Credibility calculation method and related device

[0024] The embodiment of the present invention provides a credibility calculation method and related equipment, which are used to quickly calculate a credibility result in a scenario with a super-large factor.

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work shall fall within the protection scope of this application.

[0026] See figure 1 , Is a schematic diagram of an embodiment of a possible credibility calculation method provided by an embodiment of the present invention, including:

[0027] 101. Divide the calculation factors corresponding to the target data into N calculation factor sets according to preset rules;

[0028] When calculating the credibility of the customer's data account set, the data of any set of data account sets is called the target data. When the number of calculation factors of the target data reaches 100,000, millions or even higher, In order to calculate the factor value of the calculation factor faster, the calculation factor is classified and processed in batches according to the preset rules to obtain N calculation factor sets, N is a positive integer not less than 2, where the calculation factor refers to the completion of a certain item Operation, an input item of calculation algorithm.

[0029] It should be noted that in practical applications, the calculation factors can be classified according to the dimensions of the calculation factors, for example, according to the number of times the voucher is used, the number of orders used, the frequency of voucher use, etc. For ease of understanding, assume that the calculation factors for a certain set of data sets include the following: The voucher is used more than 5 times in the current month, the voucher is used more than 10 times in the current month, the voucher is used more than 20 times in the current month, the order is used more than 5 times in the current month, the order is used more than 5 times in the current month, and the single amount of the receipt exceeds 5% of the total More than 20, according to different dimensions, we can divide the calculation factors into 3 categories, including 3 calculation factor sets, {voucher used 5 times in the current month, voucher used 10 times in the current month, voucher used 20 times in the current month}, {The order is used 5 times in the month, and the order is used 5 times in the month} and {The number of single receipts exceeding 5% of the total amount is greater than 20}. And it is understandable that there may be multiple numbers in each calculation factor set, for example, 200 to 500, etc., which is not specifically limited here.

[0030] In addition, in practical applications, there are many ways to calculate factor classification, which are not limited here.

[0031] After classifying the calculation factors corresponding to the target data, a T-type calculation factor (T is a positive integer, T is not greater than N) is obtained. For each type of calculation factor of the T-type calculation factor, there may be the same type of calculation factor When the number reaches 10,000 or even 100,000, the calculation factors of the same type are processed in batches to obtain N calculation factor sets corresponding to the T type calculation factors. It should be noted that in practical applications, there are many ways to perform batch processing, such as dividing each type of calculation factor into X batches (X is a positive integer). For example, if there are 20,000 calculation factors in a category The calculation factor is divided into 200 batches, and the number of calculation factors in each batch is 100. Of course, in actual application, it is not necessarily evenly divided into batches. Or, divide each type of calculation factor into Y batches (Y is a positive integer), and the number of each batch in the Y batch of calculation factors is not greater than the preset number, for example, if there are 20,000 calculation factors in a category Calculation factor, the preset number set by the system is 5000, and when the value of Y is 5, the number of calculation factors in batch Y can be {4500, 4500, 5000, 4000, 2000} respectively. Therefore, there are many ways to process the calculated factors in batches, which are not specifically limited here.

[0032] Optionally, in practical applications, the T-type calculation factors can be filtered according to predetermined rules for batch processing. For example, if the T-type calculation factors exist, if there are target-type calculation factors whose number is greater than the first value , The calculation factors of the target category are processed in batches, and the calculation factors of other categories are not processed in batches. For example, if there are 80 calculation factors in a type of calculation factor, and the number of batches is 100, then this type of calculation factor is not processed in batches, and there is a batch in total.

[0033] Optionally, in the embodiment of the present invention, there may be multiple target data, and when there are multiple target data, the N calculation factor sets obtained by classification are identified with an identity, and the identity may be used to identify the calculation factor The customer and the data account set of the set are distinguished from other different customers or calculation factor sets of different data account sets.

