CN111915368B

CN111915368B - System, method and medium for identifying customer ID in automobile industry

Info

Publication number: CN111915368B
Application number: CN202010753104.5A
Authority: CN
Inventors: 王海; 黄蓉蓉; 邬凯乐; 李红明; 张椿琳
Original assignee: Shanghai Shuce Software Co ltd
Current assignee: Shanghai Shuce Software Co ltd
Priority date: 2020-07-30
Filing date: 2020-07-30
Publication date: 2024-02-20
Anticipated expiration: 2040-07-30
Also published as: CN111915368A

Abstract

The invention provides a system, a method and a medium for identifying a customer ID in the automobile industry, comprising the following steps: and the basic data processing module is used for: collecting, integrating and standardizing ID data of a client host factory, and establishing a non-marketing list library for filtering ID information; ID repair module: analyzing the ID data according to a machine learning algorithm, and replacing the error ID with the correct ID; ID affinity calculation module: calculating the intimacy between IDs in the sub-network; and a transfer relation chain judging module: updating the ID pairing relation according to the intimacy between the IDs; and a visual display module: man-machine object layering of ID relationship in the sub-network and visual display of topology structure of the sub-network are carried out, so that inquiry of sub-network data and identification of ID are realized. The invention simplifies the ID relation network structure by adopting the SIL library to clean the ID data, reduces the subsequent data processing magnitude, and solves the problem of low calculation efficiency of the original graph.

Description

System, method and medium for identifying customer ID in automobile industry

Technical Field

The invention relates to the technical field of data analysis in the automobile industry, in particular to a system, a method and a medium for identifying a customer ID in the automobile industry.

Background

In recent years, with the development of the internet, especially the mobile internet, the customer demands show a personalized trend, the traditional full-coverage and bombing marketing mode in the automobile industry cannot meet the customer demands, and a host factory can truly drive the customer only by providing more differentiated products or services, and all this is based on the analysis and application of the customer data.

However, for a long time, the input and output of data analysis in the automobile industry are low, and stable value output cannot be formed, and the main problems are as follows:

(1) The vehicle-to-person relationship is not a simple one-to-one relationship between the vehicle and the vehicle owner, but a complex many-to-many relationship network of the vehicle purchasing person, the repairing person, the vehicle owner and the like exists.

(2) Since the main customer data is filled in manually by the dealer rather than acquired automatically by the machine, many customer information is erroneous.

Patent document CN109446215a (application number 201811294114.6) discloses a priority-based real-time ID pull engine method, which is mainly used for processing consumption behavior log data, extracting ID data from the consumption behavior log data, establishing a relationship between IDs, calculating weights of the ID relationships, sorting the weights, and calculating the ID attribution relationship step by step, thereby finally realizing client ID pull based on superID. This is also a method of ID identification and opening. But it is: 1. the method is only suitable for ID communication of consumption log data, and does not have the function of ID communication among multi-service data; 2. the method has no function of calculating the similarity of the IDs and repairing the error ID information.

Patent document CN110223168A (application number 201910546944.1) discloses a label propagation anti-fraud detection method and system based on enterprise relation graphs, which are mainly used for enterprise self-building blacklist libraries in the field of financial credit, and constructing blacklist relation graphs to estimate service anti-fraud probability. The construction and application method of the blacklist library is similar to the SIL list library, but does not comprise judgment of the client ID pairing relation, similarity calculation and error ID information restoration.

Disclosure of Invention

Aiming at the defects in the prior art, the invention aims to provide a system, a method and a medium for identifying a customer ID in the automobile industry.

The invention provides an automobile industry customer ID recognition system, which comprises:

and the basic data processing module is used for: collecting, integrating and standardizing ID data of a client host factory, and establishing a non-marketing list library for filtering ID information;

ID repair module: analyzing the ID data according to a machine learning algorithm, and replacing the error ID with the correct ID;

ID affinity calculation module: calculating the intimacy between IDs in the sub-network;

and a transfer relation chain judging module: updating the ID pairing relation according to the intimacy between the IDs;

and a visual display module: man-machine object layering of ID relationship in the sub-network and visual display of topology structure of the sub-network are carried out, so that inquiry of sub-network data and identification of ID are realized.

