Method and device for detecting telephone number accuracy, electronic equipment and storage medium

Abstract

The present application relates to a kind of detecting telephone number accuracy method, device, electronic equipment and storage medium.The detecting telephone number accuracy method includes the following steps:S1, sample extraction, read the telephone number information corresponding to telephone number as to be verified sample;S2, sample verification, using automatic telephone dialing test auxiliary to complete the sample verification extracted in manual way;S3, sample queue, the sample that has been successfully verified is randomly sampled using first-in first-out queue structure and is stored in telephone number information base;S4, accuracy calculation, accuracy calculation includes single-dimensional accuracy calculation and multi-dimensional accuracy calculation, the accuracy interval value in the confidence 95% of 2 times standard deviation is calculated, and dynamic output calculation result, complete the process of dynamic detection telephone number information accuracy;S5, sampling iteration.According to the detecting telephone number accuracy method of the present application, telephone number accuracy can be continuously and dynamically detected to reflect the latest state of its accuracy quality in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication, in particular, to a method and device for realizing high-confidence detection of telephone number accuracy based on dynamic small samples, electronic equipment and storage medium. BACKGROUND

[0002] When a telephone user registers a telephone number with an operator, the corresponding user information is also registered, and these pieces of information combined together constitute an important data asset specific to the operator, i.e., telephone number information.

[0003] Telephone number information can be divided into two categories according to different registered users: enterprise users register enterprise and institution information, including but not limited to unit name, address, telephone, industry classification, etc., also known as "Yellow pages"; personal users register personal information, including but not limited to name, address, etc., also known as "White pages".

[0004] The widespread application of telephone number information in the Internet era constitutes an important part of the information infrastructure of the social credit system. For example, when a commercial bank issues a credit card online or a website has a new user registration, the bank or website will query the White pages information of the operator to verify the user's real identity. Similar identity verification services using telephone number information are widely used in risk assessment models for businesses in many industries. When a user receives a call and gets a call identification prompt, the Yellow pages information of the operator is used. These are the most common application scenarios based on telephone number information.

[0005] As a basic information, telephone number information has been widely combined and applied to various computing models in various industries, including risk assessment. Therefore, the quality evaluation of its information accuracy is increasingly important, which requires effective detection of the accuracy of telephone number information.

[0006] Before the method of the present application is proposed, we first analyze the traditional statistical telephone number accuracy method and point out its defects to clarify the difference between the method of the present application and them.

[0007] The traditional telephone number accuracy calculation method has three major defects, which are: high cost of full data statistics; low efficiency of manual information verification; and inability to realize continuous dynamic detection.

[0008] First, the total statistics problem. The telephone numbers of enterprises and institutions, plus the fact that there are a large number of telephone numbers registered in the name of individuals but actually used for external contact by enterprises, the absolute number of telephone number information is very large, the number is in the hundreds of millions. The traditional method of detecting the accuracy of telephone number information is to check it with the yellow page information (i.e. the information of enterprises and institutions in different industries). For example, after checking the telephone numbers of enterprises and institutions by industry or by region (district, county, city), the accuracy of the telephone number information is calculated. This method is only suitable for the statistics of the accuracy of telephone number information in a certain industry (such as medical, education, machinery manufacturing, banking, etc.) or a certain region. In fact, it is a partial and total statistics, which is costly and time-consuming, and the accuracy obtained cannot represent the overall accuracy of the telephone number, and the confidence cannot be scientifically measured. Moreover, with the replacement of telephone directories (yellow pages) by search engines, even this partial and total data checking relying on traditional directory (yellow page) editing and publishing is difficult to carry out due to high cost.

[0009] Second, the problem of manual verification of information. Where does the telephone number information of the operator come from? How is it changed? When the user registers the telephone number at the operator's business hall, the registered information will go through the operator's manual license verification process, so the telephone number information at the time of registration can be considered accurate and authoritative, which is the original source of telephone number information. With the passage of time, the name of the enterprise will change, the owner and user of the telephone may separate, private telephones may be used for enterprises, enterprises may merge and reorganize or even go out of business, and millions of enterprise subjects are cancelled every year, which will lead to inaccurate telephone number information of the operator, which is the change of information. In order to verify whether the telephone number information is correct, manual checking through call center and user contact is needed, which is low in efficiency.

[0010] Third, in the era of big data, when we evaluate the information quality of hundreds of millions of telephone number information as a whole, unlike the traditional way of statistical accuracy once a year, we need to continuously and dynamically monitor the accuracy of telephone number information efficiently and at low cost, and we need a method that can continuously and dynamically detect the overall accuracy.

