A car owner information intelligent identification system for car washing service

Abstract

The application relates to the technical field of information recognition, and discloses a car owner information intelligent recognition system for car washing services; the system comprises a data query module, a first portrait fusion module, a second portrait construction module and a package recommendation module; the data query module extracts comprehensive car owner information from available data; the first portrait fusion module fuses user labels into a user car washing portrait; the second portrait construction module constructs a real-time service portrait with annotations; and the package recommendation module sequentially recommends service packages to a user end; the car owner information intelligent recognition system can integrate historical car washing service information and current individual demand service information of a user into one, realizes double-portrait construction of the user in a past time period and a current time period, avoids the problem that the car owner information is limited in comprehensiveness when a portrait is constructed once, and ensures that the screened service package can be highly matched with the car owner information.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information recognition, and more particularly, to a car owner information intelligent recognition system for car washing services. BACKGROUND

[0002] With the continuous growth of the number of cars and the continuous development of the Internet, the operation of integrating car washing service information online and providing reasonable car washing service recommendations to car owners has gradually been accepted by users. In order to ensure that the car washing service platform can recommend reasonable and accurate car washing service information to users, it is necessary to intelligently recognize car owner information and recommend corresponding car washing service packages based on the recognition results.

[0003] The patent application with publication number CN112579911A discloses a car washing information recommendation method and system. The license plate information and body color information of the target vehicle are determined by the image containing the target vehicle information obtained, and then the recognition degree is determined according to the license plate information and body color information and the corresponding new car information. Then, it is determined whether the target vehicle needs to be serviced according to whether the total recognition degree of the two is greater than a first preset value. If it is not greater than the first preset value, it is determined that the target vehicle is in a state that needs to be serviced, and then the corresponding car maintenance service location is matched according to the location of the target vehicle, and the name and location of the location are pushed to the target vehicle, so that the car can be maintained in time, thereby reducing the impact on the service life of the vehicle to a certain extent.

[0004] The existing recognition system identifies the car owner information on the user side by collecting multi-dimensional car washing service data of the car owner in the past time period, analyzing the collected car washing service data, and constructing a user portrait based on the user portrait. The required car washing service package is filtered based on the user portrait. However, this method still has some drawbacks. Since the user portrait is constructed based on multi-dimensional data in the past time period, the user portrait can only represent the car owner information in the past time period, and cannot effectively combine the personalized car washing needs of the car owner in the current time period. Therefore, the method of constructing a user portrait once has the problem of time period limitation, which leads to low correlation and matching degree of the subsequent car washing service package and the car owner information, and further reduces the filtering accuracy of the car washing service package.

[0005] In view of this, the present application provides a car owner information intelligent recognition system for car washing services to solve the above problems. SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a car owner information intelligent recognition system for car washing services, applied to a car washing service platform, comprising:

[0007] a data query module, configured to receive a service request of a user in a user terminal, query available data associated with the user from a database in a request state, and extract comprehensive car owner information from the available data;

[0008] a first portrait fusion module, configured to perform abnormal correction on abnormal data in the comprehensive car owner information, extract a user tag from the comprehensive car owner information after the abnormal correction, and fuse the user tag into a user car washing portrait;

[0009] a second portrait construction module, configured to determine a car washing service period, collect individual demand characteristics of the user terminal in the car washing service period, and fuse the individual demand characteristics with the user car washing portrait to construct a real-time service portrait with annotations;

[0010] a package recommendation module, configured to select a service package suitable for the real-time service portrait from a database of a service terminal based on a trained package screening model, calculate an adaptation index of the service package, and recommend the service package to the user terminal in sequence.

[0011] Further, the method for querying available data is:

[0012] the user's surname and the user's contact information corresponding to the original data in the database are queried one by one, denoted as original surname and original contact information;

[0013] the user's surname and the user's contact information corresponding to the input service request are extracted respectively, denoted as request surname and request contact information;

[0014] When the original surname and the original contact information of the original data are consistent with the request surname and the request contact information respectively, the original data is recorded as available data.

[0015] Further, the comprehensive car owner information includes service consumption data, vehicle state data and service environment data;

[0016] The service consumption data includes the number of car washes, the car wash time, the service package, the consumption amount and the discount amount;

[0017] The vehicle state data includes the vehicle type and the number of paint damage;

[0018] The service environment data includes the weather type and the ambient temperature.

[0019] Further, the method for abnormal correction of abnormal data is:

[0020] The recording time of the A service consumption data is queried one by one through the time stamp, and the A service consumption data is arranged in order according to the time;

[0021] A vehicle state data and A service environment data in the same record time with A service consumption data are respectively summarized, and A data packets are generated;

[0022] The A data packets are sequentially checked for abnormalities based on the absence of repetition and missing phenomena in the service consumption data, vehicle state data, and service environment data in the data packets;

[0023] Data packets with missing phenomena and repetition are recorded as abnormal data, and the missing data and repeated data corresponding to the missing phenomena and repetition are recorded respectively;

[0024] The repeated data in the abnormal data is removed, the original data corresponding to the missing data is traced back from the database, and the traced original data is imported into the missing position of the abnormal data to obtain the corrected comprehensive owner information.

