Target matching method, device, equipment and storage medium based on scoring portrait

Abstract

The present invention relates to the field of computer computing, and proposes a matching method based on scoring portraits, which includes: obtaining user feature information from a preset database; inputting a preset user scoring model, marking users with user scores higher than a preset value as target users; generating target user scoring portraits and target user groups; obtaining feature information of service providers; inputting a preset service provider scoring model, and generating service provider scoring portraits with service provider scores; based on the target user's historical service provider list, screening target service providers through the service provider scoring portrait groups corresponding to the target user groups of historical service providers, and matching them with the target users. The present invention realizes scoring portraits of users and service providers, and by introducing historical service providers, realizes accurate and quantitative matching, thereby improving the success rate of matching. In addition, the present invention also relates to the field of blockchain technology, and reports scoring portraits and matching results to the blockchain to prevent tampering.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a target matching method, device, equipment and storage medium based on scoring portraits. Background Art

[0002] With the in-depth development of the information age, business data is generated at a massive scale. Both users and corporate salesmen have left a huge amount of information on the system. This information constitutes a group portrait based on user consumption habits and salesmen's sales style. However, when these portraits are obtained, they are often a general name definition, which cannot be quantified and cannot be accurately defined.

[0003] After obtaining the value evaluation of users and service providers, how to match users and service providers, what should be the most fundamental consideration, how to improve the matching degree and achieve the optimization of user and service provider experience are urgent issues to be solved. Summary of the invention

[0004] The present invention provides a target matching method, device, equipment and storage medium based on scoring portraits, the main purpose of which is to achieve a higher matching success rate and realize accurate matching between users and salesmen by generating quantifiable scoring portraits of users and service providers.

[0005] In order to solve the technical problem of the inventor,

[0006] The present invention also provides a target matching method based on scoring portraits, which is used for matching a service provider with a service recipient, and is characterized in that the method comprises:

[0007] Obtain user feature information from a preset database;

[0008] Inputting the user characteristic information into a preset user rating model to obtain a user rating;

[0009] Filter out users whose user ratings are higher than the preset value and mark them as target users;

[0010] Clustering and grouping the target users with user ratings according to the target user feature information to generate a user rating portrait;

[0011] Sorting the score rankings of various groups of the target user in the target user score profile, obtaining a group with a preset ranking in the target user group score ranking, and marking it as the target user group;

[0012] Obtaining service provider characteristic information from a preset database within a preset time;

[0013] Inputting the service provider characteristic information into a preset service provider scoring model to obtain a service provider score;

[0014] Clustering and grouping service providers with service provider ratings according to the service provider characteristic information to generate a service provider rating profile;

[0015] Obtain a list of historical service providers for the target user;

[0016] Substituting the historical service provider list into the service provider group corresponding to the service provider rating portrait and the target user group, obtaining the historical service provider with a preset ranking at the top of the service provider group, and marking it as the target service provider;

[0017] The target service provider is matched with the target user.

[0018] Preferably, the preset user scoring model is a scoring model based on deep learning, and the preset user scoring model includes a neural network model and a boosted gradient tree.

[0019] Preferably, clustering the target users includes preprocessing the target user feature information, and the method for preprocessing the target user feature information includes:

[0020] Perform word segmentation and cleaning on the target user feature information to generate a user information word list;

[0021] Capture keywords in the user information word list to generate user data tags;

[0022] According to each of the user data labels, the target users are grouped;

[0023] The user score corresponding to the user tag in each target user group

[0024] Preferably, clustering and grouping the service providers includes preprocessing the service provider characteristic information, and the method for preprocessing the service provider characteristic information includes:

[0025] Perform word segmentation and cleaning on the service provider feature information to obtain a service provider information word list;

[0026] Capture keywords in the service provider information word list to generate service provider data tags;

[0027] According to each of the service provider data tags, the service providers are grouped;

[0028] The service providers in each service provider group are marked with corresponding service provider scores.

