Private number call efficiency detection method and system based on private communication platform

Abstract

The invention belongs to the technical field of privacy communication, and provides a privacy number call efficiency detection method and system based on a privacy communication platform. According to the method, a self-constructed privacy number platform delay factor detection model comprises a delay type classification model and an influence factor judgment model connected in series with the delay type classification model; and inputting the total delay time, the upstream delay, the search delay, the downstream delay and the network type in the collected current ticket data into the privacy number platform delay factor detection model to obtain the delay type of the current ticket data and the maximum influence factor influencing the sub-platform call efficiency in the current ticket data. According to the method, multiple factors in the call process are comprehensively considered, the maximum influence factor influencing the sub-platform call efficiency in the current call ticket data can be accurately determined, the occurrence area of the call quality problem and the sub-platform can be accurately positioned, and data support is provided for optimization suggestions.

Description

Technical Field

[0001] The present invention relates to the field of privacy communication technology, and in particular to a method and system for detecting call efficiency of a privacy number based on a privacy communication platform. Background Art

[0002] With the continuous development of communication technology, privacy protection has become an important issue that users are increasingly concerned about. The intermediate number service launched by China Unicom effectively protects users' real number information and avoids the risk of information leakage by providing a temporary, anonymous intermediate number for both parties of the call. However, while providing privacy protection, call quality has also become the focus of users. Problems such as call delays and unstable signals not only affect user experience, but may also cause significant losses in scenarios such as business cooperation and emergency communications.

[0003] The existing technology has the following problems: it is only a single-dimensional evaluation. The existing call efficiency evaluation technology often only focuses on the physical quality of the call (such as signal strength, delay time) or a single user feedback, ignoring the complex interactive characteristics and information transmission efficiency during the privacy number call. This single-dimensional evaluation method is difficult to fully reflect the true efficiency of the call; lack of intelligent analysis, many existing systems rely on manually set standards or simple rules to judge call efficiency, lack of intelligent analysis and adaptive capabilities, and this static evaluation method is increasingly difficult to meet actual needs; lack of geographic location analysis, geographic location information is of great significance for locating the source of call quality problems, but the existing technology lacks effective analysis and utilization of geographic location information, resulting in the inability to accurately locate the area where call quality problems occur, thereby increasing the difficulty of solving the problem.

[0004] In summary, the existing technology cannot achieve comprehensive and efficient detection of platform privacy number call quality, cannot evaluate call quality in real time and accurately, and locate the source of call quality problems. Therefore, it is necessary to develop a new method that can monitor call delays in real time, analyze geographic location data, and accurately locate call quality problems. The privacy number call quality detection method and system has important practical significance and application value. Summary of the invention

[0005] The present invention intends to provide a method and system for detecting call efficiency of a privacy number based on a privacy communication platform, so as to solve the technical problems in the prior art that the existing methods cannot evaluate the call quality in real time and accurately, and cannot accurately locate the factors affecting the call quality problems. The technical problems to be solved by the present invention are achieved through the following technical solutions.

[0006] The first aspect of the present invention proposes a method for detecting the efficiency of privacy number calls based on a privacy communication platform, comprising: collecting call list data of historical calls from the privacy communication platform, specifically including the total delay time from the call initiation of the call by the calling number to the start of the ringing, the upstream delay time, the search delay time, the downstream delay time, the network type, the province information, and the platform information during the call; the privacy communication platform includes multiple sub-platforms that can communicate with each other; constructing a privacy number platform delay factor detection model, the privacy number platform delay factor detection model includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model; classifying the delay type of each collected call list data, performing multiple classifications according to the influencing factors of the delay time of each call and marking the delay category label to establish a first training data set; using the first training data set to train the delay classification model A delay type classification model is established, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call list data to determine the delay type it belongs to; the CART algorithm is used to construct an influencing factor judgment model; the construction of the influencing factor judgment model includes: dividing the factor nodes according to the number of optional values ​​of the feature, selecting the optimal feature and the optimal segmentation point, until all nodes are split, and completing the construction of the influencing factor judgment model; real-time collection of the total delay time, upstream delay, search delay, downstream delay, network type, sub-platform code, and province code in the current call list data, and the total delay time, upstream delay, search delay, downstream delay, and network type collected in the current call list data are input into the privacy number platform delay factor detection model to obtain the delay category of the current call list data and the maximum influencing factor affecting the call efficiency of the sub-platform in the current call list data.