Preferably, when updating the ID pairing relationship, the ID pairing relationship conforming to the preset business logic is reserved, other relationships are classified into the main relationship through splitting and integration, and ID complementation is carried out on the defective sub-network.

The invention provides an automobile industry customer ID identification method, which comprises the following steps:

and a basic data processing step: collecting, integrating and standardizing ID data of a client host factory, and establishing a non-marketing list library for filtering ID information;

ID repair: analyzing the ID data according to a machine learning algorithm, and replacing the error ID with the correct ID;

ID affinity calculation step: calculating the intimacy between IDs in the sub-network;

a transfer relation chain determination step: updating the ID pairing relation according to the intimacy between the IDs;

visual presentation step: man-machine object layering of ID relationship in the sub-network and visual display of topology structure of the sub-network are carried out, so that inquiry of sub-network data and identification of ID are realized.

Preferably, the basic data processing step includes:

cleaning the customer ID information, wherein the cleaning rule comprises removing special symbols and merging identical IDs;

and establishing a non-marketing list library, wherein the non-marketing list library comprises mobile phone numbers, vehicle frame numbers, vehicle license plate numbers and identity card information.

Preferably, the ID repair step includes:

in the ID information and ID pairing relation, eliminating the client ID in the non-marketing list library, and updating the ID information and ID pairing relation;

aiming at the restoration of the identity card, the frame number and the license plate number, searching the error ID in the sub-network, finding the same kind of correct ID with the highest similarity, and replacing the error ID by using the same kind of correct ID;

aiming at repairing the mobile phone number, the repairing method comprises the following steps:

-finding the wrong ID and replacing with the same kind of correct ID with highest similarity in the same subnetwork;

and for the mobile phone numbers with the similarity higher than a certain threshold value in the same sub-network, respectively forming positive and negative samples according to the correct mobile phone numbers, wrong numbers and blank numbers confirmed by the communication operators, calculating the error probability of the mobile phone numbers through a machine learning algorithm, and replacing the mobile phone numbers with the highest possibility.

Preferably, the ID affinity calculating step includes:

based on the ID information and the ID pairing relation after ID repair, setting initial ID affinity, wherein the initial value ranges in [0,1], and the larger the initial value is, the larger the credibility of the ID pairing relation is;

setting system confidence coefficient according to service data, wherein the range is within [0,1], and the larger the confidence coefficient value is, the larger the confidence coefficient of the service data is;

and adjusting the ID affinity according to the initial ID affinity, the system confidence, the ID updating frequency and the ID freshness.

Preferably, the step of determining the shift relation chain includes:

ID relationship transfer, comprising:

-the ID relationship of the mobile phone number and the vehicle license number is transferred to the frame number and the vehicle license number;

-the ID relationship of the phone number and the car frame number is transferred to the identification card and the phone number or the identification card and the car frame number;

and (3) ID complementation, wherein if information is missing in the relationship between the identity card and the mobile phone number and the relationship between the identity card and the vehicle frame number, the virtual customer identity card ID is constructed for complementation.

Preferably, the visual presentation step includes:

a sub-network layered presentation step comprising:

-based on the pairing relationship comprising: the ID card and the mobile phone number, the ID card and the frame number, the frame number and the license plate number of the vehicle, and the ID is layered in three layers of man-machine objects;

-man-machine-object layered presentation of the sub-network associated with the entered customer ID by entering the customer ID.

Preferably, the visual presentation step includes:

and a step of inquiring statistical data: and displaying the related quantity of the sub-networks in a visual interface form, wherein the quantity comprises node information, relation information, node PR value and node access degree of the sub-networks.

According to the present invention there is provided a computer readable storage medium storing a computer program which when executed by a processor performs the steps of the method described above.

Compared with the prior art, the invention has the following beneficial effects:

1. the invention simplifies the ID relation network structure by adopting the SIL library to clean the ID data, reduces the subsequent data processing magnitude, and solves the problem of low calculation efficiency of the original graph;

2. according to the invention, the pairing of mobile phone numbers with high similarity is analyzed by formulating a customer ID data standard and combining a business rule and a machine learning algorithm, so that the problem of judging correct/incorrect mobile phone numbers is solved;

3. the invention divides the whole network into the sub-networks by adopting the sub-network calculation mode, reduces the calculation amount based on the operation of the sub-networks, and solves the problem of insufficient calculation capability of the whole network.