[0029] Preferably, before inputting the service provider characteristic information into a preset service provider scoring model to obtain a service provider score, the method further comprises:

[0030] The service provider characteristic information is segmented and cleaned according to a preset time standard to obtain cleaned service provider data, wherein the preset time standard includes a preset time length extending forward from the current time as a base point;

[0031] The cleaned service data is subjected to maximum normalization processing to obtain preprocessed data. The mathematical expression of the maximum normalization processing is:

[0032]

[0033] Among them, x * is the preprocessed data, x is the data of each service provider within the preset time standard, and x max The maximum value of the product data of the service provider within the preset time standard;

[0034] Each of the pre-processed data is weighted according to the different weights of each information in the service party data tag to obtain the intermediate processed data. The mathematical expression of the weighted calculation is:

[0035]

[0036] Among them, P is the intermediate processing data, α i is the weight of the i-th preprocessed data, and b is the deviation;

[0037] The intermediate processed data is subjected to maximum value normalization processing, and the mathematical expression of the maximum value normalization processing of the intermediate processed data is:

[0038]

[0039] Among them, p * Rate the service provider; max is the maximum value of the intermediate processing data; P is the intermediate processing data.

[0040] Preferably, the method of substituting the service provider rating profile into the service provider group corresponding to the target user group according to the historical service provider list to obtain the target service provider further includes:

[0041] Determining the number of historical service providers of the single target user according to the list of historical service providers;

[0042] If the number of historical service providers is 0,

[0043] Acquire a service provider group corresponding to the target user group;

[0044] According to the service provider rating portrait, the service provider ranked first in the service provider group is marked as a target user;

[0045] If the number of historical service providers is 1,

[0046] then determining the historical service provider as the target service provider;

[0047] If the number of historical service providers is greater than 1,

[0048] Then, the user data label of the target user ranked first in the user rating portrait is obtained;

[0049] Obtaining a service provider data tag corresponding to the user data tag ranked first in the rating;

[0050] Obtain the service provider rating portrait group under the service provider data label;

[0051] The historical service provider ranked first in the service provider rating portrait grouping is determined as the target service provider.

[0052] Preferably, it also includes uploading the scoring portrait of each target user, the scoring portrait of each service provider, and the matching result between the target user and the target service provider to the blockchain, so that the blockchain encrypts and stores the target user scoring portrait, the service provider scoring portrait, and the matching result between the target user and the target service provider.

[0053] To achieve the above-mentioned purpose, the present invention also provides a target matching device based on scoring portrait, which is applied to a background server. The matching device based on scoring portrait includes: a user feature collection module, which is used to obtain user feature information from a preset database; a user scoring module, which is used to input the user feature information into a preset user scoring model to obtain a user score; a target user screening module, which is used to screen users whose user scores are higher than a preset value and mark them as target users; a user scoring portrait module, which is used to cluster and group the target users with user scores according to the target user feature information to generate a user scoring portrait; a service provider information collection module, which is used to obtain information about the service provider in a preset database at a preset time. The target user receives the service provider's characteristic information within the specified time period; a service provider scoring module is used to input the service provider's characteristic information into a preset service provider scoring model to obtain a service provider score; a service provider scoring portrait module is used to cluster and group service providers with service provider scores according to the service provider's characteristic information to generate a service provider scoring portrait; a historical service provider screening module is used to obtain a historical service provider list of the target user; the historical service provider list is substituted into the service provider group corresponding to the service provider scoring portrait and the target user group, and a historical service provider with a preset ranking in the service provider group is obtained, and is marked as a target service provider; a matching module is used to match the target user with the target service provider.

[0054] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer device, including: a memory and a processor, wherein the processor stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes any step of the above-mentioned target matching method based on scoring portraits.

[0055] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer storage medium, and when the computer-readable instructions are executed by one or more processors, one or more of the processors execute any step in the target matching method based on scoring portraits.

[0056] The present invention proposes a target matching method, system, device and storage medium based on scoring portraits, which use user information to screen target users through a user scoring model, generate a user scoring portrait based on the target user's score, and use the salesperson information to obtain the service provider's score through a salesperson scoring model to generate a service provider scoring portrait, thereby solving the problem of quantifying the value between the user and the service provider. By introducing the dimension of the target user's historical service to the service provider, a familiar connection is established between the user and the service provider. At the same time, based on the correspondence between the quantified user and service provider portraits, accurate matching is achieved, the matching success rate is improved, and the business success rate is significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1A schematic diagram of the steps of a preferred embodiment of the target matching method based on scoring portraits of the present invention;

[0058] Figure 2 It is a schematic diagram of the structure of the target matching device based on the scoring portrait of the present invention;

[0059] Figure 3 It is a schematic diagram of the structure of the target matching device based on the scoring portrait of the present invention.