[0007] The second aspect of the present invention proposes a privacy number call efficiency detection system based on a privacy communication platform, which adopts the privacy number call efficiency detection method based on a privacy communication platform described in the first aspect of the present invention. The privacy communication anomaly detection system includes: a data acquisition module, which is used to collect call data from the privacy communication platform during historical calls, specifically including the total delay time from the call initiation of the call by the calling number to the start of the ringing, the upstream delay time, the search delay time, the downstream delay time, the network type, the province information, and the platform information during the call; the privacy communication platform includes multiple sub-platforms that can communicate with each other; a model construction module is used to construct a privacy number platform delay factor detection model, and the privacy number platform delay factor detection model includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model; each collected call list data is classified into a delay type, and multiple classifications are performed according to the influencing factors of the delay time of each call and the delay category labels are marked. Establish a first training data set; use the first training data set to train the delay type classification model, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call list data to determine the delay type it belongs to; use the CART algorithm to construct an influencing factor judgment model; the construction of the influencing factor judgment model includes: dividing the factor nodes according to the number of optional values ​​​​of the feature, selecting the optimal feature and the optimal segmentation point until all nodes are split, and completing the construction of the influencing factor judgment model; determine a module for real-time collection of the total delay time, upstream delay, search delay, downstream delay, network type, sub-platform code, and province code in the current call list data, and input the total delay time, upstream delay, search delay, downstream delay, and network type collected in the current call list data into the privacy number platform delay factor detection model to obtain the delay category of the current call list data and the maximum influencing factor affecting the call efficiency of the sub-platform in the current call list data.

[0008] The third aspect of the present invention provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the privacy number call efficiency detection method based on the privacy communication platform described in the first aspect of the present invention.

[0009] The fourth aspect of the present invention provides a computer-readable medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting the efficiency of private number calls based on a private communication platform as described in the first aspect of the present invention is implemented.

[0010] The embodiments of the present invention include the following advantages:

[0011] Compared with the existing technology, the self-built privacy number platform delay factor detection model can obtain the delay category of the current call bill data and the biggest influencing factor of the sub-platform call efficiency in the current call bill data. It can evaluate the call quality in real time and accurately, and can accurately locate the influencing factors of call quality problems.

[0012] In addition, by real-time monitoring and recording of various delay data from call initiation to ringing start, it is possible to accurately determine the most important delay factor in each call and the factor that has the greatest impact on each sub-platform, providing a solid foundation for call quality evaluation.

[0013] In addition, by using the sub-platform code and province code in the call record data, combined with the called party's geographic location information, the call quality is deeply analyzed. By correlating the delay time data and geographic location information, the area and sub-platform where the call quality problem occurs can be accurately located, providing data support for optimization suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flowchart of an example of a method for detecting the efficiency of private number calls based on a private communication platform of the present invention;

[0015] Figure 2 It is a schematic diagram of an application example of the privacy number call efficiency detection method based on the privacy communication platform of the present invention;

[0016] Figure 3 It is a structural block diagram of a privacy number call efficiency detection system based on a privacy communication platform of the present invention;

[0017] Figure 4 is a schematic structural diagram of an electronic device embodiment according to the present invention;

[0018] Figure 5 It is a schematic diagram of the structure of a computer-readable medium embodiment according to the present invention. DETAILED DESCRIPTION

[0019] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] In view of the above problems, the present invention proposes a privacy number call efficiency detection method and system based on a privacy communication platform. The method classifies the delay type of the call data through a self-constructed privacy number platform delay factor detection model; the total delay time, upstream delay, search delay, downstream delay, network type and the delay category of the current call data obtained in the current call data are input into the influencing factor judgment model to obtain the maximum influencing factor affecting the call efficiency of the sub-platform in the current call data. By comprehensively considering multiple factors in the call process, a comprehensive evaluation is constructed.

[0021] In addition, by real-time monitoring and recording of various delay data from call initiation to ringing start, it is possible to accurately determine the most important delay factor in each call and the factor that has the greatest impact on each platform, providing a solid foundation for call quality evaluation.

[0022] In addition, the call quality is deeply analyzed by combining the called party’s geographic location information. By correlating the delay time data and geographic location information, the area and sub-platform where the call quality problem occurs can be accurately located, providing data support for optimization suggestions.

[0023] Example 1

[0024] Refer to the following Figure 1 and Figure 2 , the contents of the present invention will be described in detail.

[0025] Figure 1 This is a step flow chart of an example of a method for detecting the efficiency of private number calls based on a private communication platform of the present invention.

[0026] like Figure 1 As shown, in step S101, the call data of the historical call process is collected from the privacy communication platform, specifically including the total delay time from the call initiation of the call by the calling number to the start of the ringing, the upstream delay time, the search delay time, the downstream delay time, the network type, the province information, and the platform information; the privacy communication platform includes multiple sub-platforms that can communicate with each other. Specifically, the platform information includes the privacy communication platform information and the sub-platform information, such as the sub-platform code.