[0011] In summary, how to scientifically and effectively detect the accuracy of telephone number information at low cost, how to make the detection result of the accuracy meet the predetermined confidence standard through a sampling method that meets the characteristics of telephone number information, and how to save labor cost by automating information checking to minimize the cost of obtaining statistical results, have become urgent problems to be solved. Therefore, it is urgent to design a method that can continuously and dynamically detect the accuracy to reflect the latest state of the quality in real time according to the characteristics of the continuous and dynamic update of telephone number information, and then use the accuracy index to promote the high-quality development of this social credit system infrastructure. SUMMARY

[0012] The technical problem to be solved by the present application is how to scientifically and effectively and at low cost detect the accuracy of telephone number information and how to make the detection result of the accuracy conform to the predetermined confidence standard through a sampling method conforming to the characteristics of telephone number information.

[0013] To solve the above technical problems, according to one aspect of the present application, a method for detecting the accuracy of telephone number information is provided, the telephone number information including fixed telephone number, mobile telephone number, and multi-field combination with telephone number as unique identifier, the multi-field combination including name, address, switchboard sign, industry classification and domain name, the method for detecting the accuracy of telephone number information comprising the following steps: S1, sample extraction, reading the telephone number information corresponding to the telephone number as a sample to be verified, the telephone number information having a storage time when recorded into a database, using a method of equidistant random sampling according to the storage time to sample, to ensure the randomness of sampling, wherein the data of the telephone number information is first sorted according to the storage time, then the total data amount is divided by the sampling number to obtain a sampling interval, and a number is randomly sampled within the sampling interval as a sample, the sampling is equidistant, until the sampling is completed, thereby ensuring the average distribution of the sample in the time dimension; S2, sample verification, using automatic telephone dialing test to assist the verification of the extracted sample in a manual manner, to verify whether the telephone number information is correct, wherein the automatic telephone dialing test uses an artificial intelligent robot to call the telephone user to verify the information, and outputs the verification result, to confirm whether the verification data is consistent with the original telephone number information; S3, sample queue, randomly sampling the successfully verified sample, and storing it into the telephone number information database using a first-in first-out queue structure, setting an effective time T and a sampling length L, wherein the effective time T is a time range of T time length from the start calculation time t, the sampling length L is the number of samples in a single sampling, and the effective time T and the sampling length L are prepared for the next step; S4, accuracy calculation, the accuracy calculation includes single-dimensional accuracy calculation and multi-dimensional accuracy calculation, setting a dynamic calculation period length, performing the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation according to the dynamic calculation period length, calculating the accuracy interval value within the confidence range of 95% of 2 times standard deviation based on the sample queue according to the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation, and dynamically outputting the calculation result, to complete the process of dynamically detecting the accuracy of the telephone number information, wherein the single dimension represents the corresponding relationship between the telephone number and a field in the telephone number information, and the multi-dimension represents the corresponding relationship between the telephone number and multiple fields in the telephone number information, wherein the single-dimensional accuracy calculation is used to judge the accuracy Ri of the extracted sample, if the name in the sample is completely consistent with the verification data, then Ri=1, otherwise Ri=0; the multi-dimensional accuracy calculation is used to perform fuzzy comparison on each field data of the telephone number information, and output the consistency degree weight of each dimension corresponding to each field; S5, sampling iteration, sampling is completed according to the sample queue in step S3 with an initial sample capacity n, and the sampling accuracy is calculated according to the accuracy calculation method in step S4; calculating the sample mean and sample variance, wherein when the standard deviation is greater than an experience threshold, it is judged that the sampling is unqualified, at this time, the sample capacity needs to be expanded, and steps S1 to S3 are repeated until the standard deviation is less than the experience threshold 0.01; When the standard deviation is less than the empirical threshold value, the sample mean of the telephone number information accuracy is obtained.

[0014] According to an embodiment of the application, in step S4, in the single-dimensional accuracy calculation, the formula for calculating the sample mean may be:

[0015] (1)

[0016] wherein the accuracy of a single sample is Ri, the proportion of sample verification consistency is p, the sampling quantity is n, the sample mean is , and the standard deviation is ,

[0017] The formula for calculating the standard deviation is:

[0018] (2)

[0019] The formula for calculating the single-dimensional accuracy interval μ is:

[0020] (3)

[0021] According to an embodiment of the application, in step S4, in the multi-dimensional accuracy calculation, fuzzy calculation is performed, which can include the following steps: S41, performing word segmentation and removing stop words on single-field information in multi-dimensional information to obtain training data; S42, inputting the training data, and obtaining a corpus model of the single-field information through word2vec training in a Gensim package of Python; S43, performing word segmentation on single-field data in sample records and verification result records respectively, and obtaining word vectors through corpus model adaptation; S44, after the vectors of the single-field data in the sample records and the verification records after word segmentation are simply added, the cosine similarity of the two is the accuracy of a single dimension. The value range thereof is [-1, 1], and the greater the value is, the more similar it is, and 1 represents complete consistency.