[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0062] A target matching method based on scoring profiles, comprising:

[0063] Please refer to Figure 1 , including the following steps:

[0064] S101: Acquire user feature information from a preset database;

[0065] In the daily business activities of enterprises, each service provider will import various user information into the system when serving users and customizing products. As time goes by, there will be a huge amount of information accumulated between enterprise users and service providers. Enterprises will import a huge amount of original information into the big data platform. Because the HIVE big data platform has advantages such as distribution, easy operation, compatibility with multiple languages, and fast response speed, it is very suitable for data mining. This case obtains the characteristic information of users who meet the business conditions from the big data platform HIVE. In this case, users who meet the business conditions mainly refer to users who have lost their insurance or are about to lose their insurance. One of the most important characteristics of the insurance business is that it has a warranty period, and different insurances have different warranty periods. Since the warranty period is long or short, there will be a situation where users forget the warranty period, resulting in a large number of users who lose their insurance or are about to lose their insurance. At this time, either the service provider remembers to develop users again, or the business is snatched away by competitors. In any case, it is a disaster for the performance of enterprises and service providers. It is very important to have a unified system that can filter out users who have lost their insurance or are about to lose their insurance from big data. The screening method should include comparing the date in the user's insurance policy information with the real-time date to obtain a time difference. By setting a preset time difference value, such as "less than one month", we can filter out all specific users who have lost their insurance and are less than one month away from the expiration of the insurance period. While obtaining specific users, we also obtain a lot of information related to these users' insurance policies.

[0066] S102: Inputting the user characteristic information into a preset user rating model to obtain a user rating;

[0067] In this case, the user characteristic information is divided into first information and second information.

[0068] The characteristic information of these users, such as insurance type, premium, rate and cost rate, is classified as first information. There is at least one first information and there can be multiple first information.

[0069] The user's characteristic information, such as the user's own registered province and city, enterprise type and recently uninsured insurance type, etc., is classified as the second information. Similarly, there is at least one second information, and there can be multiple.

[0070] By inputting the first information into a preset user scoring model, the score of each specific user is calculated.

[0071] Among them, the preset user rating model is based on the machine learning model. Among the many machine learning models such as support vector machines, regression trees, data clustering / dimensionality reduction, linear regression, integrated models, and naive Bayes, it can be a single model or an integration of multiple models. Finally, it is modified and adjusted through feedback to ensure that the user scores obtained by the model in practice are closer and closer to the actual situation. The feedback factors for the model should include but are not limited to the user's order conversion rate, the inspection of the competent department, and the user's rating by the service provider.

[0072] S103: Screening out users whose user scores are higher than a preset value and marking them as target users;

[0073] When the score of a specific user exceeds the preset value, they are selected and defined as target users. Target users are the focus of service providers and corporate performance and revenue.

[0074] S104: Clustering and grouping the target users with user ratings according to the target user feature information to generate a target user rating profile;

[0075] Taking the second information as the grouping basis, many groups are obtained. For example, by enterprise type, individual industrial and commercial households are divided into one group, small-scale enterprises are divided into one group, enterprises above designated size are divided into one group, foreign-controlled enterprises are divided into one group, and state-owned enterprises are divided into one group, thus obtaining many groups. Also, by the province of registration, Guangdong Province, Zhejiang Province, and Fujian Province are independently divided into groups, thus obtaining interrelated and independent groups.

[0076] User portrait is derived from the concept of user model proposed when studying the systematic method of user. User model refers to the virtual representation of real users. It is a virtual user derived from a deep understanding of real data. Through research, we understand users and divide them into different types according to their goals, behaviors and opinions. Then, typical features are extracted from each type and given descriptions such as name, photo, demographic elements and scene to form a user model. User portrait is to model and abstract the attribute label system of each user from massive user data, and produce a clear user attribute system through cleaning, mining, sorting and production. These attributes such as demographic attributes, behavior trajectories, user groups, life scenes and consumption preferences are usually called labels.

[0077] Exemplarily, in this application, by mining the second information in the insurance policy, labels with different attributes such as Guangdong Province and foreign-funded can be generated. Each label has different basic attributes. For example, users in Guangdong Province have fewer clauses on traffic caused by snowy and slippery roads than insurance in the north because there is no snow or ice in winter. Another example is the label of foreign-funded companies, which will cause more staff to take international flights and have more demand for flight-related insurance, thus forming a target user portrait composed of many basic attributes under different labels.