[0027] In a specific application example, Figure 2 As shown, the privacy communication platform includes five sub-platforms that can communicate with each other, each sub-platform corresponds to multiple users, and one or more privacy numbers bound to each user.

[0028] Specifically, call record data of historical calls are collected from the privacy communication platform, including the total delay time from the time the call is initiated by the calling number to the start of the ringing, the upstream delay time (specifically the time from the caller number sending the message to the privacy number), the search delay time (specifically the time to find the called party with the corresponding binding relationship), the downstream delay time (the time from the start of the call to the called party to the called party ringing), the network type (such as mobile phone 2G, 3G, 4G, 5G networks, IP networks, solidified traditional telephone networks, digital telephone networks, etc.), provincial information (such as the corresponding sub-platform, province code), and platform information (such as whether it is an available sub-platform, sub-platform code, etc.).

[0029] Furthermore, the call record data collected during the historical calls are pre-processed.

[0030] Data preprocessing specifically includes: using the mean to fill in the missing values ​​of various delay times in the call list data, and deleting outliers such as various delay times.

[0031] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0032] Next, in step S102, a privacy number platform delay factor detection model is constructed, and the privacy number platform delay factor detection model includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model; each collected call sheet data is classified into a delay type, and multiple classifications are performed according to the influencing factors of the delay time of each call and the delay category labels are marked to establish a first training data set; the delay type classification model is trained using the first training data set, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call sheet data to determine the delay type to which it belongs; the CART algorithm is used to construct an influencing factor judgment model; the construction of the influencing factor judgment model includes: dividing the factor nodes according to the number of optional values ​​of the feature, selecting the optimal feature and the optimal splitting point, until all nodes are split, and completing the construction of the influencing factor judgment model.

[0033] Specifically, a privacy number platform delay factor detection model is constructed, and the privacy number platform delay factor detection model includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model. Among them, the K-Nearest Neighbors (KNN) algorithm is used to construct the delay type classification model. The CART algorithm is used to construct the influencing factor judgment model (also known as the CART model).

[0034] For the construction of the delay type classification model, each collected call record data is specifically classified into a delay type, and multiple classifications are performed according to the influencing factors of the delay time of each call and the delay category labels are marked to establish the first training data set.

[0035] The historical voice call record data of the privacy communication platform is collected, and the following features are selected from it: the total delay time from the initiation of the call to the start of the ringing, upstream delay, search delay, downstream delay, network type, province area code, sub-platform code, and call record data are formed to further form a call record data set.

[0036] Each collected call record data is classified into delay types, and multiple classifications are performed according to the factors affecting the delay time of each call, and delay category labels are marked.

[0037] Specifically, the call record data set is divided into a first training set (for example, represented by D) and a test set (for example, represented by E), and a category label is added to each call record data. In this example, the delay type is divided into at least three categories of labels, including upstream impact category labels, downstream impact category labels, and search delay category labels.

[0038] Specifically, when the ratio of the upstream delay to the total delay exceeds the first threshold, it is defined as an upstream impact category label; when the ratio of the downstream delay to the total delay exceeds the second threshold, it is defined as a downstream impact category label; when the ratio of the search delay to the total delay exceeds the third threshold, it is defined as a search impact category label; or when the time for searching the called number with a corresponding binding relationship with the privacy number exceeds the time threshold, it is defined as a search delay category label.

[0039] Specifically, a call record data with a delay type tag is as follows:

[0040] D i =[O i ,S i ,F i ,T i ,C i ,N i ,P i ,y i ]

[0041] Among them, D i represents the i-th call record data; P i is the sub-platform code, O i is the total delay time of the i-th call record data; S i is the upstream delay time in the i-th call record data; F i is the search delay time in the i-th call record data, T i is the downstream delay time in the i-th call record data, Ci N is the city code to which the privacy number in the i-th call record belongs; i is the network type of the i-th call record data, y i is the delay category to which the i-th call record data belongs, where i is a positive integer, specifically 1, 2, ..., n.

[0042] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0043] Next, the delay type classification model is trained using the first training data set, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call list data to determine the delay type it belongs to.

[0044] Specifically, the distance between the call record data to be classified and each training sample in the first training data set is calculated based on the distance metric, and the data is classified into the nearest class, that is, classified into the class with the closest distance. It does not depend on the distribution form of the data and is more suitable for classifying privacy numbers according to delay time. Therefore, the KNN algorithm is used to construct the privacy number platform delay factor detection model of the present invention.

[0045] The total delay time, upstream delay, search delay, downstream delay, network type and classification label of each data in the first training data set D are used as input features of KNN.

[0046] The distance metric in the KNN algorithm can be Euclidean distance, Manhattan distance, Chebyshev distance, Minkowski distance, etc.