[0034] 102. Calculate the factor value of each calculation factor in the N calculation factor sets by parallel classification and calculation of M calculation threads;

[0035] After the calculation factors are classified into N calculation factor sets, the factor value of each calculation factor in the N calculation factor sets is calculated, and the factor value of the calculation factor is 0 or 1. For example, if a voucher in the data account set is used more than 9 times in the current month, the calculation factor A is that the voucher is used more than 5 times in the current month, and the calculation factor B is that the voucher is used more than 10 times in the current month, then the factor of factor A is calculated for the voucher The value is 1, and the factor value for calculating factor B is 0. In actual application, the embodiment of the present invention obtains the data resources of the enterprise, and then obtains the factor value of the calculation factor according to the preset calculation factor conditions. For example, if a certain voucher is used 8 times in the current month, the calculation factor A(A For the voucher to be used more than 5 times in the current month) is 1, and the factor value of the calculation factor B (B is the voucher to be used more than 10 times in the current month) is 0, that is, 1 corresponds to the data is true, and 0 means the data is false.

[0036] In addition, in the embodiment of the present invention, the factor value of each calculation factor in the N calculation factor sets is calculated in parallel through M calculation threads, where M is a positive integer not less than N. The calculation factors of different categories start multiple calculation threads according to actual needs. After the calculation threads are completed, if there are still waiting to be calculated, the next batch is automatically calculated. For example, if five calculation factor sets are obtained after classification and batch processing, which are represented as set 1 to set 5, and two calculation threads are started, which are represented as thread 1 and thread 2, assuming that thread 1 and thread 2 are calculated in parallel first For set 1 and set 2, if thread 1 calculates set 1 and there are sets 3 to 5, the next set will be calculated automatically, similarly to thread 2, until all sets are calculated.

[0037] 103. Calculate the current credibility value of the target data;

[0038] After calculating the factor value of each calculation factor in the N calculation factor set, upload all the factor values ​​to the calculation pool, and calculate the total weight of the current batch of factors and the summary value of the factor value to calculate the current credibility of the target data In practical applications, the current credibility value of the target data can be calculated as follows:

[0039] Z=(∑(a*b)+A)/(∑b+B);

[0040] The Z is used to represent the current credibility value of the target data, the a is used to represent the factor value of any calculation factor in the current set, and the b is used to represent the corresponding to any calculation factor The a*b is used to represent the standard value of any of the calculation factors; the A is used to represent the sum of the standard values ​​of the calculated factors in the calculated target set; the B is used to represent the The sum of the factor weights of the calculated factors in the calculated target set; the ∑(a*b) is the total weight of the calculated factors in the current set; where the factor weights are used to indicate the relative importance of the calculated factors, and the calculation The more important the factor, the greater the factor weight, and the factor weight of each calculation factor can be a preset number not less than 0.

[0041] It should be noted that the calculated target set is included in the above N calculation factor sets, and the sets in the calculated target set are all sets that have been calculated before calculating the current set, for example, N calculation factors The sets are set 1 to set 5. If the order of the calculation result is set 1 → set 3 → set 2 → set 5 → set 4, then if the current set is set 3, the target set is set 1; if the current set Is set 2, the target set is {set 1, set 3}; if the current set is set 5, the target set is {set 1, set 3, set 2}; if the current set is set 4, the target set is { Set 1, set 3, set 2, set 5}, after the calculation of set 4 is completed, the final credibility value of the target data is returned. Therefore, the number of sets in the target set is an integer between [0, N-1].

[0042] Each calculation factor has a corresponding factor weight. The factor weight is a preset weight value at the beginning, and can be set to the same value for each calculation factor, for example, the factor weight is 10 or 100. With the continuous acquisition of enterprise data and the continuous operation of the credibility calculation method of the present invention, the factor weights change accordingly. For example, assume that the final credibility value of a certain credibility calculation is confirmed (for example, a financial institution determines and lends ), the weight of all factors whose calculation factor value is 1 in the data account set corresponding to the final credibility value plus a preset value. The preset value can be 1 or other numbers, which are not specifically limited here, and the rest are calculated The factor weight of the factor remains unchanged. As the credibility calculation of the obtained enterprise operating data continues, the factor weights of the calculation factors are adjusted accordingly, and the proportion of the calculation factors after the weight adjustment will become larger, so that the final feasibility value obtained is more accurate.