[0078] S105: sorting the score rankings of various groups of the target user in the target user score portrait, obtaining a group with a preset ranking in the target user group score ranking, and marking it as the target user group;

[0079] Exemplarily, assuming that the second information a is 3, including E, F, and G, then E can be divided into 7 groups, F can be divided into 2 groups, and G can be divided into 4 groups, then a total of (7+2+4=) 13 groups can be obtained, which is the sum of the second information element set, and then the score of the target user is put into each group in turn. Taking the above as an example, the target users are assigned to 13 groups according to the user's information. Each target user may be in one group or in multiple groups, that is, the numerous policy information of users under multiple labels, combined with the score of the first user information, obtains the target user score portrait.

[0080] S106: Obtaining service provider characteristic information of the service provider within a preset time in a preset database;

[0081] Similarly, in this case, the company's massive user and service provider data are all on the HIVE big data platform. We screened out specific service providers and their information on the big data platform. Among them, specific service providers refer to active and stable service providers. Service providers are not affiliated with the company, but more of a cooperative contractual relationship. The service provider's employment is not certain, and there will be personnel mobility issues. Unlike user information, once a user places an order, the information is accumulated within the company. If the service provider moves, it may go to a competitor.

[0082] The information of a specific service provider refers to the information within the last three years, which not only maintains the accuracy of performance information, but also ensures that more stable service providers can be screened out. Information such as the service provider's historical insurance categories, the city of the institution, the user's registered capital range, the user's corporate type and the user's industry in the specific service provider information is classified as the fourth information, and information such as the service provider's premium and number of users in the specific service provider information is classified as the third information. Similarly, there must be at least one of the third information and the fourth information, and there can be more than one.

[0083] S107: Inputting the service provider characteristic information into a preset service provider scoring model to obtain a service provider score;

[0084] Based on the fourth information, specific service providers are first grouped to obtain different groups such as different regions and different types of insurance. Based on each group, the fourth information of each specific service provider is input into the service provider scoring algorithm. In this case, the fourth information input into the service provider scoring algorithm is the information of the service provider in the most recent year, which further limits the timeliness of the service provider data, and obtains the score of each specific service provider in different groups to obtain a service provider scoring portrait.

[0085] Exemplarily, the fourth information includes region and insurance type. The region is divided into 50 cities including Shenzhen, Guangzhou, Beijing, Shanghai, Chongqing, etc., and the insurance type is divided into 10 insurance types including auto insurance, life insurance, accident insurance, flight delay insurance, etc. Assuming that service party A works in Guangzhou and its business involves three insurance types including auto insurance and life insurance, and service party B works in Shenzhen and its business involves seven insurance types including life insurance and accident insurance, through grouping, service party A is in the group of Guangzhou and the group of three insurance types including auto insurance and life insurance, while service party B is in Shenzhen and the group of seven insurance types including life insurance and accident insurance.

[0086] The service provider scoring algorithm in this case is to perform maximum normalization processing on the information of the latest year in the third information, which refers to the number of insurance policies and users of the service provider in the latest year, to obtain pre-processed data.

[0087] Its mathematical expression is:

[0088]

[0089] Among them, x * is the maximum normalized value, x is the policy amount or number of users of the service provider in the past year, and x max The maximum policy amount or the maximum number of users of the service provider in the most recent year;

[0090] That is, the preprocessed data will obtain two values, namely, the normalized value of the number of users in the past year and the normalized value of the premium in the past year. The maximum value normalization is a quantitative processing that divides the number of all policies of each specific service provider in the past year by the maximum policy amount and the maximum number of users.

[0091] For example, taking the policy amount as an example, if the maximum policy amount of service party A in the most recent year is 50,000, and only 4 policies were issued, which are 10,000, 3,000, 50,000 and 30,000 respectively, then through maximum value normalization, the maximum normalized value of service party A's policy is 0.2, 0.06, 1 and 0.6; the maximum number of users of service party A in the most recent year is 5, and there have been three changes, which are 1, 2, 4 and 5 respectively, then through maximum value normalization, the maximum normalized value of the number of users of service party A is 0.2, 0.4, 0.8 and 1.

[0092] Then the pre-processed data is weighted according to the weights of different information in the fourth information to obtain the intermediate processed data. Here, the two maximum normalizations will be summed with different weights to become an index, that is, the weight normalization value.

