrix corresponding to the call signaling data according to the selected words.

[0121] The judgment module 502 is specifically configured to, when generating the feature matrix corresponding to the call signaling data according to the preset bag-of-words model, generate the feature matrix corresponding to the call signaling data according to the selected words.

[0122] The plurality of attribute information in the bag-of-words model comprises longitude and latitude information at the time of the call, whether the call is continuous, and number basis information of the calling number.

[0123] Optionally, the processing module 503 is further configured to:

[0124] The processing module 503 is specifically configured to, when processing the call initiated by the calling number according to the corresponding strategy, judge whether the calling number initiating the call is in the blacklist, and if so, find the label corresponding to the calling number.

[0125] The processing module 503 is specifically configured to, when processing the call initiated by the calling number according to the corresponding strategy, find the strategy corresponding to the label and process the call initiated by the calling number according to the corresponding strategy.

[0126] Optionally, the processing module 503 is specifically configured to, when processing the call initiated by the calling number according to the corresponding strategy, if the label corresponding to the calling number is a high-risk label, prohibit the calling number from making an outgoing call.

[0127] Optionally, the processing module 503 is specifically configured to, when processing the call initiated by the calling number according to the corresponding strategy, if the label corresponding to the calling number is a low-risk label, generate and send a reminder information, and dynamically adjust the limited outgoing call quantity of the calling number within a unit time.

[0128] Optionally, the processing module 503 is specifically configured to, when processing the call initiated by the calling number according to the corresponding strategy, if the label corresponding to the calling number is a low-risk label, generate and send a reminder information, and dynamically adjust the limited outgoing call quantity of the calling number within a unit time.

[0129] Optionally, the call signaling data comprises at least one of the following: user's opening account information, user's behavior information and user's call information.

[0130] The apparatus provided in the embodiment can execute the technical solutions of the method embodiments as shown in Figures 1-4 The implementation principles and technical effects are similar, and will not be repeated here.

[0131] Figure 6 A structural schematic diagram of an electronic device provided in the embodiment of the application is shown in Figure 6 The device provided in the embodiment can comprise at least one processor 61 and a memory 62.

[0132] The memory 62 stores computer execution instructions.

[0133] The at least one processor 61 executes computer-executed instructions stored in the memory 62, so that the at least one processor 61 executes the method described in any of the above embodiments.

[0134] The memory 62 and the processor 61 can be connected through the bus 63.

[0135] The specific implementation principles and effects of the device provided in the embodiment can be referred to Figures 1-4 the related descriptions and effects of the embodiments shown in the drawings, which will not be described in detail here.

[0136] Figure 7 The overall architecture diagram of an abnormal call identification and interception system provided in the embodiment. The system in the embodiment can be used to implement the method described in any of the above embodiments. As shown in the figure, Figure 7 The system in the embodiment can include a system management module, a detection and identification subsystem, an interception and processing subsystem, a billing system, and a core network.

[0137] The artificial intelligence identification and detection module, the data analysis management module, and the billing subscription module can be summarized as the detection and identification subsystem. The label management module, the policy management module, and the signaling collection and monitoring processing module can be summarized as the interception and processing subsystem.

[0138] The system management module can assign tasks to the artificial intelligence identification and detection module, the data analysis management module, and the billing subscription module.

[0139] When obtaining the number basic information of the user number, the existing billing system TCP / IP interface can be connected, the billing subscription module sends a request to the billing system, synchronizes information, and can synchronize all or part of the number basic information of the existing network users to the database in real time.

[0140] After the synchronization of the number basic information is completed, the billing subscription module synchronizes data to the data analysis management module, the data analysis management module judges the number basic information, judges whether it meets the black list or white list conditions, removes the user number that meets the white list condition from the system, and controls the user that meets the black list.

[0141] For the user number that cannot be judged, the billing system module sends data to the core network. The core network transmits the signaling to the signaling collection and monitoring processing module, obtains the call signaling data corresponding to the user number in the signaling collection and monitoring processing module, and converts the call signaling data into a statement that can be understood by the computer and inputs it to the artificial intelligence identification and detection module for judgment.

[0142] After the artificial intelligence recognition detection module judges, the number is transferred to the label management module for classification processing, the label management module is transferred to the strategy management module to complete the matching of the corresponding strategy, and after the matching is completed, the strategy management module completes the management and control of the number through the signaling collection monitoring processing module, that is, when the label number is calling, the corresponding strategy is intercepted or passed through.

[0143] The embodiment of the application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the user number processing method provided by any of the embodiments of the application.

[0144] The embodiment of the application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the user number processing method described in any of the embodiments of the application.

[0145] In several embodiments provided by the present application, it should be understood that the disclosed device and method can be implemented in other manners. For example, the described device embodiment is only schematic. For example, the division of the modules is only a logical function division. There can be another division manner for the actual implementation. For example, a plurality of modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or modules, and can be electrical, mechanical or in other forms.

[0146] The modules illustrated as separated components can or can not be physically separated, and the components illustrated as modules can or can not be physical units, that is, can be located in one place, or can be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to implement the embodiments of the present application.

[0147] In addition, each function module in each embodiment of the present application can be integrated in one processing unit, or each module can be physically present alone, or two or more modules can be integrated in one unit. The unit formed by the above modules can be realized in the form of hardware, or in the form of hardware plus software function unit.

[0148] The integrated module realized in the form of software function module can be stored in a computer readable storage medium. The software function module stored in the storage medium includes a plurality of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute part of the steps of the method described in each embodiment of the present application.