[0043] The calculation pool obtains the factor weight corresponding to any calculation factor in the current set, and calculates the sum of the factor weights of the calculation factors of the current set and the sum of the weights of the standard values ​​to obtain the current credibility value of the target data, and save the obtained ( The Σ(a*b)+A) and (Σb+B) values ​​are then used as the A and B values ​​for the next calculation of the reliability value of the target data.

[0044] 104. Display the current credibility value of the target data;

[0045] After calculating the current credibility value of the target data, display the credibility value to the customer. Since there are multiple calculation factor sets and are calculated in parallel, the current credibility value has been dynamically changing , When the factor values ​​of each calculation factor in the N calculation factor sets are all calculated, the current credibility value of the target data is the final credibility value of the target data, and the final credibility value of the target data is closer to 1 , Which means that the authenticity of the target data is higher.

[0046] It should be noted that, in order to prompt the user of the final credibility value, the display color of the final credibility value can be set to be different from other credibility values, or the final credibility value can be dynamically displayed repeatedly. Therefore, in practical applications, the specific display mode of the final credibility value is not limited in this application.

[0047] In addition, since there is a scenario where a customer has multiple data account sets in practical applications, the final credibility value of each data account set can be calculated, and the data account set with the highest final credibility value is recommended to the financial Institutions to perform operations such as credit review. Optionally, if the data account set with the highest final credibility value is confirmed by a financial institution, all calculation factors in the data account set with the highest final credibility value are 1. The weight of the factor plus a preset value, the preset value can be 1 or other numbers, which are not specifically limited here.

[0048] The embodiment of the present invention can quickly calculate the credibility value in the scene of the super-large factor, and based on the elastic calculation, the result of the credibility value is gradually and dynamically fed back, eliminating the user's lack of waiting.

[0049] In addition, in the embodiment of the present invention, the calculation factors corresponding to the target data are classified and processed in batches. It can be understood that: 1. In the case of the same number, the time required for the server to calculate the factor value of the same type of calculation factor It is smaller than the factor value for calculating different types of calculation factors; 2. When the number of calculation factors is as large as 100,000 or even millions, the calculation factors may be classified. There may be a large number of calculation factors in a certain type of calculation factor. Therefore, in the embodiment of the present invention, each type of calculation factor can be processed in batches to obtain N calculation factor sets, and then M threads are used to calculate the factor value of each calculation factor in the N calculation factor sets in parallel. There is one thread processing in the technology. When the number of calculation factors is the same and the processing capabilities of each thread are the same, the calculation speed of multiple threads at the same time is greater than the calculation speed of one thread. Therefore, the embodiment of the present invention is in a scenario with a large factor It can speed up the calculation of the credibility value of the target data.

[0050] In order to facilitate a better understanding of this application, the following description will be combined with specific practical application scenarios, as follows:

[0051] First, define several types of factors. To simply illustrate the situation, the number of each type of factor is only 3. In actual operation, this value is recommended to be between 200-500:

[0052] Factors of the first type: (behavior model, by times)

[0053] A1: The voucher has been used more than 5 times in the current month;

[0054] A2: The voucher has been used more than 20 times in the current month;

[0055] A3: The voucher has been used more than 50 times in the current month;

[0056] The second type of factor: (behavior model, according to frequency)

[0057] B1: The voucher is used every week in the current month;

[0058] B2: The voucher will be used for 2 weeks in the current month;

[0059] B3: The voucher is only used for 1 week in the current month;

[0060] The third type of factors: (consistency model, fraud prevention)

[0061] C1: The proportion of sales order-sales delivery-receipt collection is greater than 80%;

[0062] C2: The proportion of sales order-sales delivery-receipt related generation is greater than 50%;

[0063] C3: The ratio of sales order-sales delivery-receipt collection is greater than 30%;

[0064] The fourth type of factor: (anti-fraud model);

[0065] D1: The amount of a single receipt sheet exceeds 5% of the total amount of the current month and the number is greater than 10;

[0066] D1: The amount of a single receipt sheet exceeds 3% of the total amount of the current month and the number is greater than 20;