[0093] Its mathematical expression is:

[0094]

[0095] Among them, P is the intermediate processing data, α i is the weight of the i-th preprocessed data, and b is the deviation

[0096] For example, in a certain embodiment, the service provider characteristic information has only two parameters, and the specific mathematical expression is:

[0097] P=0.7x1+0.3x2

[0098] Among them, P is the intermediate processing data, x1 is the normalized value of the number of users in the past year, and x2 is the normalized value of the premium in the past year.

[0099] Exemplarily, x1 of service provider A is 0.2, 0.06, 1 and 0.6, and x2 is 0.6, 0.4 and 1; then P is 0.02, 0.162, 0.94 and 0.72.

[0100] Finally, the intermediate processed data is normalized, and its mathematical expression is:

[0101]

[0102] Among them, p * is the weight normalization value; p max is the maximum value of the intermediate processing data; P is the intermediate processing data.

[0103] Similarly, according to the example, the weighted normalized value p of salesperson A in the last year is * It can be calculated as 0.021, 0.172, 1 and 0.766.

[0104] S108: Clustering and grouping the service providers with service provider ratings according to the service provider characteristic information to generate a service provider rating portrait;

[0105] Sum and sort within the group to get the score range of the service provider within the group, that is, the score of service provider A in the Shenzhen group is 1.959. Similarly, the scores of all service providers such as service providers B and C in the past year can be obtained, and sorted within each group to construct a service provider score profile.

[0106] In this case, the information of specific users and specific service providers in the context of big data was quantified by algorithms to obtain scoring portraits and rankings. Here, each algorithm has its own particularity, and will iteratively evolve in practice to produce an algorithm that is more in line with the actual situation. There is also a large amount of other information about users and service providers on the big data platform. In the practice of the model, some binary information will be introduced or deleted, striving to achieve the best simulation of the model to reality.

[0107] S109: Obtain a list of historical service providers for the target user;

[0108] Based on the rating profile of the target user, that is, the target user's insurance policy information, there are service providers who have provided services to the target user, that is, the insurance agents in the insurance policy. By crawling, we can obtain all the insurance agents of the target user in history, that is, the historical service providers.

[0109] In this case and in reality, the user's closest contact with the old partners should be the user's historical service providers. However, the number of historical service providers for the selected target users is uncertain. Generally speaking, there are three situations, namely 0, 1, and greater than 1. There are different matching methods for these three situations. The following will introduce the three matching methods.

[0110] S110: Substituting the historical service provider list into the service provider group corresponding to the service provider rating portrait and the target user group, obtaining a historical service provider with a preset ranking at the top of the service provider group, and marking it as a target service provider;

[0111] When the number of historical service providers is 0, determine all ranking orders of the target user in each of the second information groups, perform weighted addition of all ranking orders of each target user in each of the second information groups to obtain the target user ranking, and correspond the target user ranking to the scoring interval of the historical service provider in the third information group to obtain the target service provider;

[0112] When the number of historical service providers is 1, the historical service provider is the target service provider;

[0113] When the number of historical service providers is greater than 1, each historical service provider will be ranked according to recent performance and insurance type priority, and the target service provider will be the service provider with the highest ranking.

[0114] The optimal matching method takes the target user's historical service as the most important consideration, and the matching method can also be fine-tuned with other considerations.

[0115] S111: Matching the target service provider with the target user;

[0116] For example, if the target user Xiaohong is in Guangzhou, the insurance type that is cancelled is auto insurance, and the historical service provider has resigned, that is, the number of historical service providers is 0, then the first method is used for matching, and the ranking obtained by adding the ranking of Xiaohong in Guangzhou and the auto insurance group is matched with the service providers in the same interval in the same group to obtain the target service provider, and then the target service provider is matched with the target user;

[0117] For example, if the target user Xiaobai has only one historical service provider, then this service provider is Xiaobai's target matching service provider;

[0118] For example, if the target user Xiao Ming has 5 historical service providers and Xiao Ming is in Shenzhen and the insurance type that he has lost is auto insurance, that is, the number of historical service providers is greater than 1, and the third method is used to match Xiao Ming with the best service provider, Xiao Ming will be assigned to two service provider groups, namely Shenzhen and auto insurance service provider groups. The top-ranked service providers in these two groups will be ranked according to their recent performance. If the performance of two people is the same, they will be ranked according to which service provider has a higher auto insurance priority, so that the target service provider is matched with the target user Xiao Ming.