[0022] According to an embodiment of the application, in step S4, in the multi-dimensional accuracy calculation, the weight of each dimension is Wi, the accuracy of each dimension is Di, and the number of dimensions is d,

[0023] wherein the formula for calculating the accuracy of a single multi-dimensional sample is:

[0024] (4)

[0025] The accuracy of a single sample is , the sampling capacity is n, the sample mean is , and the standard deviation is ,

[0026] The formula for calculating the sample mean The formula is:

[0027] (5)

[0028] Calculate the standard deviation The formula is:

[0029] (6)

[0030] Calculate the multidimensional accuracy interval The formula is:

[0031] (7)

[0032] According to an embodiment of the present invention, in step S5, the initial sample size n can be set to 100.

[0033] According to an embodiment of the present invention, in step S1, when reading the telephone number information corresponding to the telephone number as a sample to be verified, dynamic sampling is performed on the data at a default time interval of 30 minutes. A default set number of 1000 telephone numbers is randomly selected from the telephone number information database, and the telephone number information corresponding to the telephone number is read as a sample to be verified.

[0034] According to a second aspect of the present application, there is provided a device for detecting the accuracy of a telephone number, comprising: a sample extraction module for reading telephone number information corresponding to a telephone number as a sample to be verified, the telephone number information having a storage time when recorded into a database, and the sample extraction module using a method of equidistant random sampling according to the storage time to ensure randomness of the sampling, wherein the data of the telephone number information is first sorted according to the storage time, then the total data amount is divided by the sampling number to obtain a sampling interval, and a number is randomly selected within the sampling interval as a sample, and the equidistant sampling is performed according to the sampling interval until the sampling is completed, thereby ensuring the average distribution of the sample in the time dimension; a sample verification module for verifying the extracted sample in an artificial manner with the aid of automatic telephone dialing test, and verifying whether the telephone number information is correct, wherein the automatic telephone dialing test uses an artificial intelligent robot to call the telephone user to verify the information, and outputs a verification result to confirm whether the verified data is consistent with the original telephone number information; a sample queue module for randomly sampling the successfully verified sample and storing the sample into a telephone number information database using a first-in-first-out queue structure, and setting an effective time T and a sampling length L, wherein the effective time T is a time range of T time length before the start of the calculation time t, the sampling length L is the number of samples in a single sampling, and the effective time T and the sampling length L are used for the next step; and an accuracy calculation module, the accuracy calculation including single-dimensional accuracy calculation and multi-dimensional accuracy calculation, the accuracy calculation module being configured to set a dynamic calculation period length, perform the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation according to the dynamic calculation period length, calculate an accuracy interval value within a 95% confidence range of 2 times of a standard deviation based on the sample queue according to the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation, and dynamically output the calculation result to complete the process of dynamically detecting the accuracy of the telephone number information.

[0035] According to an embodiment of the present application, the device further comprises a sampling iteration module for performing sampling iteration calculation based on the sample queue module and the accuracy calculation module to output the accuracy of the telephone number information satisfying an empirical threshold.

[0036] According to a third aspect of the present application, there is provided an electronic device, comprising a memory, a processor, and a detection telephone number accuracy program stored in the memory and executable on the processor, and the detection telephone number accuracy program implements the steps of the detection telephone number accuracy method when executed by the processor.

[0037] According to a fourth aspect of the present application, there is provided a computer storage medium, wherein the computer storage medium stores a detection telephone number accuracy program, and the detection telephone number accuracy program implements the steps of the detection telephone number accuracy method when executed by a processor.