[0025] Further, the user label includes a car wash frequency value, a package preference, a price sensitivity, a vehicle health degree, and a service environment attribute;

[0026] The fusion method of the user car wash profile is:

[0027] A basic profile with three inner and outer wrapped distribution ring contours is constructed, and the three ring contours are sequentially recorded as an environment unit, a vehicle unit, and a user unit in an inner-to-outer manner;

[0028] The service environment attribute is imported into the environment unit, the vehicle health degree is imported into the vehicle unit, and the car wash frequency value, the package preference, and the price sensitivity are imported into the user unit;

[0029] Data channels are established between the environment unit and the vehicle unit, and between the vehicle unit and the user unit, respectively, to facilitate the conversion of the basic profile to the user car wash profile.

[0030] Further, the determination method of the car wash service period is:

[0031] All original data of the user end are queried from the database, the data attributes of the original data are identified one by one, and the original data with data attributes of consultation and browsing are recorded as valid data;

[0032] The recording time of all valid data is queried one by one through the time stamp, the recording time located at the last position is recorded as the calibration time according to the chronological order, and the calibration time and the recording time of the valid data are imported into the same time line;

[0033] The first recording time after the calibration time is recorded as the period starting point, the current time is recorded as the period ending point, and the period from the period starting point to the period ending point is recorded as the car wash service period.

[0034] Further, the individual demand features include type features, price features and preferential features;

[0035] The collection method of the type features, the price features and the preferential features is as follows:

[0036] The natural language processing technology is used to identify the character groups and the number groups of the B effective data in the car washing service cycle one by one;

[0037] The effective data with the character groups of fine washing, general washing and rough washing are recorded as type data, the effective data with the character groups of member and non-member are recorded as price data, and the effective data with the character groups of coupons, full reduction and discount are recorded as preferential data;

[0038] The time stamps are used to mark the reading and browsing time of the type data, the price data and the preferential data one by one, and the first time length, the second time length and the third time length are obtained by adding all the reading and browsing times;

[0039] The character groups and the number groups corresponding to the maximum value of the first time length, the maximum value of the second time length and the maximum value of the third time length are combined to generate the type features, the price features and the preferential features.

[0040] Further, the construction method of the real-time service portrait is as follows:

[0041] Three feature boxes are set at equal angles on the user unit of the user car washing portrait, and the three feature boxes are sequentially numbered as No. 1, No. 2 and No. 3 to generate No. 1 feature box, No. 2 feature box and No. 3 feature box;

[0042] The type features, the price features and the preferential features are respectively introduced into the No. 1 feature box, the No. 2 feature box and the No. 3 feature box to generate feature interpretation;

[0043] Two interpretation positions are established between the No. 1 feature box and the No. 2 feature box, and between the No. 2 feature box and the No. 3 feature box, and the user surname and the user contact information are respectively introduced into the two interpretation positions to generate identity interpretation, thereby constructing the real-time service portrait with remark interpretation.

[0044] Further, the training method of the package screening model is as follows:

[0045] A plurality of real-time service portraits and corresponding service packages are collected in advance, and the car washing frequency value, the package preference, the price sensitivity, the vehicle health degree, the service environment attribute, the type features, the price features and the preferential features in the real-time service portraits are summarized and marked as feature vectors to obtain a plurality of feature vectors;

[0046] The service packages are converted into labels corresponding to feature vectors, with one feature vector corresponding to one label, forming a set of training data. Multiple sets of training data constitute a training set. The labeled training features are divided into a training set and a test set. The training set is used to train the package selection model, and the test set is used to test the package selection model. A preset error threshold is set. When the mean of the prediction errors of all training features in the test set is less than the preset error threshold, the package selection model is obtained.

[0047] Furthermore, the adaptation index is calculated as follows:

[0048] The service items in the D service packages are separated one by one using word segmentation technology, and each service item is compared with the real-time service profile.

[0049] Service items that overlap with package preferences, price sensitivity, vehicle health, service environment attributes, type characteristics, price characteristics, and discount characteristics are recorded as valid items, and the number of valid items is counted to obtain D valid values;

[0050] After comparing the D valid values ​​with the total number of service items in the D service packages, D valid ratios are obtained.

[0051] Find out the package amount and actual payment amount for each of the D service packages, then compare the difference between the package amount and the actual payment amount to calculate the D discount rates;

[0052] After assigning corresponding proportional coefficients to the D effective ratios and D discount ratios, they are summed to obtain the D adaptation indices.