[0119] In an optional implementation, it is also possible to upload the target user rating portrait, the service provider rating portrait, and the matching result between the target user and the target service provider to the blockchain, so that the blockchain encrypts and stores the target user rating portrait, the service provider rating portrait, and the matching result between the target user and the target service provider.

[0120] The matching result between the target user and the target service provider indicates the contact relationship between the target user and the target service provider and the order result achieved based on the matching result. The accuracy of the target matching method based on the scoring portrait is judged through the matching result. Due to the tamper-proof nature of the blockchain, a more real and objective evaluation of the accuracy of the target matching method based on the scoring portrait is achieved.

[0121] Specifically, the summary information corresponding to the target user's rating portrait and the service provider's rating portrait is obtained by, for example, processing using the sha256s algorithm. Uploading the summary information to the blockchain can ensure its security and fairness and transparency to users. Users can download the summary information from the blockchain to verify whether the user's and the service provider's rating portraits and content have been tampered with. The blockchain referred to in this example is a new application model of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. The blockchain can include the blockchain underlying platform, the platform product service layer, and the application service layer.

[0122] In this embodiment, a target matching device based on a scoring portrait is provided, and the target matching device based on a scoring portrait corresponds one-to-one to the target matching method based on a scoring portrait in the above embodiment. Figure 2 As shown, the target matching device based on scoring portrait includes a user feature collection module 91, a user scoring module 92, a target user screening module 93, a user scoring portrait module 94, a target user grouping module 95, a service provider information collection module 96, a service provider scoring module 97, a service provider scoring portrait module 98, a historical service provider screening module 99, a target service provider acquisition module 100 and a matching module 101.

[0123] The detailed description of each functional module is as follows:

[0124] The user feature collection module 91 is used to obtain user feature information from a preset database.

[0125] The user scoring module 92 is used to input user feature information into a preset user scoring model to obtain a user score.

[0126] The target user screening module 93 is used to screen users whose user scores are higher than a preset value and mark them as target users.

[0127] The user rating portrait module 94 is used to cluster and group target users with user ratings according to target user feature information to generate user rating portraits.

[0128] The target user grouping module 95 is used to sort the target user's various group score rankings in the target user's score profile, obtain a group with a preset ranking in the target user group score ranking, and mark it as the target user group.

[0129] The service provider information collection module 96 is used to obtain the service provider characteristic information of the service provider in a preset database within a preset time.

[0130] The service provider scoring module 97 is used to input the service provider characteristic information into a preset service provider scoring model to obtain a service provider score.

[0131] The service provider rating portrait module 98 is used to cluster and group the service providers with service provider ratings according to the service provider feature information to generate a service provider rating portrait.

[0132] The historical service provider screening module 99 is used to obtain the target user's historical service provider list.

[0133] The target service provider acquisition module 100 is used to substitute the historical service provider list into the service provider group corresponding to the service provider rating portrait and the target user group, obtain the historical service provider with a preset ranking in the service provider group, and mark it as the target service provider.

[0134] Matching module 101: used to match a target user with a target service provider.

[0135] Specifically, the user rating portrait module 94 includes a target user feature cleaning unit, a user data label generating unit, a target user grouping unit, and a target user rating portrait generating unit.

[0136] The target user feature cleaning unit is used to perform word segmentation cleaning on the target user feature information to generate a user information word list.

[0137] The user data tag generating unit is used to capture keywords in the user information word list and generate user data tags.

[0138] The target user grouping unit is used to group the target users according to each of the user data tags.

[0139] The target user rating portrait generating unit is used to generate a target user rating portrait based on the user rating corresponding to the user tag in each target user group.

[0140] Specifically, the service provider rating portrait model 98 includes a service provider feature cleaning unit, a service provider data label generating unit, a service provider grouping unit, and a service provider rating portrait generating unit.

[0141] The service provider feature cleaning unit is used to perform word segmentation and cleaning on the service provider feature information to obtain a service provider information word list.

[0142] The service provider data label generating unit is used to capture the keywords in the service provider information word list and generate the service provider data label.

[0143] The service provider grouping unit is used to group the service providers according to each of the service provider data tags.

[0144] The service provider rating portrait generating unit is used to generate a service provider rating portrait by scoring the service provider corresponding to the service provider mark in each of the service provider groups.

[0145] Specifically, the target service provider acquisition module 100 includes a historical service provider quantity judgment unit and a target service provider determination unit.