[0038] Compared with the prior art, the technical scheme provided by the embodiment of the present application can at least achieve the following beneficial effects:

[0039] As the basic data of the digital platform of various industries in today's big data era, telephone number information plays an important role in various models including risk assessment, and the evaluation of its accuracy has important practical significance. The present application introduces the basic principles of probability and mathematical statistics into the detection of the accuracy of telephone number information, improves the traditional high-cost, low-efficiency and unreliable local full-detection method, and can scientifically detect the accuracy of the telephone number information of the operator in the global range. Under the premise of ensuring that the accuracy confidence is higher than 95%, the required sample set size can be automatically and dynamically adjusted through iterative calculation to reduce manual intervention. The modeling method and accuracy calculation formula for two types of telephone number information (i.e. single dimension and multi-dimension) are given. And a specially designed sample queue is used to automatically extract telephone number information samples to achieve the purpose of low-cost and dynamic continuous calculation of accuracy.

[0040] The present application uses the central limit theorem to calculate the accuracy of telephone number information through sampling statistics, and uses the "equal interval of storage time" random sampling method to ensure the randomness of sample extraction. For the accuracy of hundreds of millions of telephone number information, the application of the present application has high reliability, and the confidence is set to be higher than 95%. The accuracy is continuously and dynamically calculated, and automatic sampling is realized to achieve low-cost and dynamic automatic detection.

[0041] The present application uses a specially designed sample queue to automatically extract telephone number information samples to achieve the purpose of continuously and dynamically obtaining the interval estimation of the accuracy of the telephone number information within the target confidence. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments will be briefly introduced below. Obviously, the drawings described below only relate to some of the embodiments of the present application, but not limit the present application.

[0043] Figure 1 is a flow chart showing a method for detecting the accuracy of telephone number according to the embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme of the embodiments of the present application will be clearly and completely described below in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, but not all the embodiments. Based on the described embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application.

[0045] Unless otherwise defined, technical terms or scientific terms used herein shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terms "first", "second", and similar terms as used herein do not denote any order, quantity, or importance, but are used to identify different components. Also, the terms "a" or "an", as used herein, do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items.

[0046] The method for detecting the accuracy of telephone numbers is used to detect the accuracy of telephone number information including fixed telephone numbers, mobile telephone numbers, and multi-field combinations with telephone numbers as unique identifiers, including name, address, switchboard sign, industry classification, and domain name.

[0047] Figure 1 is a flow chart showing a method for detecting the accuracy of telephone numbers according to an embodiment of the application.

[0048] As shown in Figure 1 The method for detecting the accuracy of telephone numbers includes the following steps:

[0049] S1, sample extraction, read the telephone number information corresponding to the telephone number as a sample to be verified, the telephone number information has an entry time when recorded into the database, and a random sampling method according to the entry time is used to ensure the randomness of the sampling, wherein the data of the telephone number information is first sorted according to the entry time, then the total data volume is divided by the sampling number to obtain a sampling interval, and a number is randomly selected within the sampling interval as a sample, and the sampling is performed at equal intervals according to the sampling interval until the sampling is completed, thereby ensuring the average distribution of the sample in the time dimension.

[0050] Telephone number: can be a fixed telephone number containing an area code; can also be a mobile telephone number. For example: (021-12345678, 18912345678).

[0051] Telephone number information: a combination of corresponding multi-fields with telephone numbers as unique identifiers (primary keys), i.e. (telephone number: name, address, switchboard sign, industry classification, domain name). For example: (021-12345678: Shanghai Third Machinery Company, No. 100, Nanjing West Road, 'Y', 'Mechanical Manufacturing', "www.shdsjx.cn")

[0052] S2, sample verification, using automatic telephone dialing test to assist in verifying the extracted sample in a manual manner, whether the telephone number verification information is correct, wherein the automatic telephone dialing test uses an artificial intelligent robot to telephone the user to check the information, and outputs the verification result to confirm whether the verified data is consistent with the original telephone number information.

[0053] S3, sample queue, randomly sampling the successfully verified samples, storing them into the phone number information database using a first-in first-out queue structure, setting an effective time T and a sampling length L, wherein the effective time T is a T time range before the start of the calculation time t, the sampling length L is the number of samples for a single sampling, and the effective time T and the sampling length L are prepared for the next step.

[0054] S4, accuracy calculation, the accuracy calculation includes single-dimensional accuracy calculation and multi-dimensional accuracy calculation, setting a dynamic calculation period length, performing single-dimensional accuracy calculation and multi-dimensional accuracy calculation according to the dynamic calculation period length, calculating the accuracy interval value within the 95% confidence range of 2 times the standard deviation based on the sample queue according to the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation, and dynamically outputting the calculation result to complete the process of dynamically detecting the accuracy of the phone number information.