[0067] D1: The amount of a single receipt sheet exceeds 2% of the total amount of the current month and the number is greater than 50;

[0068] According to the description of the above method, the factors in this case are divided into 4 types, A, B, C, and D. These 4 types of factors are basically similar, and faster calculation results will be obtained in batch calculation;

[0069] Assuming that the initial weight of each factor is 10, now start to calculate A customer;

[0070] Consider that there are only 4 factor classifications, assuming that 2 calculation threads are enabled;

[0071] The result of thread 1 calculation is A1(1), A2(1), A3(0). After this result is transferred to the calculation pool in batches, 004 calculates the current factor value summary=1*10+1*10+0*10= 20. The weight is summarized as 10+10+10=30;

[0072] At this time, there is no other data in the calculation pool, and the returned result = 20/30 = 0.6667;

[0073] The result of thread 2 calculation is B1(1), B2(1), B3(1). After this result is sent to the calculation pool in batches, 004 calculates the current factor value summary=1*10+1*10+1*10= 30, the weight is summarized as 10+10+10=30;

[0074] At this time, the calculation pool has data for A, and the return result = (20+30)/(30+30) = 0.8333;

[0075] Save all current factor values ​​(50) and factor weights (60) at this time;

[0076] Thread 1 finds that there are still C and D not calculated, then continue to calculate C, the result is C1(1), C2(1), C3(1);

[0077] After this result is transferred to the calculation pool in batches, 004 calculates that the current factor value summary=1*10+1*10+1*10=30, and the weight summary is 10+10+10=30;

[0078] At this time, the calculation pool has summary data of A and B, and the returned result = (50+30)/(60+30) = 0.8888;

[0079] Now save all current factor values ​​(80) and factor weights (90);

[0080] Thread 2 also calculates D, the result is D1(0), D2(1), D3(0);

[0081] After this result is transferred to the calculation pool in batches, 004 calculates the current factor value summary=0*10+1*10+0*10=10, and the weight summary is 10+10+10=30;

[0082] At this time, the calculation pool has summary data of A/B/C, and the return result = (80+10)/(90+30) = 0.75

[0083] Save all current factor values ​​(90) and factor weights (120) at this time

[0084] Assuming that the current customer K has three data A/C sets of K1, K2, and K3, calculated according to the above algorithm:

[0085] The reliability of K1 is 0.75;

[0086] The credibility of K2 is 0.8;

[0087] The credibility of K3 is 0.85;

[0088] Then it is recommended that the data account set of K3 has high credibility and is recommended to financial institutions. If the financial institution determines and lends money, the weight of all factors whose calculation factor value is 1 in K3 is increased by 1.

[0089] The calculation method of credibility in the embodiment of the present invention is described above, and the calculation device in the embodiment of the present invention is described below. Please refer to figure 2 , The computing device in the embodiment of the present invention includes:

[0090] The allocation unit 201 is configured to divide the calculation factors corresponding to the target data into N calculation factor sets according to preset rules, where N is a positive integer greater than 1;

[0091] The first calculation unit 202 is configured to calculate the factor value of each calculation factor in the N calculation factor sets in parallel through M calculation threads, where M is a positive integer not greater than N;

[0092] The second calculation unit 203 is configured to calculate and output the current credibility value of the target data according to the factor value of each calculation factor in the current set and the factor value of each calculation factor in the calculated target set. The set is one of the N calculation factor sets, and the calculated target set is included in the N calculation factor sets.

[0093] For ease of understanding, the computing device in the embodiment of the present invention is described in detail below. figure 2 On the basis shown, please refer to image 3 , image 3 It is a schematic diagram of another embodiment of a computing device in an embodiment of the present invention. Optionally, the second computing unit 303 includes:

[0094] The calculation module 3031 is configured to calculate the current credibility value of the target data in the following manner:

[0095] Z=(∑(a*b)+A)/(∑b+B);

[0096] The Z is used to represent the current credibility value of the target data, the a is used to represent the factor value of any calculation factor in the current set, and the b is used to represent the corresponding to any calculation factor The factor weight of the a*b is used to represent the standard value of any of the calculation factors; the A is used to represent the sum of the standard values ​​of the calculated factors in the calculated target set; the B is used to represent all The sum of the factor weights of the calculated factors in the calculated target set;

[0097] The display module 3032 is used to display the current credibility value of the target data.