[0146] The historical service provider quantity determination unit is used to determine the number of service providers on the historical service provider list. There are three target service provider determination units according to the historical service provider quantity being 0, 1, and greater than 1, which respectively correspond to the first target service provider determination unit when the historical service provider quantity is 0, the second target service provider determination unit when the historical service provider quantity is 1, and the third target service provider determination unit when the historical service provider quantity is greater than 1.

[0147] The first target service provider determination unit is configured to obtain a service provider group corresponding to the target user group, and mark a service provider ranked first in the service provider group as a target user according to the service provider rating portrait.

[0148] The second target service provider determining unit is used to determine the historical service provider as the target service provider.

[0149] The third target service provider determination unit is used to obtain the user data tag of the target user that ranks first in the user rating portrait; obtain the service provider data tag corresponding to the user data tag that ranks first in the rating; obtain the service provider rating portrait group under the service provider data tag; and determine the historical service provider that ranks first in the service provider rating portrait group as the target service provider.

[0150] For the specific definition of the target matching device based on the scoring portrait, please refer to the definition of the target matching method based on the scoring portrait above, which will not be repeated here. Each module in the above-mentioned target matching device based on the scoring portrait can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0151] Please refer to Figure 3 , the present application also provides a computer device, including:

[0152] A memory and a processor, wherein computer-readable instructions are stored in the processor, and when the computer-readable instructions are executed by the processor, the processor executes any step of the above-mentioned target matching method based on scoring portraits.

[0153] The present application also provides a computer-readable storage medium. When the computer-readable instructions are executed by one or more processors, the one or more processors execute any step in the target matching method based on the scoring portrait. It can be understood that the readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0154] In this case, the huge amount of user and service provider data that has been retained in the enterprise in the past was screened and quantified using different specific algorithms. The scores of users and service providers in different groups were obtained to construct target user score portraits and service provider score portraits, which achieved the method of reducing the dimension of data and quantifying it into a single indicator. Then, based on whether there is a user's historical service provider, three different matching methods of target users and service providers were generated. The score portrait obtained by mining the value of users and service providers through big data effectively defined the value of users and service providers. Through the matching consideration of the dimension of historical service providers, users and service providers are more familiar with each other, which improves users' comfort, reduces the hard matching feeling of cold data, improves the success rate of insurance policies, and better serves users. Through the scoring of service providers, it even achieves the quantification of the assessment of service provider capabilities.

[0155] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, device, article or method including the element.

[0156] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. Through the description of the above implementation modes, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware, but in many cases the former is a better implementation mode. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0157] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A target matching method based on scoring portrait, characterized in that: The method comprises: Obtain user feature information from a preset database; Inputting the user characteristic information into a preset user rating model to obtain a user rating; Filter out users whose user ratings are higher than the preset value and mark them as target users; Clustering and grouping the target users with user ratings according to the characteristic information of the target users to generate a target user rating portrait; clustering and grouping the target users with user ratings according to the characteristic information of the target users to generate a target user rating portrait includes: performing word segmentation and cleaning on the characteristic information of the target users to generate a user information word list; capturing keywords in the user information word list to generate user data tags; grouping the target users according to each of the user data tags; generating a target user rating portrait by performing the user rating corresponding to the user tag in each of the target user groups; Sorting the various group score rankings of the target user in the target user score profile, obtaining a group of the target user with a preset ranking in the group score ranking, and marking it as the target user group; Obtaining service provider characteristic information from a preset database within a preset time; Inputting the service provider characteristic information into a preset service provider scoring model to obtain a service provider score; Clustering and grouping service providers with service provider ratings according to the service provider characteristic information to generate a service provider rating profile; Obtain a list of historical service providers for target users in the target user group; Substituting the historical service provider list into the service provider group corresponding to the service provider rating portrait and the target user group, obtaining the historical service provider with a preset ranking at the top of the service provider group, and marking it as the target service provider; The target service provider is matched with the target user.

2. The target matching method based on scoring portrait according to claim 1, characterized in that: The preset user scoring model is a scoring model based on deep learning, and the preset user scoring model includes a neural network model or a boosted gradient tree.

3. The target matching method based on scoring portrait according to claim 1 is characterized in that: The clustering and grouping of service providers with service provider ratings according to the service provider characteristic information to generate a service provider rating portrait comprises: Perform word segmentation and cleaning on the service provider feature information to obtain a service provider information word list; Capture keywords in the service provider information word list to generate service provider data tags; According to each of the service provider data tags, the service providers are grouped; The service provider score corresponding to the service provider mark in each of the service provider groups is used to generate a service provider score portrait.