[0055] wherein single dimension represents the correspondence between the phone number and a field in the phone number information, and multi-dimension represents the correspondence between the phone number and multiple fields in the phone number information; the single-dimensional accuracy is the accuracy of the phone number corresponding to a field in the phone number information, and the multi-dimensional accuracy is the weighted average accuracy of the phone number corresponding to all fields in the phone number information.

[0056] wherein the single-dimensional accuracy calculation is used to judge the accuracy Ri of a certain sample, if the name in the sample is completely consistent with the verification data, then Ri = 1, otherwise Ri = 0; the multi-dimensional accuracy calculation is used to compare the fuzzy comparison of each field data of the phone number information, and output the consistency degree weight of each dimension corresponding to each field.

[0057] S5, sampling iteration, according to the sample queue of step S3, completing the sampling with an initial sample capacity n, calculating the sampling accuracy according to the accuracy calculation method of step S4; calculating the sample mean and sample variance of the sampling,

[0058] wherein when the standard deviation is greater than the experience threshold, it is judged that the sampling is unqualified, at this time the sample capacity needs to be expanded, and steps S1 to S3 are repeated until the standard deviation is less than the experience threshold 0.01; when the standard deviation is less than the experience threshold, the sample mean is the accuracy of the phone number information.

[0059] The application utilizes the central limit theorem to calculate the accuracy of the telephone number information by sampling statistics, and guarantees the randomness of sample extraction according to the "equidistant storage time" random sampling method. For the accuracy of hundreds of millions of telephone number information, the application of the method has high reliability, and the confidence level is set to be more than 95%. The accuracy is continuously and dynamically calculated, and automatic sampling is realized to achieve low-cost and dynamic automatic detection. The application combines the use of a specially designed sample queue to automatically extract telephone number information samples, and achieves the purpose of continuously and dynamically obtaining the interval estimation of the accuracy of the telephone number information within the target confidence level.

[0060] According to one or some embodiments of the application, in step S4, in the single-dimensional accuracy calculation, the sample mean is calculated The formula is:

[0061] (1)

[0062] Wherein, the accuracy of a single sample is Ri, the proportion of sample verification consistency is p, the sampling number is n, the sample mean is , and the standard deviation is .

[0063] The formula for calculating the standard deviation is:

[0064] (2)

[0065] The formula for calculating the single-dimensional accuracy interval mu is:

[0066] (3).

[0067] According to one or some embodiments of the application, in step S4, the fuzzy calculation is performed in the multi-dimensional accuracy calculation, including the following steps:

[0068] S41, the single field information in the multi-dimensional information is segmented and the stop words are removed to obtain training data.

[0069] S42, the training data is input, and the word2vec training in the Gensim package of Python is performed to obtain the corpus model of the single field information.

[0070] S43, the single field data in the sample record and the verification result record are segmented respectively, and the word vectors are obtained through the corpus model adaptation.

[0071] S44, the cosine similarity of the vectors of the single field data in the sample record and the verification record after simple addition is the accuracy of the single dimension. The value range is [-1, 1], and the greater the value is, the more similar it is. 1 represents complete consistency.

[0072] According to one or some embodiments of the present application, in step S4, in the multi-dimensional accuracy calculation, let the weight of each dimension be Wi, the accuracy of each dimension be Di, and the number of dimensions be d,

[0073] The formula for calculating the accuracy of a single multi-dimensional sample is:

[0074] (4)

[0075] The accuracy of a single sample is , the sample size is n, the sample mean is , and the standard deviation is ,

[0076] The formula for calculating the sample mean is:

[0077] (5)

[0078] The formula for calculating the standard deviation is:

[0079] (6)

[0080] The formula for calculating the multi-dimensional accuracy interval is:

[0081] (7).

[0082] According to one or some embodiments of the present application, in step S5, the initial sample size n is set to 100.

[0083] According to one or some embodiments of the present application, in step S1, when reading the phone number information corresponding to the phone number as the sample to be verified, the default setting is to dynamically sample data every 30 minutes, and 1000 phone numbers are randomly selected from the phone number information database as the default number of samples to be verified.

[0084] The telephone number information as the basic data of the digital platform of various industries in the current big data era plays an important role in various models including risk assessment, and the assessment of its accuracy has important practical significance. The present application introduces the basic principles of probability and mathematical statistics into the detection of the accuracy of the telephone number information, improves the traditional high-cost low-efficiency unreliable local full detection method, and can scientifically detect the accuracy of the telephone number information of the operator in the global range. Under the premise of ensuring that the accuracy confidence is higher than 95%, the required sample set size can be automatically dynamically adjusted through iterative calculation to reduce manual intervention. The modeling method and accuracy calculation formula for two types of telephone number information (i.e. single dimension and multi dimension) are given. And a specially designed sample queue is used to automatically extract telephone number information samples to realize the purpose of low-cost dynamic continuous calculation of accuracy.