[0047] In this example, Euclidean distance is used as the distance measurement method. The specific expression is as follows:

[0048] The distance between each training sample in the first training set and the central sample of any delay type is calculated using the following formula:

[0049]

[0050] Where d(a,b) represents the Euclidean distance between any training sample a in the first training set and the central sample b of any delay type, a i and b i are the values ​​of any training sample a and the central sample b of any delay type on the i-th feature, i is a positive integer, specifically 1, 2, ...n, and n is the number of input features of the delay type classification model.

[0051] Through the above distance calculation, three sets corresponding to the upstream impact category label, the downstream impact category label, and the search delay category label are formed respectively.

[0052] Preferably, outlier removal is performed while forming three sets corresponding to the upstream impact category labels, the downstream impact category labels, and the search delay category labels, respectively.

[0053] Next, the construction process of the influencing factor judgment model will be explained.

[0054] The CART algorithm is used to construct an influencing factor judgment model, which includes: dividing the factor nodes according to the number of optional values ​​of the feature, selecting the optimal feature and the optimal segmentation point, until all nodes are split, and completing the construction of the influencing factor judgment model.

[0055] Factor nodes are divided according to the number of optional values ​​of the feature.

[0056] The factor nodes are divided specifically according to the number of optional values ​​of the following features: platform code, sub-platform code, and delay category of each call record data. The optional value specifically refers to the value of each input feature.

[0057] The weighted sum of the Gini impurity of the influencing factor judgment model is further repeatedly calculated to perform multiple divisions of the data set, and the features and segmentation points that minimize the calculated weighted sum of the Gini impurity are selected as the optimal division of the current node.

[0058] The following expression is used to calculate the weighted sum of the Gini impurity of the influencing factor judgment model, and the first division of the data set is performed to obtain the first data set D1 and the second data set D2:

[0059]

[0060] Among them, Gini(D) represents the weighted sum of Gini impurity of the influencing factor judgment model; C k is the sample set belonging to the kth class in the call record dataset D, |C k |It is C k The number of samples, k is a positive integer, specifically 1, 2, ..., K.

[0061] After the first partition of the data set, the following expression is used to calculate the weighted sum of the Gini impurity after the partition for subsequent partitioning of the data set:

[0062]

[0063] Gini split(D, A, a) represents the weighted sum of the Gini impurity of the influencing factor judgment model after segmentation based on the optimal features and the optimal segmentation points; D represents the call bill data set; A represents the input features of the delay factor detection model of the privacy number platform; a represents the optional value of each feature, which is determined based on each input feature; D1 represents the first data set obtained after the first division; D2 represents the second data set obtained after the first division.

[0064] It should be noted that the specific segmentation is the division of the data set, that is, data partitioning or data segmentation, from which sample to start dividing the data set.

[0065] The selection of the optimal feature and the optimal split point includes: traversing all features and all split points, and selecting the feature and split point that minimize the weighted Gini impurity after splitting as the optimal split of the current node. For example, the first training data set D is split according to the category feature "upstream delay", two child nodes are generated, and feature splitting is recursively selected for the two child nodes until the split child node data set has only one feature.

[0066] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0067] Next, in step S103, the total delay time, upstream delay, search delay, downstream delay, network type, sub-platform code, and province code in the current call bill data are collected in real time, and the total delay time, upstream delay, search delay, downstream delay, and network type collected in the current call bill data are input into the privacy number platform delay factor detection model to obtain the delay category of the current call bill data and the biggest influencing factor affecting the sub-platform call efficiency in the current call bill data.

[0068] Based on the call record data collected in real time, the delay type classification model and influencing factor judgment model of the trained privacy number platform delay factor detection model are used to determine the delay category of the current call record data and the factors that have the greatest impact on the call efficiency of the sub-platform in the current call record data.

[0069] Specifically, the trained delay type classification model is used to calculate the distance between each sample in the first training set and the sample to be classified. According to the calculated distance, k samples are selected as the nearest neighbors of the sample to be classified in order of distance, and the value of k is selected according to the size of the training set, and the value of k is 3 to 8.

[0070] For the selected k nearest neighbors, the delay categories to which they belong are counted, and the category with the largest number of occurrences is used as the classification category of the call record data to be classified (ie, the sample to be classified).

[0071] Optionally, the value of k is 5.

[0072] The following formula is used to calculate the distance between each training sample in the first training set and the sample to be classified:

[0073]

[0074] Where d(x,c) represents the Euclidean distance between any training sample x in the first training set and the sample to be classified c, x i and c i are the values ​​of the training sample x and the sample to be classified c on the i-th feature, respectively. i is a positive integer, specifically 1, 2, ...n, where n is the number of input features of the delay type classification model.

[0075] Through the above model calculation, the category of the training sample that is the nearest neighbor to the current call record data is determined to obtain the delay category of the current call record data.