[0149] It should be appreciated that the above processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the application can be directly embodied as hardware processor execution, or executed by a combination of hardware and software modules in the processor.

[0150] The memory can include a high-speed RAM memory, and can also include a non-volatile storage NVM, such as at least one disk memory, and can also be a U disk, a mobile hard disk, a read-only memory, a magnetic disk or an optical disk, etc.

[0151] The bus can be an industry standard architecture (ISA) bus, a peripheral component interconnect (PCI) bus, or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, the bus in the drawings of the present application does not limit to only one bus or one type of bus.

[0152] The above storage medium can be realized by any type of volatile or non-volatile storage device or their combination, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0153] An exemplary storage medium is coupled to the processor, so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and the storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the storage medium can also exist as discrete components in an electronic device or a host device.

[0154] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The foregoing program can be stored in a computer readable storage medium. The program executes the steps of the above-mentioned method embodiments when executed; and the foregoing storage medium includes: ROM, RAM, magnetic disk or optical disk and various media that can store program codes.

[0155] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims.

[0156] It is to be understood that the application is not limited to the precise details of design and construction that have been described and exemplified above and that various modifications and changes can be made by those skilled in the art without departing from the scope of the application. The scope of the application is limited only by the claims that follow.

Claims

1. A method of handling a subscriber number, characterized by, The method comprises the following steps: acquiring the number basic information of the user number to be analyzed, the number basic information comprising the user name, the ID number, the handling time, the handling channel, the handling location, the handling package, the number starting location, and the reported situation; determining whether the number basic information meets the blacklist condition or the whitelist condition, if not, acquiring the call signaling data corresponding to the user number, the call signaling data comprising at least one of the following: the user's opening account information, the user's behavior information, and the user's call information, and generating the corresponding feature vector through the bag-of-words model and feature weight calculation according to the call signaling data; inputting the feature vector into the neural network model to obtain the result of whether the user number needs to be monitored; if it is determined that the user number needs to be monitored, adding the user number to the blacklist and determining the corresponding label for the user number, wherein the label comprises a high-risk label and a low-risk label, and intercepting or passing the call of the user number according to the corresponding strategy of the label; further comprising determining whether the calling number initiating the call is in the blacklist, if so, searching for the label corresponding to the calling number; searching for the corresponding strategy of the label and processing the call initiated by the calling number according to the corresponding strategy, if the label corresponding to the calling number is a high-risk label, the calling number is prohibited from making an outgoing call; if the label corresponding to the calling number is a low-risk label, the reminding information is generated and sent, and the number of outgoing calls of the calling number in a unit time is dynamically adjusted.

2. The method of claim 1, wherein, According to the call signaling data, determining whether the user number needs to be monitored based on the neural network model, comprising: generating the corresponding feature vector through the bag-of-words model and feature weight calculation according to the call signaling data; inputting the feature vector into the neural network model to obtain the result of whether the user number needs to be monitored.

3. The method of claim 2, wherein, According to the call signaling data, generating the corresponding feature vector through the bag-of-words model and feature weight calculation, comprising: generating a corresponding word set according to the call signaling data; generating a feature matrix corresponding to the call signaling data according to the preset bag-of-words model; calculating the feature weight corresponding to each element in the feature matrix, and obtaining the feature vector corresponding to the call signaling data according to the calculated feature weight and the feature matrix.

4. The method of claim 3, wherein, The bag-of-words model comprises a plurality of attribute information for representing the call signaling data; According to the preset bag-of-words model, generating a feature matrix corresponding to the call signaling data, comprising: selecting the words corresponding to the attribute information in the bag-of-words model from the word set, and generating a feature matrix corresponding to the call signaling data according to the selected words; wherein the plurality of attribute information in the bag-of-words model comprises: the latitude and longitude information at the time of the call, whether it is a continuous call, and the number basic information of the calling number.

5. A subscriber number handling apparatus, characterized by The device comprises: an acquisition module for acquiring the number basic information of the user number to be analyzed, the number basic information comprising the user name, the ID number, the handling time, the handling channel, the handling location, the handling package, the number starting location, and the reported situation; The judgment module is configured to judge whether the number base information meets a blacklist condition or a whitelist condition, and if not, acquire call signaling data corresponding to the user number, the call signaling data including at least one of the following: user opening account information, user behavior information, and user call information, and generate a corresponding feature vector through a bag-of-words model and feature weight calculation according to the call signaling data; input the feature vector into a neural network model to obtain a result of whether the user number needs to be monitored. The processing module is configured to add the user number to a blacklist and determine a corresponding label for the user number when it is determined that the user number needs to be monitored, wherein the label includes a high-risk label and a low-risk label, and intercept or pass through a call of the user number according to a strategy corresponding to the label. The processing module is further configured to judge whether a calling number initiating a call is in the blacklist, and if so, find a label corresponding to the calling number; find a strategy corresponding to the label, and process the call initiated by the calling number according to the corresponding strategy. The processing module is configured to, when processing the call initiated by the calling number according to the corresponding strategy, specifically prohibit the calling number from making an outgoing call if the label corresponding to the calling number is a high-risk label, and generate and send a reminder information and dynamically adjust the number of outgoing calls of the calling number within a unit time if the label corresponding to the calling number is a low-risk label.

6. An electronic device, comprising: At least one processor and a memory; The memory stores computer execution instructions; The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the user number processing method in any one of claims 1-4. The computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the user number processing method in any one of claims 1-4.

7. A computer readable storage medium characterized in that, ​

Images