[0085] According to the second aspect of the present application, a device for detecting the accuracy of the telephone number is provided, which comprises a sample extraction module, a sample verification module, a sample queue module and an accuracy calculation module.

[0086] The sample extraction module is used to read the telephone number information corresponding to the telephone number as the sample to be verified, the telephone number information has the storage time when being recorded into the database, and the equidistant random sampling method according to the storage time is used for sampling to ensure the randomness of sampling, wherein the data of the telephone number information is first sorted according to the storage time, then the sampling interval is obtained by dividing the total data amount by the sampling number, and a number is randomly extracted in the sampling interval as a sample, and the equidistant sampling is performed according to the sampling interval until the sampling is completed, so as to ensure the average distribution of the sample in the time dimension.

[0087] The sample verification module is used to complete the verification of the extracted sample in an artificial manner with the aid of automatic telephone dialing test, and whether the telephone number verification information is correct, wherein the automatic telephone dialing test is in the form of an artificial intelligent robot to telephone the user to verify the information, and outputs the verification result to confirm whether the verified data is consistent with the original telephone number information.

[0088] The sample queue module is used to randomly sample the successfully verified sample, and store it into the telephone number information database using the first-in first-out queue structure, set the effective time T and the sampling length L, wherein the effective time T is the T time range before the start calculation time t, the sampling length L is the number of samples in a single sampling, and the effective time T and the sampling length L are prepared for the next step.

[0089] The accuracy calculation module includes single-dimension accuracy calculation and multi-dimension accuracy calculation, and is configured to set a dynamic calculation period length, perform single-dimension accuracy calculation and multi-dimension accuracy calculation according to the dynamic calculation period length, calculate an accuracy interval value within a 95% confidence range of 2 times of a standard deviation based on the sample queue according to the single-dimension accuracy calculation and the multi-dimension accuracy calculation, and dynamically output the calculation result to complete the process of dynamically detecting the telephone number information accuracy.

[0090] According to one or some embodiments of the present application, the device further comprises a sampling iteration module configured to perform sampling iteration calculation based on the sample queue module and the accuracy calculation module to output telephone number information accuracy satisfying an experience threshold.

[0091] The method and device provided by the present application can be used to dynamically detect the accuracy of a telephone number information database of a certain operator and calculate single-dimension accuracy and multi-dimension accuracy respectively.

[0092] (1) Single-dimension accuracy

[0093] Suppose that the sampling number n=1000. A sample set R {0, 1, 1, …, 0, 1, 1} is obtained. According to the sample mean formula

[0094] , the sample mean (i.e. single-dimension accuracy) is calculated as =91.47%.

[0095] According to the standard deviation formula: (wherein the sample verification consistent proportion p=0.9147 and the sampling number n=1000), the standard deviation σ=0.008833 is obtained. The sample standard deviation is less than the experience value 0.01. The single-dimension accuracy result is valid.

[0096] According to the accuracy interval formula

[0097] ,

[0098] The single-dimension accuracy interval is obtained as 91.4153%≤u≤91.5247%.

[0099] (2) Multi-dimension accuracy

[0100] According to the single-sample accuracy formula: , according to the following table,

[0101]

[0102] The single-sample accuracy is calculated as =0.91876. Taking sample size n=100, 10 times of sampling accuracy set is obtained: (0.91876, 0.88738, 0.92768, 0.87454, 0.86325, 0.88913 … 0.95622, 0.93329, 0.85881, 0.87699). According to sample mean formula: , sample mean =89.8605%. According to standard deviation formula: , standard deviation =0.031416, greater than experience threshold 0.01. Multidimensional accuracy result is invalid.

[0103] Taking sample size n to 1000, 1000 times of sampling accuracy set (0.91876, 0.92326, 0.91989, …, 0.89648) is obtained. According to sample mean formula: , sample mean =90.2711%, according to standard deviation formula: , standard deviation =0.005913, less than experience threshold 0.01. Multidimensional accuracy result is valid.

[0104] According to accuracy interval formula , multidimensional accuracy interval is obtained: 90.2345%≤ ≤90.3077%.