[0053] The technical effects and advantages of the intelligent identification system for car owner information in car wash services according to the present invention are as follows:

[0054] The application can collect the comprehensive car washing service information of the user in the past time period in multiple dimensions by extracting the comprehensive car owner information, expand the data dimension of the intelligent identification of the car owner information, and eliminate and supplement the abnormal data with missing and repeated phenomena in the comprehensive car owner information by the way of abnormal correction of the comprehensive car owner information, improve the integrity and uniqueness of the comprehensive car owner information, and through the way of building the user car washing portrait and real-time service portrait, the historical car washing service information and the current individual demand service information of the user can be fused, thereby realizing the dual construction effect of the portrait of the user in the past time period and the current time period, avoiding the problem of the limitation of the comprehensiveness of the car owner information in the construction of the portrait once, greatly improving the intelligent identification accuracy of the car owner information, and combining the intelligent screening processing of the package screening model, the service package matched with the car owner information and associated can be quickly and accurately screened out, ensuring that the screened service package can maintain high matching and association effect with the car owner information, and providing reasonable and accurate recommendation results for the subsequent car washing service of the user. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A schematic diagram of the architecture of a car owner information intelligent identification system for car washing service is provided for the first embodiment of the application.

[0056] Figure 2 A flowchart of a car owner information intelligent identification method for car washing service is provided for the second embodiment of the application. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0058] Embodiment one: please refer to Figure 1 The car owner information intelligent identification system for car washing service described in the embodiment is applied to a car washing service platform, which comprises:

[0059] The data query module receives the service request of the user in the user terminal, queries the available data associated with the user from the database in the request state, and extracts the comprehensive car owner information from the available data, wherein the comprehensive car owner information comprises service consumption data, vehicle state data and service environment data.

[0060] The user end refers to a client port capable of providing data input, data feedback and result reception for a user, and is used to realize the data bidirectional interaction effect between the user and the car washing service platform. Specifically, the user end includes but is not limited to an APP installed on a mobile phone of the user, a WeChat public number and a WeChat mini program and the like.

[0061] The service request refers to the request data sent by the user end and containing the information related to the car washing demand, i.e., can be used as the data identification basis corresponding to the car washing demand sent by the user to the car washing service platform. Specifically, the service request includes but is not limited to a car washing service item query demand, a car washing service price query request, a car washing service discount query demand and the like. When the user request is input to the user end, the user can input by means of text editing, voice input, project button selection and the like.

[0062] The request state refers to the real-time working state of the car washing service platform after receiving the service request of the user, i.e., can distinguish the current working state and mode of the car washing service platform, and ensure that the car washing service platform is in the real-time opened request state only after receiving the service request of the user end, so that the car washing service platform can be opened when needed, and remain silent or closed state when not needed, which can not only avoid the risk of increasing data information leakage due to the long-time opening of the user end, but also prevent the car washing service platform from sending pop-up ads to the user end at irregular intervals and causing unnecessary negative interference to the user.

[0063] When the car washing service platform is in the request state, the diversified original data associated with the user need to be queried from the database at this time, and the diversified original data is recorded as available data, and all the car washing service conditions of the user saved and recorded in the car washing service platform are represented by the available data;

[0064] When judging whether the data is associated with the user, the consistency of the attributes of the original data in the database and the attributes of the user needs to be judged. The attribute is used to specifically represent the type of data. For example, the attributes include the user's surname, the user's contact information, the user's address, the user's vehicle type and the like.

[0065] Specifically, the user's surname and the user's contact information corresponding to the original data in the database are queried one by one, and are recorded as the original surname and the original contact information.

[0066] The user's surname and the user's contact information corresponding to the input service request are extracted respectively, and are recorded as the request surname and the request contact information.

[0067] When the original surname and the original contact information are consistent with the requested surname and the requested contact information respectively, it indicates that the original data recorded and saved in the database is associated with the user, and the original data consistent with the requested surname and the requested contact information is recorded as available data.

[0068] The comprehensive car owner information refers to the data in the available data that can represent the car washing service conditions of the user in the historical time period.

[0069] Specifically, the comprehensive car owner information includes service consumption data, vehicle state data and service environment data.

[0070] The service consumption data refers to the data in the available data that can specifically represent the relevant dynamic conditions of the user in the car washing service process, and is the data in the car owner information that is specific to the user.

[0071] Specifically, the service consumption data includes the number of car washes, the car wash time, the service package, the consumption amount and the discount amount; wherein the number of car washes refers to the total number of all car washing services of the user in the past time period recorded and saved in the car washing service platform, the car wash time refers to the specific time corresponding to each car wash of the user recorded and saved in the car washing service platform, the service package refers to the car wash package type corresponding to each car wash of the user recorded and saved in the car washing service platform, the consumption amount refers to the actual payment price corresponding to each car wash of the user recorded and saved in the car washing service platform, and the discount amount refers to the discount range corresponding to each car wash of the user recorded and saved in the car washing service platform.

[0072] The vehicle state data refers to the specific data in the available data that can represent the relevant dynamic conditions of the user's vehicle in the car washing service process, and is the data in the car owner information that is specific to the user's vehicle.

[0073] The vehicle state data includes the vehicle type and the number of paint damage; wherein the vehicle type refers to the specific type of the vehicle driven by the user recorded and saved in the car washing service platform, and the vehicle type includes but is not limited to sedan, urban off-road vehicle, small truck, etc.; the number of paint damage refers to the number of damaged locations on the appearance paint of the vehicle of the user at each car wash recorded and saved in the car washing service platform.