Drawings

Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:

FIG. 1 is a diagram of a system functional module relationship;

FIG. 2 is a flow chart of a basic data processing module;

FIG. 3 is a data flow of the ID repair module;

FIG. 4 is a data flow of the ID affinity calculation module;

fig. 5 is a data flow of the transfer relation chain determination module.

Detailed Description

The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the present invention, but are not intended to limit the invention in any way. It should be noted that variations and modifications could be made by those skilled in the art without departing from the inventive concept. These are all within the scope of the present invention.

Example 1:

the system collects, integrates and standardizes the customer IDs by opening up a plurality of service system data of a host factory, classifies and merges the man-vehicle relations, repairs the wrong IDs by combining a machine learning algorithm and service rules, establishes a customer ID relation network based on graph calculation, and realizes layered display of man (identity card) machine (mobile phone number) objects (frame numbers and vehicle license numbers).

The data quality of the customer ID of the host factory can be improved through the system; the data analysis team can query the ID relationship network in real time, so that the data analysis efficiency is improved; marketing team can implement marketing campaign based on more accurate customer ID, promotes the customer and touches the effect.

As shown in FIG. 1, the system comprises a basic data processing module, an ID repairing module, an ID affinity calculation module, a transfer relation chain judging module and a visual display module, wherein the total number of the modules is 5:

1. and the basic data processing module is used for:

-input: original business system data;

-output of: the cleaned ID information table, the ID pairing table and the SIL library;

ID repair module

-input: the cleaned ID information table, the ID pairing table and the SIL library;

-output of: a repaired ID information table and a repaired ID matching table;

ID affinity calculation

-input: initializing the affinity of the repaired ID pairing table;

-output of: an ID matching table with the intimacy adjusted;

4. transfer relation chain judging module

-input: an ID matching table with the intimacy adjusted;

-output of: the relation chain is linked with the ID matching table;

5. visualization module

-input: an ID matching table after the transfer of the relation chain and an ID information table after the repair;

-output of: visual query of ID relation chain;

application environment:

-big data platform: spark, hive, solr, mysql;

-an application server: django, nigix;

-big data platform: calculated in billions of data, requiring 200core, 1T memory;

-an application server: 16G memory, 8core.

1. A base data processing module for:

1. the collection, integration and standardization of the customer ID data of the host factory are realized.

2. A SIL list library (Suspected Identity Library non-marketing list library) is established for subsequent ID information filtering.

As shown in fig. 2, the data flow of the present module includes the following steps:

(1) The host factory service data required for positioning the system comprises sales data, after-sales data, customer relationship management data and member data.

(2) Defining customer ID information including a vehicle purchaser ID number, a mobile phone number, a car frame number registered in sales data, a repair person mobile phone number, a car frame number, a car license number registered in after-sales data, a customer mobile phone number, a car license number registered in customer relationship management data, a member registered mobile phone number in member data, and the like, which are defined as customer ID information.

(3) Other important information required for the system calculation is defined, including the date of after-sale store of the customer, the date of purchase of the customer, and the time when the customer ID information is established in the database.

(4) Customer ID information and other important information is accessed through the interface.

(5) And cleaning the client ID information, wherein the cleaning rule comprises special symbol removal and merging of the same ID.

(6) Initializing descriptive information for establishing a client ID, including:

ID compliance classification, the ID classification is set to "correct ID/false ID" by judging whether the client ID information meets the data standard. The data criteria for the customer ID information include: identification card data standard, mobile phone number data standard, frame number data standard and vehicle license plate number data standard.

id category including identification card, cell phone number, car frame number, vehicle license number.

ID information sources including sales data table, after-sales data table, customer relationship management data table, membership data table.

Whether the ID is a primary key or not is determined according to the service logic of the host factory and the source of the ID information.

(7) And (3) establishing an ID information table, and after finishing definition and cleaning of the ID information, extracting the customer ID information and descriptive information in the service data, and importing the ID information table.