[0105] According to still another aspect of the present application, a device for detecting telephone number accuracy is provided, comprising: a memory, a processor, and a program for detecting telephone number accuracy stored in the memory and executable on the processor, the program for detecting telephone number accuracy, when executed by the processor, implements the steps of the above-mentioned method for detecting telephone number accuracy.

[0106] According to the present application, a computer storage medium is further provided.

[0107] The program for detecting telephone number accuracy is stored on the computer storage medium, and the program for detecting telephone number accuracy, when executed by the processor, implements the steps of the above-mentioned method for detecting telephone number accuracy.

[0108] The method implemented when the program for detecting telephone number accuracy executed on the processor can refer to each embodiment of the method for detecting telephone number accuracy of the present application, and will not be repeated here.

[0109] The present application further provides a computer program product.

[0110] The computer program product of the present application comprises a telephone number accuracy detection program, which, when executed by a processor, implements the steps of the telephone number accuracy detection method as described above.

[0111] The method implemented by the telephone number accuracy detection program running on the processor can refer to each embodiment of the telephone number accuracy detection method of the present application, and will not be described here.

[0112] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software and a necessary general hardware platform, and of course, can also be implemented by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, or an optical disc) as described above, and includes a plurality of instructions for causing an end device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0113] The above description is only exemplary embodiments of the present application, and is not intended to limit the protection scope of the present application, which is defined by the appended claims.

Claims

1. A method for detecting the accuracy of telephone number information, the telephone number information including fixed telephone numbers, mobile telephone numbers, and multi-field combinations using telephone numbers as unique identifiers, the multi-field combinations including name, address, switchboard sign, industry classification, and domain name, the method comprising the following steps: S1, sampling, the data of the telephone number information is first sorted by the time of storage, then the sampling interval is obtained by dividing the total data amount by the sampling number, and a number is randomly selected as a sample within the sampling interval, the sampling is performed at equal intervals within the sampling interval until the sampling is completed, thereby ensuring the average distribution of the sample in the time dimension; S2, sample verification, the verification of the extracted sample is completed in an artificial manner with the aid of automatic telephone dialing test, and the telephone number verification information is checked, wherein the automatic telephone dialing test checks the telephone user in the form of an artificial intelligence robot, outputs the verification result, and confirms whether the checked data is consistent with the original telephone number information; S3, sample queue, the successfully checked sample is randomly sampled and stored in the telephone number information database using a first-in first-out queue structure, and the effective time T and the sampling length L are set, wherein the effective time T is a time range of T time length from the start of the calculation time t, the sampling length L is the number of samples for a single sampling, and the effective time T and the sampling length L are prepared for the next step; S4, accuracy calculation, the accuracy calculation includes single-dimensional accuracy calculation and multi-dimensional accuracy calculation, a dynamic calculation period is set, the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation are performed according to the dynamic calculation period, the accuracy interval value within the 95% confidence range of 2 times the standard deviation is calculated based on the sample queue according to the single-dimensional accuracy calculation and the multi-dimensional accuracy calculation, and the calculation result is dynamically output, thereby completing the process of dynamically detecting the accuracy of the telephone number information, wherein single dimension represents the corresponding relationship between the telephone number and a field in the telephone number information, and multi dimension represents the corresponding relationship between the telephone number and multiple fields in the telephone number information; the single-dimensional accuracy is the accuracy of a field in the telephone number information corresponding to the telephone number, and the multi-dimensional accuracy is the weighted average accuracy of all fields in the telephone number information corresponding to the telephone number; wherein the single-dimensional accuracy calculation is used to judge the accuracy Ri of a sample, if the name in the sample is completely consistent with the verification data, then Ri = 1, otherwise Ri = 0; the multi-dimensional accuracy calculation is used to perform fuzzy comparison on each field data of the telephone number information, and output the consistency degree weight of each dimension corresponding to each field; S5, sampling iteration, the sample queue in step S3 is used to complete the sampling with an initial sample capacity n, and the sampling accuracy is calculated according to the accuracy calculation method in step S4; the sample mean and the sample variance are calculated, wherein when the standard deviation is greater than the experience threshold, it is judged that the sampling is unqualified, at this time, the sample capacity needs to be expanded, and steps S1 to S3 are repeated until the standard deviation is less than the experience threshold 0.01; when the standard deviation is less than the experience threshold, the sample mean is the accuracy of the telephone number information. S1, sample extraction, read the telephone number information corresponding to the telephone number as a sample to be verified, the telephone number information has the storage time when it is recorded into the database, and the sampling method of equidistant random sampling according to the storage time is adopted to ensure the randomness of sampling, wherein, ​ ​ ​ ​ ​ ​ ​ ​ 2. The method of claim 1, wherein in step S4, the single-dimension accuracy calculation comprises: Compute sample mean The formula is: （1） wherein The accuracy of a single sample is Ri, the proportion of sample verification consistency is p, the sampling number is n, the sample mean is , and the standard deviation is , The formula for calculating the standard deviation is: The formula for calculating the standard deviation is: （2） The formula for calculating the single-dimension accuracy interval μ is: （3）。 3. The method of claim 1, wherein in step S4, the multi-dimension accuracy calculation comprises fuzzy calculation, comprising the following steps: S41. Tokenizing and removing stop words from single-field information in multi-dimension information to obtain training data; S42. Inputting the training data to obtain a corpus model of single-field information through word2vec training in the Gensim package of Python; S43. Tokenizing single-field data in sample records and verification result records respectively, and obtaining word vectors through corpus model adaptation; S44. Adding the vectors of the tokenized single-field data in sample records and verification records, and the cosine similarity of the two is the accuracy of a single dimension, which ranges from -1 to 1, and the greater the value, the more similar, and 1 represents complete consistency.