[0074] The service environment data refers to the specific data in the available data that can represent the external environment state of the user at each car wash.

[0075] The service environment data includes the weather type and the environmental temperature; wherein the weather type refers to the weather condition corresponding to each car wash of the user recorded and saved in the car washing service platform, and the weather type includes but is not limited to sunny, rainy, snowy, etc.; the environmental temperature refers to the external environmental temperature corresponding to each car wash of the user recorded and saved in the car washing service platform.

[0076] The first image fusion module corrects the abnormal data in the comprehensive car owner information, extracts a user label from the corrected comprehensive car owner information, and fuses the user label into a user car washing image.

[0077] After obtaining the comprehensive car owner information, it is necessary to ensure the integrity, accuracy and reasonableness of the comprehensive car owner information, so as to avoid abnormal data such as abnormal value size and repeated data type in the comprehensive car owner information. Therefore, it is necessary to perform an abnormal correction operation on the comprehensive car owner information, that is, to perform operations such as elimination, supplementary recording and arrangement on the abnormal data in the comprehensive car owner information.

[0078] In order to ensure the pertinence and accuracy of the abnormal correction operation, it is necessary to identify the service consumption data, vehicle state data and service environment data in the comprehensive car owner information, and perform an abnormal correction operation on the existing abnormal data according to the identified results.

[0079] The abnormal correction method of the abnormal data is as follows:

[0080] The time stamp is used to query the record time of the A service consumption data one by one, and the A service consumption data is arranged in order according to the time sequence.

[0081] A data package is generated by respectively summarizing A vehicle state data and A service environment data at the same record time as the A service consumption data. The data package is used to summarize different types of available data, so that there is only one service consumption data, vehicle state data and service environment data in a data package.

[0082] The A data packages are sequentially checked for abnormalities based on the absence of repetition and absence of the service consumption data, vehicle state data and service environment data in the data package.

[0083] The data package with missing data and repeated data is recorded as abnormal data, and the missing data and repeated data are recorded as missing data and repeated data, respectively.

[0084] The repeated data in the abnormal data is eliminated, the original data corresponding to the missing data is traced back from the database, and the traced original data is imported into the missing position of the abnormal data to obtain the corrected comprehensive car owner information. The missing position refers to the specific position of the missing data in the data package corresponding to the abnormal data, and serves as the supplementary import position of the subsequently traced original data.

[0085] It should be noted that the missing data refers to all data in the data packet without containing service consumption data, vehicle state data and service environment data, and the repeated data refers to the existence of two same service consumption data, vehicle state data and service environment data in the data packet. When supplementing data from the database, the record time corresponding to the data packet is taken as the standard, and other service consumption data, vehicle state data and service environment data in the data packet are analyzed to determine the specific data corresponding to the missing data, and then the integrity of each different type of data in the comprehensive car owner information is realized.

[0086] After the comprehensive car owner information is abnormally corrected, the user label needs to be extracted from the abnormally corrected comprehensive car owner information at this time, so that the user label can intuitively, simply and accurately represent the recorded and saved car washing service of the user in the car washing service platform;

[0087] The user label includes a car washing frequency value, a package preference, a price sensitivity, a vehicle health degree and a service environment attribute.

[0088] Specifically, the car washing frequency value refers to the car washing frequency of the user in the comprehensive car owner information in a unit of time, that is, the car washing frequency of the user in a unit of time can be represented. In this embodiment, the time when the service consumption data is generated for the first time is recorded as the starting time, the time when the service consumption data is generated for the last time is recorded as the ending time, and the car washing frequency value is obtained by comparing the number of car washings with the time length between the starting time and the ending time.

[0089] The package preference refers to the service package with the most number of times when the user washes the car in the comprehensive car owner information, that is, the preference of the user for the specific type of car washing can be represented. In this embodiment, the package preference is recorded as the service package corresponding to the maximum number by counting the number corresponding to all service packages.

[0090] The price sensitivity refers to the proportion of the number of times of using the car washing coupon when the user washes the car in the comprehensive car owner information, that is, the price sensitivity of the user for car washing can be represented. In this embodiment, the price sensitivity is obtained by comparing the number of times of using the coupon with the number of car washings by counting all the times of using the coupon when washing the car.

[0091] The vehicle health degree refers to the representation of the severity of the damage to the appearance of the vehicle when the user washes the car in the comprehensive car owner information, that is, the health degree of the appearance of the vehicle can be represented. In this embodiment, the average value is obtained by counting the number of damaged vehicle paint surfaces each time the car is washed.

[0092] The service environment attribute refers to an indication of the user's environmental preference during car washing in the comprehensive car owner information, that is, the type of weather environment during car washing can be indicated; in this embodiment, the service environment attribute is obtained by querying the weather type and environmental temperature during each car washing, combining them into different service environment attributes, and then selecting the service environment attribute corresponding to the maximum value after counting the number of each service environment attribute; for example, when the weather type is sunny and the environmental temperature is 30 degrees Celsius, the service environment attribute formed at this time is "sunny: 30".