(8) An ID pairing relation table is established, ID pairing in service data tables such as a sales data table, an after-sales data table, a customer relation management data table, a member data table and the like is extracted (ID pairing means that two customer IDs appear in the same record of the service data table), and the ID pairing relation table is formed through integration.

(9) An ID pairing relationship is initialized (ID pairing relationship refers to a business relationship in which five ID types are paired with each other). The ID pairing relationship is defined in terms of business logic as follows: "OWNER (i.e., OWNER)", "MEMBER (i.e., MEMBER)", "DELIVERER (i.e., repair man)", "BUYER (i.e., BUYER)", and "CONTACTOR (i.e., contact man)", and performs a merge deduplication of the ID pairing relationships repeated therein.

(10) And establishing an SIL list library, wherein the SIL list library comprises mobile phone numbers, vehicle frame numbers, vehicle license plate numbers and identity card information. The data of the SIL list library has the following two sources:

i. and importing the mobile phone number of the marketing personnel of the dealer and the mobile phone number of the staff of the host factory into the SIL list library in a file form.

And ii, carrying out association on the ID pairing relation table according to the primary key, and counting the association quantity between the client IDs, wherein the method comprises the following steps: the number of the identification cards related to different frame numbers, the number of the identification cards related to different mobile phone numbers, the number of the mobile phone numbers related to the frame numbers, the number of the vehicle license plate numbers related to the mobile phone numbers and the number of the vehicle license plate numbers related to the frame numbers. And setting corresponding thresholds for different ID association numbers based on internal business rules of the host factory, and importing the customer ID information exceeding the thresholds into the SIL list library.

2. An ID repair module for:

the basic information based on the ID information table is combined with a machine learning algorithm to analyze the ID information, and the correct ID is used for replacing the error ID.

As shown in fig. 3, the data flow of the present module is as follows:

1. and judging whether the client IDs of the ID information table and the ID pairing relation table are simultaneously present in the SIL list library.

2. And eliminating the client ID in the SIL list base in the ID information table and the ID pairing relation table, and updating the ID information table and the ID pairing relation table.

3. And establishing an integral ID relation network by a graph calculation method based on the filtered ID information table and the ID pairing relation table. The node in the whole ID relation network is the client ID, and the connecting line is the ID pairing relation.

4. Splitting a plurality of sub-networks from the whole ID relation network, wherein the sub-networks are mutually related through client IDs, and the sub-networks are not mutually communicated.

5. PR values (Page Rank value) are calculated, and the PR values of all IDs in the sub-network are calculated by using an algorithm, wherein the PR values represent the importance of a specific ID in the sub-network.

6. And (3) similarity calculation, namely calculating the similarity between all IDs in the same sub-network based on a text similarity algorithm, wherein the similarity calculation is based on the same type of ID (the same type of ID refers to the same identity card or the same mobile phone number and the like).

7. ID repair, which is divided into 2 parts:

i. aiming at the restoration of the ID card, the frame number and the license plate number (the data standard of the client ID is strict, the error ID can be efficiently detected), the error ID in the sub-network is searched, the similar correct ID with the highest similarity is found, and the similar correct ID is used for replacing the error ID.

Aiming at the repair of mobile phone numbers (only a small number of error IDs can be detected based on data standards, and a large number of error IDs cannot be identified according to the data standards), the following two repair methods exist:

(a) And finding out the wrong ID in the same repair mode as the ID card, the frame number and the license plate number, and replacing the wrong ID with the same kind of correct ID with the highest similarity in the same sub-network.

(b) And for the (two or more) mobile phone numbers with the similarity higher than a certain threshold value in the same sub-network, judging the mobile phone number with the highest error possibility through an algorithm. The threshold here is a numerical boundary commonly confirmed by business experience and quantitative analysis; the algorithm for judging the accuracy of the mobile phone number is to respectively form positive and negative samples according to the correct mobile phone number, wrong number and blank number confirmed by a communication operator, and calculate the error probability of the mobile phone number through a machine learning algorithm. After the judgment is completed, the mobile phone number with the highest error probability is marked as an error ID, and then the operation of the method a) is repeated.