[0076] Next, the sub-platform code, province code in the current call bill data and the delay category of the current call bill data obtained by the delay type classification model are input into the influencing factor judgment model to obtain the greatest influencing factor affecting the sub-platform call efficiency in the current call bill data.

[0077] By using the sub-platform code and province code in the call data, that is, combining the geographical location information of the called party, the call quality is deeply analyzed. That is, by associating the geographical location, delay time, and delay category of the province where the sub-platform is located, the area and sub-platform where the call quality problem occurs can be accurately located, providing data support for optimization suggestions. The sub-platform code and province code in the call data are specifically used to further determine the factors that have the greatest impact on the call efficiency of the sub-platform, and the area and sub-platform where the call quality problem occurs can be accurately located, providing data support for optimization suggestions.

[0078] In other implementations, the delay category of the current call list data and the sub-platform corresponding to the privacy number may also be obtained. The above is only described as a preferred example and should not be construed as a limitation to the present invention.

[0079] It should be noted that, in this example, n is the number of input features of the privacy number platform delay factor detection model. The call efficiency detection method of this application includes a privacy number platform delay factor detection model (i.e., a classification model, i.e., KNN) and an influencing factor judgment model (i.e., a CART model). The two models are in series, and the input of each model is different. The above is only explained as an optional example and cannot be understood as a limitation of the present invention.

[0080] In addition, the drawings are only schematic illustrations of the processes included in the method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It is easy to understand that the processes shown in the drawings do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0081] Compared with the existing technology, the self-built privacy number platform delay factor detection model can obtain the delay category of the current call bill data and the biggest influencing factor of the sub-platform call efficiency in the current call bill data. It can evaluate the call quality in real time and accurately, and can accurately locate the influencing factors of call quality problems.

[0082] In addition, by real-time monitoring and recording of various delay data from call initiation to ringing start, it is possible to accurately determine the most important delay factor in each call and the factor that has the greatest impact on each sub-platform, providing a solid foundation for call quality evaluation.

[0083] In addition, by using the sub-platform code and province code in the call record data, combined with the called party's geographic location information, the call quality is deeply analyzed. By correlating the delay time data and geographic location information, the area and sub-platform where the call quality problem occurs can be accurately located, providing data support for optimization suggestions.

[0084] Example 2

[0085] The following is a system embodiment of the present invention, which can be used to implement the method embodiment of the present invention. For details not disclosed in the system embodiment of the present invention, please refer to the method embodiment of the present invention.

[0086] Figure 3 It is a structural diagram of an example of a privacy number call efficiency detection system based on a privacy communication platform of the present invention.

[0087] Reference Figure 3 The second aspect of the present disclosure provides a privacy number call efficiency detection system based on a privacy communication platform, which adopts the privacy number call efficiency detection method based on a privacy communication platform described in the first aspect of the present invention.

[0088] Specifically, the privacy number call efficiency detection system 300 includes a data collection module 310 , a model building module 320 and a determination module 330 .

[0089] In a specific embodiment, the data collection module 310 is used to collect call record data of historical calls from the privacy communication platform, specifically including the total delay time from the time the call is initiated by the calling number to the start of the ringing, the upstream delay time, the search delay time, the downstream delay time, the network type, the province information, and the platform information; the privacy communication platform includes multiple sub-platforms that can communicate with each other. The model building module 320 is used to build a privacy number platform delay factor detection model, which includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model; each collected call sheet data is classified into a delay type, and multiple classifications are performed according to the influencing factors of the delay time of each call and the delay category labels are marked to establish a first training data set; the delay type classification model is trained using the first training data set, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call sheet data to determine the delay type to which it belongs; the CART algorithm is used to construct an influencing factor judgment model; the construction of the influencing factor judgment model includes: dividing the factor nodes according to the number of optional values ​​of the feature, selecting the optimal feature and the optimal split point, until all nodes are split, and completing the construction of the influencing factor judgment model. The determination module 330 is used to collect the total delay time, upstream delay, search delay, downstream delay, network type, sub-platform code, and province code in the current call bill data in real time, and input the collected total delay time, upstream delay, search delay, downstream delay, and network type in the current call bill data into the privacy number platform delay factor detection model to obtain the delay category of the current call bill data and the biggest influencing factor affecting the sub-platform call efficiency in the current call bill data.

[0090] According to an optional implementation manner, the factor node division according to the number of optional values ​​of the features includes: factor node division according to the number of optional values ​​of the following features: platform code, sub-platform code, delay category of each call record data; the optional value specifically refers to the value of each input feature.

[0091] The weighted sum of the Gini impurity of the influencing factor judgment model is further repeatedly calculated to perform multiple divisions of the data set, and the features and segmentation points that minimize the calculated weighted sum of the Gini impurity are selected as the optimal division of the current node.