[0093] After obtaining the user label, the user label at this time can comprehensively represent the user's car washing situation in the past time period, and through the ordered fusion of the user label, a user car washing portrait that conforms to the user's past car washing performance is constructed, ensuring that the user car washing portrait can comprehensively represent the user's personal preference for car washing, car washing vehicle condition and car washing environmental factors;

[0094] The fusion method of the user car washing portrait is:

[0095] A basic portrait with three inner and outer wrapped distribution annular contours is constructed, and the three annular contours are sequentially recorded as an environmental unit, a vehicle unit and a user unit from inside to outside; the basic portrait refers to a blank portrait without any substantial content, the annular contour is a component unit used to construct the basic portrait, and also provides an independent position limitation for the subsequent import of the user label, and the structure of the annular contour with inner and outer wrapped distribution can ensure the independence of each annular contour, preventing mutual interference between the data in different annular contours;

[0096] The service environment attribute is imported into the environmental unit, the vehicle health degree is imported into the vehicle unit, and the car washing frequency value, the package preference and the price sensitivity are imported into the user unit;

[0097] Data channels are established between the environmental unit and the vehicle unit, and between the vehicle unit and the user unit, respectively, to facilitate the conversion of the basic portrait to the user car washing portrait. The data channel is used to provide a correlated data transmission channel between the two adjacent units, so that the user labels between the two adjacent units can maintain a dynamic correlation effect.

[0098] It should be noted that the user car washing portrait is not fixed, and the user car washing portrait can be updated and replaced in the environmental unit, the vehicle unit and the user unit according to the different car washing frequency value, package preference, price sensitivity, vehicle health degree and service environment attribute of the user generated in real time in the car washing service platform, so as to ensure that the user car washing portrait is consistent with the real-time car washing service related data of the user, thereby avoiding the negative effects caused by the untimely updating of the user car washing portrait data.

[0099] The second image construction module determines a car washing service period, collects the individual demand characteristics of the user end in the car washing service period, and fuses the individual demand characteristics with the user car washing image to construct a real-time service image with annotations;

[0100] The car washing service period refers to the maximum time span of the user-related data recorded and saved by the car washing service platform, i.e., the time limit for collecting the relevant data generated and recorded by the user in the car washing service platform, thereby playing a time limiting role for the subsequent collection of individual demand characteristics;

[0101] The determination method of the car washing service period is:

[0102] All original data of the user end are queried from the database, the data attributes of the original data are identified one by one, and the original data with data attributes for reference and browsing are recorded as valid data;

[0103] The recording time of all valid data is queried one by one through the time stamp, the recording time located at the last position is recorded as the calibration time according to the chronological order, and the calibration time and the recording time of the valid data are imported into the same time line;

[0104] The first recording time after the calibration time is recorded as the cycle starting point, the current time is recorded as the cycle ending point, and the period between the cycle starting point and the cycle ending point is recorded as the car washing service period.

[0105] After determining the car washing service period, the relevant car washing service information concerned and referred to by the user in the car washing service period needs to be collected, so as to simply and accurately represent the individualized car washing demand of the user in the car washing service period;

[0106] The individual demand characteristics include type characteristics, price characteristics, and preferential characteristics. Specifically, the type characteristics are used to represent the car washing package service type concerned by the user in the car washing service period, the price characteristics are used to represent the actual payment price of the car washing package concerned by the user in the car washing service period, and the preferential characteristics are used to represent the preferential strength of the car washing package concerned by the user in the car washing service period;

[0107] The collection method of the type characteristics, the price characteristics, and the preferential characteristics is:

[0108] The text groups and the number groups of the B valid data in the car washing service period are identified one by one through natural language processing technology; the text groups and the number groups are respectively used to simply represent the text part and the number part of each service item in the valid data, and serve as the smallest component part for constructing the type characteristics, the price characteristics, and the preferential characteristics;

[0109] The effective data of the character group is denoted as type data, the effective data of the character group of member and non-member is denoted as price data, and the effective data of the character group of coupon, full reduction and discount is denoted as preferential data;

[0110] The reading and browsing time of the type data, the price data and the preferential data is marked one by one through the timestamp, and the first duration, the second duration and the third duration are obtained by accumulating all the reading and browsing times;

[0111] The character group and the number group corresponding to the maximum value of the first duration are composed to generate a type feature;

[0112] The character group and the number group corresponding to the maximum value of the second duration are composed to generate a price feature;

[0113] The character group and the number group corresponding to the maximum value of the third duration are composed to generate a preferential feature.

[0114] After obtaining the type feature, the price feature and the preferential feature, the type feature, the price feature and the preferential feature can accurately represent the related data of the car washing service package type that the user is interested in recently, and serve as the basis for subsequent comprehensive representation of the user's car washing service demand, thereby constructing a real-time service portrait capable of representing the car washing service related information that the user is interested in recently.