4. The method of claim 1, wherein in step S4, the multi-dimension accuracy calculation comprises setting the weight of each dimension as Wi, the accuracy of each dimension as Di, and the number of dimensions as d, wherein, The formula for calculating the accuracy of a single multi-dimension sample is: （4） The accuracy of a single sample is , the sample size is n, the sample mean is , the standard deviation is , and the formula for calculating the sample mean is （5） The formula for calculating the standard deviation is: σ = √[(Σ (xi - μ)2) / ( （6） Computing multidimensional accuracy intervals The formula is: （7）。 5. The method of claim 1, wherein in step S5, the initial sample capacity n is set to 100.

6. The method of claim 1, wherein in step S1, when reading the phone number information corresponding to the phone number as the sample to be verified, the default setting is to dynamically sample every 30 minutes of time interval data, and 1000 phone numbers are randomly sampled from the phone number information database as the default number of samples, and the phone number information corresponding to the phone number is read as the sample to be verified.

7. A device for detecting the accuracy of a phone number, comprising: a sample extraction module for reading the phone number information corresponding to the phone number as the sample to be verified, the phone number information having an entry time into the database, and using an equidistant random sampling method according to the entry time to ensure randomness of the sampling, wherein the data of the phone number information is first sorted according to the entry time, then the total data volume is divided by the sampling number to obtain a sampling interval, and a number is randomly sampled within the sampling interval as a sample, and the equidistant sampling is performed according to the sampling interval until the sampling is completed, thereby ensuring the average distribution of the sample in the time dimension; a sample verification module for verifying the extracted sample in an automatic telephone dialing test assisted by manual verification to determine whether the phone number verification information is correct, wherein the automatic telephone dialing test uses an artificial intelligence robot to call the user of the phone number to verify the information, and outputs the verification result to confirm whether the verified data is consistent with the original phone number information; a sample queue module for randomly sampling the successfully verified sample and storing it in the phone number information database using a first-in-first-out queue structure, setting an effective time T and a sampling length L, wherein the effective time T is a T time range from the start of the calculation time t, the sampling length L is the number of samples in a single sampling, and the effective time T and the sampling length L are prepared for the next step. An accuracy calculation module, the accuracy calculation includes single-dimensional accuracy calculation and multi-dimensional accuracy calculation, the accuracy calculation module is used for setting dynamic calculation period length, carries out single-dimensional accuracy calculation and multi-dimensional accuracy calculation according to dynamic calculation period length, calculates the accuracy interval value in the confidence 95% range of 2 times standard deviation based on sample queue according to single-dimensional accuracy calculation and multi-dimensional accuracy calculation, and dynamically outputs the calculation result, completes the process of dynamically detecting telephone number information accuracy.

8. The apparatus of claim 7, further comprising: A sampling iteration module is used for sampling iteration calculation based on the sample queue module and the accuracy calculation module to output telephone number information accuracy meeting the experience threshold.

9. An electronic device comprising: A memory, a processor and a detection telephone number accuracy program stored on the memory and executable on the processor, when the detection telephone number accuracy program is executed by the processor, the steps of the detection telephone number accuracy method in any one of claims 1 to 6 are realized.

10. A computer storage medium, wherein, The computer storage medium has a detection telephone number accuracy program stored, when the detection telephone number accuracy program is executed by the processor, the steps of the detection telephone number accuracy method in any one of claims 1 to 6 are realized.

Images