[0092] According to an optional implementation method, the total delay time, upstream delay, search delay, downstream delay, and network type collected in the current call bill data are input into the privacy number platform delay factor detection model to obtain the delay category of the current call bill data and the biggest influencing factor affecting the call efficiency of the sub-platform in the current call bill data, including: inputting the total delay time, upstream delay, search delay, downstream delay, and network type collected in the current call bill data into the delay type classification model of the privacy number platform delay factor detection model, and outputting the delay category corresponding to the current call bill data.

[0093] Then, the sub-platform code, province code and delay category output by the delay type classification model in the current call bill data are input into the influencing factor judgment model to output the biggest influencing factor affecting the call efficiency of the sub-platform in the current call bill data.

[0094] According to an optional implementation manner, the method further includes: using the following expression to calculate the weighted sum of the Gini impurity of the influencing factor judgment model, performing a first division of the data set, and obtaining a first data set D1 and a second data set D2:

[0095]

[0096] Among them, Gini(D) represents the weighted sum of Gini impurity of the influencing factor judgment model; C k is the sample set belonging to the kth class in the call record dataset D, |C k |It is C k The number of samples, k is a positive integer, specifically 1, 2, ..., K.

[0097] After the first partition of the call record data set, the following expression is used to calculate the weighted sum of the Gini impurity after the partition for subsequent partitioning of the data set:

[0098]

[0099] Gini split (D, A, a) represents the weighted sum of the Gini impurity of the influencing factor judgment model after segmentation based on the optimal features and the optimal segmentation points; D represents the call bill data set; A represents the input features of the delay factor detection model of the privacy number platform; a represents the optional value of each feature, which is determined based on each input feature; D1 represents the first data set obtained after the first division; D2 represents the second data set obtained after the first division.

[0100] According to an optional implementation, the selection of the optimal feature and the optimal segmentation point includes: traversing all features and all segmentation points, and selecting the feature and segmentation point that minimize the weighted Gini impurity after segmentation as the optimal partition of the current node. For example, the first training data set is divided according to the category feature "upstream delay", two child nodes are generated, and feature partitioning is recursively selected for the two child nodes until the divided child node data set has only one feature.

[0101] According to an optional implementation manner, each collected call record data is classified into a delay type, and multiple classifications are performed according to the influencing factors of the delay time of each call and the delay category labels are marked.

[0102] The delay types are grouped into at least three categories of labels, including:

[0103] When the ratio of the upstream delay to the total delay exceeds the first threshold, it is defined as an upstream impact category label; when the ratio of the downstream delay to the total delay exceeds the second threshold, it is defined as a downstream impact category label;

[0104] When the ratio of the search delay to the total delay exceeds a third threshold, it is defined as a search impact category label.

[0105] According to an optional implementation, it further includes: calculating the distance between each sample in the first training set and the sample to be classified based on the calculated distance, selecting k samples as the nearest neighbors of the sample to be classified in order of distance, and selecting the value of k according to the size of the training set, and the value of k is 3 to 8.

[0106] For the selected k nearest neighbors, the delay categories to which they belong are counted, and the category with the largest number of occurrences is used as the classification category of the sample to be classified.

[0107] According to an optional implementation manner, the method further includes: using the following formula to calculate the distance between each training sample in the first training set and the sample to be classified:

[0108] The following formula is used to calculate the distance between each training sample in the first training set and the sample to be classified:

[0109]

[0110] Where d(x,c) represents the Euclidean distance between any training sample x in the first training set and the sample to be classified c, x i and c i are the values ​​of the training sample x and the sample to be classified c on the i-th feature, respectively. i is a positive integer, specifically 1, 2, ...n, where n is the number of input features of the delay type classification model.

[0112] Compared with the existing technology, the self-built privacy number platform delay factor detection model can obtain the delay category of the current call bill data and the biggest influencing factor of the sub-platform call efficiency in the current call bill data. It can evaluate the call quality in real time and accurately, and can accurately locate the influencing factors of call quality problems.

[0113] In addition, by real-time monitoring and recording of various delay data from call initiation to ringing start, it is possible to accurately determine the most important delay factor in each call and the factor that has the greatest impact on each sub-platform, providing a solid foundation for call quality evaluation.

[0114] In addition, by using the sub-platform code and province code in the call record data, combined with the called party's geographic location information, the call quality is deeply analyzed. By correlating the delay time data and geographic location information, the area and sub-platform where the call quality problem occurs can be accurately located, providing data support for optimization suggestions.

[0115] It should be noted that the privacy number call efficiency detection method executed by the privacy number call efficiency detection system based on the privacy communication platform in Example 2 is the same as the privacy number call efficiency detection method in Example 1, so the description of the same parts is omitted.