[0098] Optionally, the allocation unit 301 is specifically configured to:

[0099] Classify the calculation factors corresponding to the target data according to the dimensions to obtain a T-type calculation factor, where the T is a positive integer not greater than the N;

[0100] The various calculation factors in the T-type calculation factors are processed in batches to obtain the N calculation factor sets.

[0101] Above Figure 2 to Figure 3 The computing devices in the embodiments of the present invention are described separately from the perspective of modular functional entities. The following describes the computing devices in the embodiments of the present invention in detail from the perspective of hardware processing, please refer to Figure 4 , An embodiment of the computing device 400 in the embodiment of the present invention includes:

[0102] Input device 401, output device 402, processor 403 and memory 404 (the number of processors 403 can be one or more, Figure 4 Take a processor 403 as an example). In some embodiments of the present invention, the input device 401, the output device 402, the processor 403, and the memory 404 may be connected by a bus or other means, where: Figure 4 Take the bus connection as an example.

[0103] Wherein, by calling the operating instructions stored in the memory 404, the processor 403 is configured to perform the following steps:

[0104] Divide the calculation factors corresponding to the target data into N calculation factor sets according to preset rules, where N is a positive integer greater than 1;

[0105] Calculate the factor value of each calculation factor in the N calculation factor sets in parallel by M calculation threads, where M is a positive integer not greater than N;

[0106] Calculate and output the current credibility value of the target data according to the factor value of each calculation factor in the current set and the calculated factor value of each calculation factor in the target set, and the current set is the N calculation factors A set in the set, and the calculated target set is included in the N calculation factor sets.

[0107] Optionally, the processor 403 is specifically configured to:

[0108] Calculate the current credibility value of the target data as follows:

[0109] Z=(∑(a*b)+A)/(∑b+B);

[0110] The Z is used to represent the current credibility value of the target data, the a is used to represent the factor value of any calculation factor in the current set, and the b is used to represent the corresponding to any calculation factor The factor weight of the a*b is used to represent the standard value of any of the calculation factors; the A is used to represent the sum of the standard values ​​of the calculated factors in the calculated target set; the B is used to represent all The sum of the factor weights of the calculated factors in the calculated target set;

[0111] The output device 402 is specifically configured to display the current credibility value of the target data.

[0112] Optionally, the processor 403 is specifically configured to: classify the calculation factors corresponding to the target data according to dimensions to obtain a T-type calculation factor, where T is a positive integer not greater than N;

[0113] The various calculation factors in the T-type calculation factors are processed in batches to obtain the N calculation factor sets.

[0114] The embodiment of the present invention also provides a computing system. The computing system includes a client and a server. The client is used to divide the calculation factors corresponding to the target data into N calculation factor sets according to preset rules. N is a positive integer greater than 1; the factor value of each calculation factor in the N calculation factor sets is calculated by parallel classification by M calculation threads, and the M is a positive integer not greater than N;

[0115] The client uploads the calculated factor value of each calculation factor in the N calculation factor sets to the server;

[0116] The server is used to calculate the current credibility value of the target data based on the factor value of each calculation factor in the current set and the factor value of each calculation factor in the calculated target set, and the current set is the One of the N calculation factor sets, and the calculated target set is included in the N calculation factor sets;

[0117] The server sends the current credibility value of the target data to the client;

[0118] The client outputs the current credibility value of the target data.

[0119] Those skilled in the art can clearly understand that, for the convenience and conciseness of the description, the specific working process of the above-described system, device, and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

[0120] In the several embodiments provided in this application, it should be understood that the disclosed system, device, and method may be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components may be combined or It can be integrated into another system, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.

[0121] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

[0122] In addition, the functional units in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above-mentioned integrated unit can be realized in the form of hardware or software functional unit.

[0123] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present invention essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , Including several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

[0124] As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions recorded in the embodiments are modified, or some of the technical features are equivalently replaced; these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.