4. The target matching method based on scoring portrait according to claim 3 is characterized in that: Before inputting the service provider characteristic information into a preset service provider scoring model to obtain a service provider score, the method further includes: The service provider characteristic information is segmented and cleaned according to a preset time standard to obtain cleaned service provider data, wherein the preset time standard includes a preset time length extending forward from the current time as a base point; The cleaned service data is subjected to maximum normalization processing to obtain preprocessed data. The mathematical expression of the maximum normalization processing is: in, To preprocess the data, Provides data for each service provider within the preset time standard. The maximum value of the product data of the service provider within the preset time standard; Each of the pre-processed data is weighted according to the different weights of each information in the service party data tag to obtain the intermediate processed data. The mathematical expression of the weighted calculation is: in, For intermediate processing of data, is the weight of the i-th preprocessed data, and b is the deviation; The intermediate processed data is subjected to maximum value normalization processing, and the mathematical expression of the maximum value normalization processing of the intermediate processed data is: in, Rate the service provider; is the maximum value of the intermediate processing data; p is the intermediate processing data.

5. The target matching method based on scoring portrait according to claim 4 is characterized in that: Substituting the service provider rating profile into the service provider group corresponding to the target user group according to the historical service provider list, before obtaining the target service provider, the method further includes: Determine the number of historical service providers for a single target user according to the list of historical service providers; If the number of historical service providers is 0, obtaining a service provider group corresponding to the target user group; According to the service provider rating portrait, the service provider ranked first in the service provider group is marked as a target service provider; If the number of the historical service providers is 1, the historical service provider is determined to be the target service provider; If the number of the historical service providers is greater than 1, obtaining the user data label of the target user ranked first in the user rating portrait; Obtaining a service provider data tag corresponding to the user data tag ranked first in the rating; Obtain the service provider rating portrait group under the service provider data label; The historical service provider ranked first in the service provider rating portrait grouping is determined as the target service provider.

6. The target matching method based on scoring portrait according to claim 1, characterized in that: Also includes: The scoring portrait of each of the target users, the scoring portrait of each of the service providers, and the matching results between the target users and the target service providers are uploaded to the blockchain, so that the blockchain encrypts and stores the scoring portrait of the target users, the scoring portrait of the service providers, and the matching results between the target users and the target service providers.

7. A target matching device based on scoring portrait, characterized in that: Applied to backend servers, including: A user feature collection module, used to obtain user feature information from a preset database; A user rating module, used to input the user feature information into a preset user rating model to obtain a user rating; The target user screening module is used to screen users whose user scores are higher than a preset value and mark them as target users; The user rating portrait module is used to cluster and group the target users with user ratings according to the feature information of the target users to generate user rating portraits; the user rating portrait module includes a target user feature cleaning unit, a user data label generating unit, a target user grouping unit, and a target user rating portrait generating unit; the target user feature cleaning unit is used to perform word segmentation and cleaning on the feature information of the target user to generate a user information word list; the user data label generating unit is used to capture keywords in the user information word list to generate user data labels; the target user grouping unit is used to group the target users according to each user data label; the target user rating portrait generating unit is used to generate a target user rating portrait for the user score corresponding to the user label in each target user group; A target user grouping module is used to sort the various group score rankings of the target user in the target user score portrait, obtain a group of the target user with a preset ranking in the group score ranking, and mark it as the target user group; A service provider information collection module is used to obtain the service provider characteristic information of the service provider in a preset database within a preset time; A service provider scoring module, used to input the service provider characteristic information into a preset service provider scoring model to obtain a service provider score; A service provider rating portrait module is used to cluster and group service providers with service provider ratings according to the service provider feature information to generate a service provider rating portrait; A historical service provider screening module, used to obtain a list of historical service providers for target users in the target user group; A target service provider acquisition module is used to substitute the historical service provider list into the service provider group corresponding to the service provider rating portrait and the target user group, obtain the historical service provider with a preset ranking at the top of the service provider group, and mark it as the target service provider; A matching module is used to match the target user with the target service provider.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the target matching method based on scoring portraits as described in any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a target matching program based on a scoring portrait. When the matching program based on a scoring portrait is executed by a processor, the steps of the target matching method based on a scoring portrait as described in any one of claims 1 to 6 are implemented.

Images