[0115] In constructing the real-time service portrait, the type feature, the price feature and the preferential feature need to be effectively fused with the user car portrait, so that the user car portrait can represent the user's car washing preference in the past time period, and the type feature, the price feature and the preferential feature can represent the user's car washing attention preference in the recent period, thereby improving the individualization and accuracy of the real-time service portrait.

[0116] The note interpretation is a specific representation for comprehensively noting the individual demand characteristics, the user contact information and the user surname, that is, the actual situation of each user on the user side can be accurately and comprehensively displayed; specifically, the note interpretation includes characteristic interpretation and identity interpretation; wherein the characteristic interpretation is used to represent the individual demand characteristics of the user, and the identity interpretation is used to represent the car owner identity information of the user;

[0117] The construction method of the real-time service portrait is:

[0118] Three feature boxes are set on the user unit of the user car portrait at equal angles, and the three feature boxes are sequentially numbered 1, 2 and 3 to generate a No. 1 feature box, a No. 2 feature box and a No. 3 feature box; the feature box is used to provide a position limit for importing individual demand characteristics, so as to ensure that the type feature, the price feature and the preferential feature can be associated and fused with the user car portrait in order and independently.

[0119] The type feature, the price feature and the preferential feature are respectively introduced into the No. 1 feature box, the No. 2 feature box and the No. 3 feature box to generate feature interpretation;

[0120] Two interpretation positions are respectively established between the No. 1 feature box and the No. 2 feature box and between the No. 2 feature box and the No. 3 feature box, and the user surname and the user contact information are respectively introduced into the two interpretation positions to generate identity interpretation, thereby constructing the real-time service image with remark interpretation. The interpretation position is a position limit for introducing the user surname and the user contact information, and can intuitively and accurately display the corresponding owner identity information of the real-time service image.

[0121] The package recommendation module is configured to select a service package suitable for the real-time service image from a database of a service end based on a trained package screening model, calculate an adaptation index of the service package, and recommend the service package to a user end in sequence.

[0122] The service end is a client port for providing the user with a car washing service package and other related data queries, and is configured to realize a data bidirectional interaction effect between a merchant and a car washing service platform. Specifically, the service end includes, but is not limited to, an APP installed on a mobile phone of the merchant, a WeChat public account and a WeChat mini-program and the like.

[0123] The service package refers to a related car washing service package stored in the service end and capable of being associated with the real-time service image and being suitable for the real-time service image, and serves as a package for subsequent car washing service recommendation to the user, so that the service package can meet the car washing service demand of the user as much as possible. The target package is obtained by collecting a large number of different car washing packages selected by the real-time service image.

[0124] When the service package is selected, the package screening model trained in advance can be used for intelligent recognition and screening operation, so that the package screening model can accurately select the service package according to the real-time service image.

[0125] The training method of the package screening model is as follows:

[0126] A plurality of real-time service images and corresponding service packages are collected in advance, and the car washing frequency value, the package preference, the price sensitivity, the vehicle health degree, the service environment attribute, the type feature, the price feature and the preferential feature in the real-time service image are summarized and marked as a feature vector to obtain a plurality of feature vectors.

[0127] The service package is converted into a label corresponding to the feature vector, one feature vector corresponds to one label, a group of training data is formed, a plurality of training data constitutes a training set, and the labeled training features are divided into a training set and a test set.

[0128] Use 70% of the training features as the training set and 30% as the test set. Train the package selection model using the training set and test the package selection model using the test set. Set a preset error threshold. When the mean of the prediction error of all training features in the test set is less than the preset error threshold, the package selection model is obtained.

[0129] For example, the package selection model can be either a support vector machine model or a random forest model; the preset error threshold is set in advance according to the actual accuracy required by the package selection model.

[0130] After training the package selection model, the obtained real-time service profile can be input into the package selection model, thereby enabling the selection of service packages that match the real-time service profile from the server's database.

[0131] Since the number of selected service packages is not unique, and the degree of matching between each service package and the real-time service profile is different, it is necessary to calculate the degree of matching between each service package and the real-time service profile, and use the matching index to express the degree of matching numerically.

[0132] The adaptation index is calculated as follows:

[0133] The service items in the D service packages are separated one by one using word segmentation technology, and each service item is compared with the real-time service profile.

[0134] Service items that overlap with package preferences, price sensitivity, vehicle health, service environment attributes, type characteristics, price characteristics, and discount characteristics are recorded as valid items, and the number of valid items is counted to obtain D valid values;

[0135] After comparing the D valid values ​​with the total number of service items in the D service packages, D valid ratios are obtained.