[0116] Figure 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

[0117] like Figure 4 As shown, the electronic device is presented in the form of a general computing device. The processor may be one or more and work in coordination. The present invention does not exclude distributed processing, that is, the processor may be dispersed in different physical devices. The electronic device of the present invention is not limited to a single entity, but may also be the sum of multiple physical devices.

[0118] The memory stores a computer executable program, which is usually a machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least part of the steps in the method.

[0119] The memory includes a volatile memory, such as a random access memory unit (RAM) and / or a cache memory unit, and may also be a non-volatile memory, such as a read-only memory unit (ROM).

[0120] Optionally, in this embodiment, the electronic device further includes an I / O interface, which is used for the electronic device to exchange data with an external device. The I / O interface can represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0121] It should be understood that Figure 4 The electronic device shown is only an example of the present invention, and the electronic device of the present invention may also include elements or components not shown in the above examples. For example, some electronic devices also include display units such as display screens, and some electronic devices also include human-computer interaction elements such as buttons, keyboards, etc. As long as the electronic device can execute the computer-readable program in the memory to implement the method of the present invention or at least part of the steps of the method, it can be considered as an electronic device covered by the present invention.

[0122] Through the above description of the implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by combining software with necessary hardware. Figure 5 As shown, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes a number of commands to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present invention.

[0123] The software product may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0124] The computer readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with a command execution system, device, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0125] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0126] The computer-readable medium carries one or more programs (eg, computer-executable programs). When the one or more programs are executed by a device, the computer-readable medium implements the method of the present disclosure.

[0127] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed accordingly and only used in one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0128] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several commands to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

[0129] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structure, configuration or implementation method described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. A method for detecting the call efficiency of a private number based on a private communication platform, characterized in that: include: Collecting call data from historical calls from the privacy communication platform, specifically including the total delay time from the call initiation of the call by the calling number to the start of the ringing, the upstream delay time, the search delay time, the downstream delay time, the network type, the province information, and the platform information during the call; the privacy communication platform includes multiple sub-platforms that can communicate with each other; Construct a privacy number platform delay factor detection model, the privacy number platform delay factor detection model includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model; classify each collected call list data into a delay type, perform multiple classifications according to the influencing factors of each call delay time and mark the delay category label to establish a first training data set; use the first training data set to train the delay type classification model, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call list data to determine the delay type to which it belongs; The CART algorithm is used to build an influencing factor judgment model; The construction of the influencing factor judgment model includes: dividing the factor nodes according to the number of optional values ​​of the feature, selecting the optimal feature and the optimal segmentation point, until all nodes are split, and the construction of the influencing factor judgment model is completed; The total delay time, upstream delay, search delay, downstream delay, network type, sub-platform code and province code in the current call bill data are collected in real time, and the total delay time, upstream delay, search delay, downstream delay and network type collected in the current call bill data are input into the privacy number platform delay factor detection model to obtain the delay category of the current call bill data and the biggest influencing factor affecting the sub-platform call efficiency in the current call bill data.

2. The privacy number call efficiency detection method based on the privacy communication platform according to claim 1 is characterized in that: The factor node division according to the number of optional values ​​of the feature includes: Factor nodes are divided according to the number of optional values ​​of the following features: platform code, sub-platform code, and delay category of each call record data; the optional value specifically refers to the value of each input feature; The weighted sum of the Gini impurity of the influencing factor judgment model is further repeatedly calculated to perform multiple divisions of the data set, and the features and segmentation points that minimize the calculated weighted sum of the Gini impurity are selected as the optimal division of the current node.

3. The privacy number call efficiency detection method based on the privacy communication platform according to claim 1 is characterized in that: The total delay time, upstream delay, search delay, downstream delay, and network type in the collected current call list data are input into the privacy number platform delay factor detection model to obtain the delay category of the current call list data and the biggest influencing factor affecting the call efficiency of the sub-platform in the current call list data, including: Input the total delay time, upstream delay, search delay, downstream delay, and network type in the collected current call bill data into the delay type classification model of the privacy number platform delay factor detection model, and output the delay category corresponding to the current call bill data; Then, the sub-platform code, province code and delay category output by the delay type classification model in the current call bill data are input into the influencing factor judgment model to output the biggest influencing factor affecting the call efficiency of the sub-platform in the current call bill data.