8. And outputting the ID information table and the ID pairing relation table after ID repair.

3. An ID affinity calculation module for:

and calculating the intimacy between IDs in the subnetwork, and taking the intimacy as a data basis of a transfer relation chain judging module.

As shown in fig. 4, the data flow of the present module is as follows:

1. based on the ID information table and the ID pairing relation table after ID repair, setting initial ID affinity (affinity refers to the measurement of the association relation between IDs in the same sub-network, namely, the possibility that the IDs of different types belong to the same person or organization is judged), and comprehensively determining an initial value by the ID pairing relation and business experience, wherein the range is within [0,1], and the larger the value is, the larger the credibility of the ID pairing relation is indicated for subsequent affinity calculation.

2. And setting the system confidence according to the service data reliability determined by service investigation, wherein the larger the value is in the range of [0,1], the larger the reliability of the service data is, and the larger the reliability of the ID derived from the service data is.

3. The update ID affinity is calculated based on the initial ID affinity, system confidence, ID update frequency, and ID freshness (the length of time since the last update was from the current time point).

4. ID affinity adjustment, the ID affinity is adjusted based on 3 categories between ID pairs, the 3 categories are:

i. the relation between the frame number and the license plate number of the vehicle can only be one-to-one in the same time point.

And ii, the relationship between the identity card and the mobile phone number, and the relationship between the identity card and the vehicle frame number can be one-to-many in the same time point.

The relation between the frame number and the mobile phone number can be in a unified time point

Based on the above 3 types of relationships, a master relationship and a slave relationship are calculated by using an algorithm, wherein the ID relatedness of the master relationship is adjusted to be 1, and the ID relatedness of the slave relationship is smaller than 1 and is at least 0.

5. And outputting the ID pairing relation table with the adjusted ID affinity.

4. A transfer relationship chain determination module for:

and reserving an ID pairing relation conforming to the business logic, classifying other relations into a main relation through splitting and integrating, carrying out ID complementation on the defective sub-network, and updating the ID pairing relation.

As shown in fig. 5, the data flow of the present module is as follows:

1. ID relation screening, in a sub-network, a plurality of ID pairing relations exist, and three relations conforming to business logic are finally reserved: the identification card and the mobile phone number, the identification card and the vehicle frame number, and the vehicle frame number and the vehicle license plate number.

2.ID relationship transfer, there are 2 transfer types:

i. the ID relationship between the mobile phone number and the vehicle license plate number is transferred to the frame number and the vehicle license plate number

ii, transferring the ID relationship between the mobile phone number and the frame number to the identity card and the mobile phone number or the identity card and the frame number

3. And if the information is missing in the relation between the identity card and the mobile phone number and the relation between the identity card and the vehicle frame number, constructing a virtual customer identity card ID for complement.

4. And updating the ID pairing relation, judging the ID pairing relation based on the adjusted affinity in the ID pairing table, and updating the ID pairing relation into a 'pre owner' relation if the adjusted affinity is 0.

5. The ID pairing relationship is updated with the following 3 update types:

i. the unique relation judgment is that the relation of 'Owner' and 'BUYER' in the same sub-network has uniqueness, and if a plurality of 'Owner' or 'BUYER' occur, the party with high affinity is reserved.

if a plurality of relationships exist in one ID pair, the relationship with high affinity is obtained.

if there is only one relation in the same sub-network, it is set as OWNER.

5. A visual presentation module for:

and man-machine object layering of ID relationship in the sub-network and visual display of the topology structure of the sub-network are carried out, so that the inquiry of the sub-network data is realized.

The module comprises the following 2 sub-modules:

(1) The sub-network layer presents sub-modules for:

i. the person (identity card) machine (mobile phone number) object (vehicle frame number, vehicle license number) is layered, and in the subnetwork, the ID is layered in three layers of the person and the machine object based on 3 pairing relations (identity card and mobile phone number, identity card and vehicle frame number, vehicle frame number and vehicle license number).

And ii, carrying out sub-network layered presentation, namely carrying out man-machine object layered presentation on the sub-network related to the input client ID by inputting the client ID.

