User information pushing method and device, electronic equipment and computer readable medium

Abstract

The embodiment of the invention discloses a user information pushing method and device, electronic equipment and a computer readable medium. A specific embodiment of the method comprises the steps of extracting user feature data of a target user according to historical mailing data of the target user; according to the historical business data of the plurality of candidate operation objects, extracting object feature data of the plurality of candidate operation objects; based on the matching degree between the user feature data and the object feature data, selecting a target operation object from the plurality of candidate operation objects; and pushing the user information of the target user to a terminal of the target operation object, so that the target operation object serves the target user. The implementation mode is related to a big data analysis technology, and the matching degree of the feature data between the potential users and the sales personnel is extracted and analyzed, so that the users can be allocated to proper sales for service. In this way, the success probability of user signing can be improved, and the number of new users using logistics services is increased.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of big data analysis technology, and more particularly to the field of logistics market development technology. Specifically, the present disclosure relates to a method, apparatus, electronic device, and computer-readable medium for pushing user information. Background Art

[0002] In the operation of logistics B-side (Business, usually representing enterprise user merchants) customers, in order to expand the market and increase the operating income of logistics, front-line operators (such as merchant sales, commercial sales, station masters, etc.) usually conduct new signing visits to potential merchants within the scope of their responsibilities to promote their signing.

[0003] However, the inventors have found that the current new signing expansion of the logistics system highly depends on offline sales and station masters, and relies on the discovery of offline merchant leads. This method often results in a limited number of newly signed users.

[0004] The above information disclosed in this background art section is only used to enhance the understanding of the background of the inventive concept. Therefore, it may include information that does not form the prior art known to ordinary skilled artisans in the relevant field in this country. Summary of the Invention

[0005] The content part of the present disclosure is used to briefly introduce concepts that will be described in detail in the subsequent detailed implementation part. The content part of the present disclosure is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0006] Some embodiments of the present disclosure propose a method, apparatus, electronic device, computer-readable medium, and computer program product for pushing user information to solve one or more of the technical problems mentioned in the above background art section.

[0007] In a first aspect, some embodiments of the present disclosure provide a method for pushing user information, including: extracting user feature data of a target user based on the historical mailing data of the target user, where the user feature data includes at least one of the following: character-type feature data, numerical-type feature data, and time-series feature data; extracting object feature data of a plurality of candidate operation objects based on the historical business data of the plurality of candidate operation objects; selecting a target operation object from the plurality of candidate operation objects based on the matching degree between the user feature data and the object feature data; and pushing the user information of the target user to the terminal of the target operation object so that the target operation object serves the target user.

[0008] In some embodiments, according to the historical mailing data of the target user, user feature data of the target user is extracted, including: determining character-type data with a string length greater than a specified value as long string-type data; processing the long string-type data using a Word2vec word vector model to obtain a low-dimensional first character-type feature vector; and determining character-type data with a string length not greater than the specified value as short string-type data; processing the short string-type data using an embedding technique to obtain a low-dimensional second character-type feature vector.

[0009] In some embodiments, according to the historical mailing data of the target user, user feature data of the target user is further extracted, including: concatenating the first character-type feature vector and the second character-type feature vector to obtain character-type feature data of the target user.

[0010] In some embodiments, according to the historical mailing data of the target user, user feature data of the target user is further extracted, including: normalizing the attribute values of the same attribute for numerical data representing size relationships; performing key information extraction based on the normalized numerical data to obtain numerical feature data of the target user.

[0011] In some embodiments, according to the historical mailing data of the target user, user feature data of the target user is further extracted, including: performing key feature extraction on time series data to obtain time series feature data of the target user, where the time series feature data includes feature data in a first time period and feature data in a second time period, and the duration of the first time period is greater than the duration of the second time period.

[0012] In some embodiments, the method further includes: respectively extracting key feature data from the character-type feature data, numerical feature data, and time series feature data of the user feature data; fusing the key feature data of the three types of feature data for determining the matching degree.

[0013] In some embodiments, the method further includes: inputting the user feature data and object feature data into a pre-trained matching model to obtain the matching degree between the target user and the operation object, where the matching model is used to analyze the similarity between the input feature data.

[0014] In some embodiments, the training samples of the matching model are obtained by the following method: performing normalization processing on each type of feature data in the acquired user feature data to obtain sample user feature data, where the dimension of each type of feature data in the sample user feature data is at least the product of the sample quantity and the quantity of that type of feature data; performing normalization processing on each type of feature data in the acquired object feature data to obtain sample object feature data; and using the sample user feature data and the sample object feature data as training samples.

[0015] In a second aspect, a user information pushing device includes: a user feature extraction unit configured to extract user feature data of a target user according to the historical mailing data of the target user, where the user feature data includes at least one of the following: character type feature data, numerical type feature data, and time series feature data; an object feature extraction unit configured to extract object feature data of a plurality of candidate operation objects according to the historical service data of the plurality of candidate operation objects; a selection unit configured to select a target operation object from the plurality of candidate operation objects based on the matching degree between the user feature data and the object feature data; and a pushing unit configured to push the user information of the target user to the terminal of the target operation object so that the target operation object serves the target user.

[0016] In some embodiments, the user feature extraction unit is further configured to determine character type data with a string length greater than a specified value as long string type data; process the long string type data using a Word2vec word vector model to obtain a low-dimensional first character type feature vector; and determine character type data with a string length not greater than the specified value as short string type data; process the short string type data using an embedding technique to obtain a low-dimensional second character type feature vector.