[0136] The expression for the effective ratio is:

[0137] ;

[0138] In the formula, For the first The effective ratio of each service package =1,2,...,D, For the first The effective value of each service package For the first The total number of service items in each service package;

[0139] The package amount and the actual payment amount of the D service packages are queried one by one, and the package amount and the actual payment amount are compared after being subtracted, and D discount ratios are calculated;

[0140] The expression of the discount ratio is:

[0141] ;

[0142] In the formula, is the discount ratio of the i-th service package, is the package amount of the i-th service package, is the actual payment amount of the i-th service package; D effective ratios and D discount ratios are respectively assigned to corresponding ratio coefficients and added to obtain D adaptation indexes;

[0143] The expression of the adaptation index is:

[0144] ;

[0145]

[0146] In the formula, is the adaptation index of the i-th service package, , are respectively the ratio coefficients of the effective ratio and the discount ratio, and , , are all greater than 0.

[0147] After calculating the adaptation indexes of all the service packages, the adaptation degree between the service packages and the user's car washing service demand can be represented according to the size of the adaptation index, and used as a basis for subsequent recommendation of service packages to users;

[0148] Specifically, the greater the adaptation index, the higher the adaptation degree between the service package and the real-time service portrait, and the higher the correlation degree between the service package and the user's actual car washing service demand, and vice versa. Therefore, when recommending service packages to users, the D service packages are sequentially recommended to users in order according to the adaptation index from large to small, so as to ensure that the car washing service platform can recommend service packages to users in a matching and correlated degree from high to low, and help users accurately, quickly and reasonably select service packages that meet their own car washing service demand.

[0149] ​​​In the embodiment, by extracting comprehensive car owner information, the user's comprehensive car washing service information in the past time period can be collected in multiple dimensions, the data dimension of the car owner information for intelligent identification is expanded, by abnormally correcting the comprehensive car owner information, the abnormal data with missing phenomenon and repetition phenomenon in the comprehensive car owner information can be eliminated and supplemented, the integrity and uniqueness of the comprehensive car owner information are improved, and by constructing the user car washing portrait and real-time service portrait, the user's historical car washing service information and current individual demand service information can be fused, thereby realizing the double construction effect of the portrait of the user in the past time period and the current time period, avoiding the problem of limitation of the comprehensiveness of the car owner information in the construction of the portrait once, greatly improving the intelligent identification accuracy of the car owner information, and combining the intelligent screening processing of the package screening model, the service package matched with the car owner information and associated can be quickly and accurately screened out, ensuring that the screened service package can have high matching and association effect with the car owner information, and providing reasonable and accurate recommendation results for the subsequent car washing service of the user.

[0150] Embodiment two: please refer to Figure 2 As shown in the embodiment, some parts not described in detail are described in embodiment one, a car owner information intelligent identification method for car washing service is provided, applied to a car washing service platform, realized based on a car owner information intelligent identification system for car washing service, comprising:

[0151] S1: receiving the service request of the user in the user end, in the request state, querying the available data associated with the user from the database, and extracting the comprehensive car owner information from the available data;

[0152] S2: abnormally correcting the abnormal data in the comprehensive car owner information, extracting the user label from the abnormally corrected comprehensive car owner information, and fusing the user label into the user car washing portrait;

[0153] S3: determining the car washing service period, collecting the individual demand characteristics of the user end in the car washing service period, and fusing the individual demand characteristics with the user car washing portrait to construct the real-time service portrait with annotation;

[0154] S4: based on the completed package screening model, screening the service package matched with the real-time service portrait from the database of the service end, calculating the adaptation index of the service package, and recommending the service package to the user end in sequence.

[0155] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A car owner information intelligent identification system for car washing service, applied to a car washing service platform, characterized in that, The method comprises the following steps: a data query module is used to receive a service request of a user in a user terminal, query available data associated with the user from a database in a request state, and extract comprehensive car owner information from the available data; a first portrait fusion module is used to correct abnormal data in the comprehensive car owner information, extract a user tag from the corrected comprehensive car owner information, and fuse the user tag into a user car washing portrait; a second portrait construction module is used to determine a car washing service period, collect individual demand characteristics of the user terminal in the car washing service period, fuse the individual demand characteristics with the user car washing portrait, and construct a real-time service portrait with annotations; the individual demand characteristics include type characteristics, price characteristics, and preferential characteristics; the collection method of the type characteristics, the price characteristics, and the preferential characteristics is as follows: the natural language processing technology is used to identify the character groups and the number groups of B effective data in the car washing service period one by one; the effective data with the character groups of fine washing, general washing, and rough washing are recorded as type data, the effective data with the character groups of member and non-member are recorded as price data, and the effective data with the character groups of coupons, full reduction, and discount are recorded as preferential data; the reading and browsing time of the type data, the price data, and the preferential data is marked one by one through the time stamp, and the first duration, the second duration, and the third duration are obtained by adding all the reading and browsing times; the character groups and the number groups corresponding to the maximum value of the first duration, the maximum value of the second duration, and the maximum value of the third duration are combined to generate the type characteristics, the price characteristics, and the preferential characteristics; a package recommendation module is used to select a service package suitable for the real-time service portrait from the database of the server based on a trained package screening model, calculate the adaptation index of the service package, and recommend the service package to the user terminal in sequence; the training method of the package screening model is as follows: a plurality of real-time service portraits and corresponding service packages are collected in advance, the washing frequency value, the package preference, the price sensitivity, the vehicle health degree, the service environment attribute, the type characteristics, the price characteristics, and the preferential characteristics in the real-time service portrait are summarized and marked as a feature vector to obtain a plurality of feature vectors; the service package is converted into a label corresponding to the feature vector, one feature vector corresponds to one label, a group of training data is constructed, a plurality of training data constitutes a training set, the labeled training features are divided into a training set and a test set, the training set is used to train the package screening model, the test set is used to test the package screening model, a preset error threshold is set, when the average of the prediction errors of all training features in the test set is less than the preset error threshold, the package screening model is obtained. 2.The car owner information intelligent identification system for car washing service according to claim 1, characterized in that, the query method of the available data is as follows: the user surname and the user contact method corresponding to the original data in the database are queried one by one and recorded as an original surname and an original contact method; the user surname and the user contact method corresponding to the input service request are extracted and recorded as a request surname and a request contact method; when the original surname and the original contact method of the original data are consistent with the request surname and the request contact method respectively, the original data is recorded as available data. 3.The car owner information intelligent identification system for car washing service according to claim 2, characterized in that, The comprehensive car owner information includes service consumption data, vehicle state data and service environment data; The service consumption data includes car washing frequency, car washing time, service package, consumption amount and discount amount; The vehicle state data includes vehicle type and paint damage number; The service environment data includes weather type and environment temperature.