(2) A statistics query sub-module for:

the related quantity of the sub-network is displayed in the form of a visual interface, wherein the quantity comprises node information, relation information, node PR value and node access degree (a measure of the importance of the connection of the nodes in the network) of the sub-network.

Example 2:

the system is provided for a data analysis department of a host factory and is applied to accurate marketing of vehicle owners or customer relationship management. The system is responsible for updating or repairing the client ID information according to the calculation logic every day, and simultaneously provides a visual function to realize layered display of the client ID relationship subnetwork. The system changes the original working mode of integrating and analyzing business data by hands, improves the data analysis efficiency, positions the problem data in time, corrects the problem data and realizes accurate customer access.

Before the system is implemented, enterprises do not recognize SIL lists, and after the system is optimized, the following main implementation effects are achieved:

(1) The SIL list of the identified frame numbers accounts for 0.03% of the total frame numbers.

(2) The SIL list of identification numbers accounts for 0.01% of all identification numbers.

(3) The list of identification mobile phone numbers SIL accounts for 0.5% of all mobile phone numbers.

(4) The identification license plate number SIL list accounts for 0.02% of the total license plate numbers.

Before the system is implemented, enterprises do not repair the customer ID information, and after the system is optimized, the following main implementation effects are achieved:

(1) The successful repair mobile phone number accounts for 75% of the similar mobile phone number, and the repair identity card accounts for 78% of the similar identity card.

(2) The repair license plate number accounts for 16% of the similar license plate number.

Those skilled in the art will appreciate that the systems, apparatus, and their respective modules provided herein may be implemented entirely by logic programming of method steps such that the systems, apparatus, and their respective modules are implemented as logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc., in addition to the systems, apparatus, and their respective modules being implemented as pure computer readable program code. Therefore, the system, the apparatus, and the respective modules thereof provided by the present invention may be regarded as one hardware component, and the modules included therein for implementing various programs may also be regarded as structures within the hardware component; modules for implementing various functions may also be regarded as being either software programs for implementing the methods or structures within hardware components.

The foregoing describes specific embodiments of the present invention. It is to be understood that the invention is not limited to the particular embodiments described above, and that various changes or modifications may be made by those skilled in the art within the scope of the appended claims without affecting the spirit of the invention. The embodiments of the present application and features in the embodiments may be combined with each other arbitrarily without conflict.

Claims

1. A vehicle industry customer ID identification system, comprising:

and a visual display module: man-machine object layering of ID relationship in the sub-network and visual display of topology structure of the sub-network are carried out, so that inquiry of sub-network data and identification of ID are realized;

when updating the ID pairing relation, reserving the ID pairing relation which accords with the preset business logic, classifying other relations into a main relation through splitting and integrating, and carrying out ID complementation on the defective sub-network;

the non-marketing list library comprises mobile phone numbers, frame numbers, vehicle license plate numbers and identity card information;

the ID repair module includes:

the method for repairing the mobile phone number comprises the following steps:

for the mobile phone numbers with the similarity higher than a certain threshold value in the same sub-network, respectively forming positive and negative samples according to the correct mobile phone numbers, wrong numbers and blank numbers confirmed by a communication operator, calculating the error probability of the mobile phone numbers through a machine learning algorithm, and replacing the mobile phone numbers with the highest possibility;

the ID affinity calculation module includes:

2. A method for identifying a customer ID in the automotive industry, comprising:

visual presentation step: man-machine object layering of ID relationship in the sub-network and visual display of topology structure of the sub-network are carried out, so that inquiry of sub-network data and identification of ID are realized;

the basic data processing step comprises the following steps:

establishing a non-marketing list library, wherein the non-marketing list library comprises mobile phone numbers, frame numbers, license plate numbers of vehicles and identity card information;

the ID repair step includes:

the ID affinity calculation step comprises the following steps:

3. The automotive industry customer ID recognition method according to claim 2, characterized in that the transfer relation chain determination step includes:

ID relationship transfer, comprising:

4. The automotive industry customer ID recognition method of claim 2, wherein the visual presentation step includes:

a sub-network layered presentation step comprising:

5. The automotive industry customer ID recognition method of claim 2, wherein the visual presentation step includes:

6. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the method of any one of claims 2 to 5.