4. The privacy number call efficiency detection method based on the privacy communication platform according to claim 2 is characterized in that: Further including: The following expression is used to calculate the weighted sum of the Gini impurity of the influencing factor judgment model, and the first division of the data set is performed to obtain the first data set D1 and the second data set D2: Among them, Gini(D) represents the weighted sum of Gini impurity of the influencing factor judgment model; C k is the sample set belonging to the kth class in the call record dataset D, |C k |It is C k The number of samples, k is a positive integer, specifically 1, 2, ..., K; After the first partition of the call record data set, the following expression is used to calculate the weighted sum of the Gini impurity after the partition for subsequent partitioning of the data set: Gini split (D, A, a) represents the weighted sum of the Gini impurity of the influencing factor judgment model after segmentation based on the optimal features and the optimal segmentation points; D represents the call bill data set; A represents the input features of the delay factor detection model of the privacy number platform; a represents the optional value of each feature, which is determined based on each input feature; D1 represents the first data set obtained after the first division; D2 represents the second data set obtained after the first division.

5. The privacy number call efficiency detection method based on the privacy communication platform according to claim 1 is characterized in that: The selecting of the optimal features and the optimal segmentation points comprises: Traverse all features and all segmentation points, and select the features and segmentation points that minimize the weighted Gini impurity after segmentation as the optimal partition of the current node.

6. The privacy number call efficiency detection method based on the privacy communication platform according to claim 1 is characterized in that: The delay type classification of each collected call record data is carried out, and multiple classifications are carried out according to the factors affecting the delay time of each call and the delay category labels are marked, including: The delay types are grouped into at least three categories of labels, including: When the ratio of the upstream delay to the total delay exceeds the first threshold, it is defined as an upstream impact category label; when the ratio of the downstream delay to the total delay exceeds the second threshold, it is defined as a downstream impact category label; When the ratio of the search delay to the total delay exceeds a third threshold, it is defined as a search impact category label.

7. The privacy number call efficiency detection method based on the privacy communication platform according to claim 1 is characterized in that: Further including: According to the calculated distance, the distance between each sample in the first training set and the sample to be classified is calculated, and k samples are selected as the nearest neighbors of the sample to be classified according to the distance sorting. The value of k is selected according to the size of the training set, and the value of k is 3 to 8; For the selected k nearest neighbors, the delay categories to which they belong are counted, and the category with the largest number of occurrences is used as the classification category of the sample to be classified.

8. The privacy number call efficiency detection method based on the privacy communication platform according to claim 7 is characterized in that: Further including: The following formula is used to calculate the distance between each training sample in the first training set and the sample to be classified: Where d(x,c) represents the Euclidean distance between any training sample x in the first training set and the sample to be classified c, x i and c i are the values ​​of the training sample x and the sample to be classified c on the i-th feature, respectively. i is a positive integer, specifically 1, 2, ...n, where n is the number of input features of the delay type classification model.

9. A privacy number call efficiency detection system based on a privacy communication platform, characterized in that: It implements the privacy number call efficiency detection method based on the privacy communication platform according to any one of claims 1 to 8, and the privacy number call efficiency detection system includes: A data collection module is used to collect call data from the privacy communication platform during historical calls, specifically including the total delay time from the caller number initiating the call to the start of the ringing, the upstream delay time, the search delay time, the downstream delay time, the network type, the province information, and the platform information during the call; the privacy communication platform includes multiple sub-platforms that can communicate with each other; A model building module is used to build a privacy number platform delay factor detection model, the privacy number platform delay factor detection model includes a delay type classification model and an influencing factor judgment model connected in series with the delay type classification model; each collected call list data is classified into a delay type, and multiple classifications are performed according to the influencing factors of the delay time of each call and the delay category labels are marked to establish a first training data set; the delay type classification model is trained using the first training data set, so that the delay type classification model learns the degree of influence of different influencing factors on the call efficiency of each call list data to determine the delay type to which it belongs; the CART algorithm is used to build an influencing factor judgment model; the construction of the influencing factor judgment model includes: dividing the factor nodes according to the number of optional values ​​of the feature, selecting the optimal feature and the optimal splitting point, until all nodes are split, and completing the construction of the influencing factor judgment model; The determination module is used to collect the total delay time, upstream delay, search delay, downstream delay, network type, sub-platform code, and province code in the current call bill data in real time, and input the collected total delay time, upstream delay, search delay, downstream delay, and network type in the current call bill data into the privacy number platform delay factor detection model to obtain the delay category of the current call bill data and the biggest influencing factor affecting the sub-platform call efficiency in the current call bill data.

10. The privacy number call efficiency detection system based on the privacy communication platform according to claim 9 is characterized in that: The factor node division according to the number of optional values ​​of the feature includes: Factor nodes are divided according to the number of optional values ​​of the following features: platform code, sub-platform code, and delay category of each call record data; the optional value specifically refers to the value of each input feature; The weighted sum of the Gini impurity of the influencing factor judgment model is further repeatedly calculated to perform multiple divisions of the data set, and the features and segmentation points that minimize the calculated weighted sum of the Gini impurity are selected as the optimal division of the current node.

Images