4. The car owner information intelligent recognition system for car washing service according to claim 3, characterized in that, The abnormal correction method of the abnormal data is: Query the record time of the A service consumption data one by one through the time stamp, and arrange the A service consumption data in order according to the time sequence; After the A vehicle state data and the A service environment data at the same record time of the A service consumption data are summarized respectively, A data packets are generated; According to the standard that there is no repetition and missing phenomenon in the service consumption data, vehicle state data and service environment data in the data packet, the A data packets are sequentially checked for abnormalities; The data packet with missing phenomenon and repetition is recorded as abnormal data, and the missing phenomenon and the repetition are recorded as missing data and repeated data respectively; The repeated data in the abnormal data is removed, the original data corresponding to the missing data is traced back from the database, and the traced back original data is imported into the missing position of the abnormal data to obtain the corrected comprehensive car owner information.

5. The car owner information intelligent recognition system for car washing service according to claim 4, characterized in that, The user label includes car washing frequency value, package preference, price sensitivity, vehicle health degree and service environment attribute; The fusion method of the user car washing portrait is: A basic portrait with three inner and outer wrapped distributions of annular contours is constructed, and the three annular contours are sequentially recorded as environment unit, vehicle unit and user unit from inside to outside; The service environment attribute is imported into the environment unit, the vehicle health degree is imported into the vehicle unit, and the car washing frequency value, package preference and price sensitivity are imported into the user unit; Data channels are established between the environment unit and the vehicle unit, and between the vehicle unit and the user unit, so as to convert the basic portrait into the user car washing portrait.

6. The car owner information intelligent recognition system for car washing service according to claim 5, characterized in that, The determination method of the car washing service cycle is: All the original data of the user end are queried through the database, the data attributes of the original data are identified one by one, and the original data with the data attribute of browsing and browsing are recorded as valid data; The record time of all the valid data is queried one by one through the time stamp, the last record time is recorded as the calibration time according to the time sequence, and the calibration time and the record time of the valid data are imported into the same time line; The first record time after the calibration time is recorded as the cycle starting point, the current time is recorded as the cycle ending point, and the period between the cycle starting point and the cycle ending point is recorded as the car washing service cycle.

7. The car owner information intelligent recognition system for car washing service according to claim 6, characterized in that, The construction method of the real-time service portrait is: Three feature boxes are set at equal angles on the user unit of the user car washing portrait, and the three feature boxes are sequentially numbered 1, 2 and 3 to generate No. 1 feature box, No. 2 feature box and No. 3 feature box; The type feature, price feature and discount feature are respectively imported into No. 1 feature box, No. 2 feature box and No. 3 feature box to generate feature interpretation; Two interpretation positions are respectively established between the first feature box and the second feature box and between the second feature box and the third feature box, and the user's surname and the user's contact information are respectively introduced into the two interpretation positions to generate identity interpretation and build a real-time service image with remark interpretation. 8.The car owner information intelligent identification system for car washing service of claim 7, wherein, The calculation method of the adaptation index is: The service items in the D service packages are split one by one through the word segmentation technology, and the service items are compared with the real-time service image one by one; The service items that are repeated with the package preference, price sensitivity, vehicle health degree, service environment attribute, type feature, price feature and preferential feature are recorded as effective items, and the number of effective items is counted to obtain D effective values; After comparing the D effective values with the total amount of service items in the D service packages, D effective ratios are obtained; The package amount and the actual payment amount of the D service packages are queried one by one, and the difference between the package amount and the actual payment amount is compared to calculate D preferential ratios; After the D effective ratios and the D preferential ratios are respectively assigned to the corresponding proportion coefficients and added, D adaptation indexes are obtained.

Images