[0017] In some embodiments, the user feature extraction unit is further configured to splice the first character type feature vector and the second character type feature vector to obtain the character type feature data of the target user.

[0018] In some embodiments, the user feature extraction unit is further configured to perform normalization processing on the attribute values of the same attribute for numerical type data representing size relationships; and perform key information extraction based on the normalized numerical type data to obtain the numerical type feature data of the target user.

[0019] In some embodiments, the user feature extraction unit is further configured to perform key feature extraction on time series data to obtain the time series feature data of the target user, where the time series feature data includes feature data in a first time period and feature data in a second time period, and the duration of the first time period is greater than the duration of the second time period.

[0020] In some embodiments, the device further includes a feature fusion unit configured to respectively extract key feature data from the character-based feature data, numerical feature data, and time series feature data of the user feature data, and fuse the key feature data of the three types of feature data for determining the matching degree.

[0021] In some embodiments, the device further includes a matching degree determination unit configured to input the user feature data and the object feature data into a pre-trained matching model to obtain the matching degree between the target user and the operation object, where the matching model is used to analyze the similarity between the input feature data.

[0022] In some embodiments, the device further includes a training sample generation unit configured to perform normalization processing on each type of feature data in the acquired user feature data to obtain sample user feature data, where the dimension of each type of feature data in the sample user feature data is at least the sample number multiplied by the number of this type of feature data; perform normalization processing on each type of feature data in the acquired object feature data to obtain sample object feature data; and use the sample user feature data and the sample object feature data as training samples.

[0023] In a third aspect, some embodiments of the present disclosure provide an electronic device, including: one or more processors; a storage device storing one or more programs thereon, and when the one or more programs are executed by the one or more processors, the one or more processors implement the user information push method described in any implementation manner of the first aspect above.

[0024] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium storing a computer program thereon, where the computer program implements the user information push method described in any implementation manner of the first aspect above when executed by a processor.

[0025] In a fifth aspect, some embodiments of the present disclosure provide a computer program product including a computer program, and the computer program implements the user information push method described in any implementation manner of the first aspect above when executed by a processor.

[0026] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: The user information push method of some embodiments of the present disclosure can greatly increase the number of potential users mined and improve the probability of successful user signing. Specifically, for the logistics market expansion business scenario mentioned in the background art part, there are usually three solutions: One is that customers actively seek out the stationmasters and salespersons of logistics, describe their needs, and then sign contracts with the logistics; Another is that front-line couriers (little brothers) collect customer intentions when delivering express deliveries and submit them to the stationmaster / salesperson. The salesperson / stationmaster uses their professional advantages to communicate with the customers and enable them to sign contracts with JD Logistics; There is also a way of customer referral. Front-line salespersons rely on their extensive personal networks to attract new customers for the company.

[0027] However, these methods generally rely on human factors, and the potential users mined are often limited. For example, in the first method where customers actively seek out the stationmasters and salespersons of logistics, the initiative is given to the customers, and they submit reports on their own. This method can meet the needs of customers to a certain extent, but it is not conducive to the development of the logistics market. For the second and third methods, the areas responsible for delivery by each courier are usually fixed. That is to say, the users they come into contact with are limited. In the initial stage, some new users may be mined, but as more and more users in this area sign contracts, the number of users that can be mined will become fewer and fewer, and it is impossible to continuously mine new users for a long time. And through the method of promoting customer signing by visiting, the companies that front-line personnel can reach are limited, and they cannot contact the core decision-making personnel.

[0028] Based on this, the user information push method of some embodiments of the present disclosure can use big data to design a brand-new lead distribution plan. It can collect available B-side potential customers from all over the country, select high-score potential customers for cultivation using the recall sorting algorithm, and screen high-quality potential customers using the method of artificial cultivation. That is, it can use big data to obtain the historical mailing data of users, so as to extract user characteristic data. At the same time, it can obtain the historical business data of employees (i.e., the operation objects), so as to extract employee characteristic data. By calculating the matching degree between the two characteristic data, suitable employees can be selected. Furthermore, user information can be pushed to the matching sales employees, which is conducive to improving the user service experience and satisfaction, and thus helps to increase the probability of successful user signing. And using big data can greatly increase the number of potential customers mined, so as to achieve continuous signing of new customers. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In combination with the accompanying drawings and with reference to the following specific embodiments, the above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic, and the elements and elements are not necessarily drawn to scale.

[0030] Figure 1 is a flowchart of some embodiments of the user information push method of the present disclosure;

[0031] Figure 2 is a schematic diagram of some application scenarios of the user information push method of the present disclosure;

[0032] Figure 3 is a schematic diagram of the architecture of some embodiments of the matching model of the present disclosure;

[0033] Figure 4 is a schematic diagram of the structure of some embodiments of the user information push device of the present disclosure;

[0034] Figure 5 is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Embodiments

[0035] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not used to limit the protection scope of the present disclosure.

[0036] In addition, it should be noted that for the sake of convenience of description, only parts related to the relevant invention are shown in the drawings. Without conflict, the embodiments in the present disclosure and the features in the embodiments can be combined with each other.

[0037] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependent relationships.

[0038] In addition, the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly specified in the context, it should be understood as "one or more".

[0039] Figure 1 Shows process 100 of some embodiments of the user information push method according to the present disclosure. The method may include the following steps:

[0040] Step 101, extracting user feature data of the target user according to the historical mailing data of the target user.

[0041] In some embodiments, the execution subject of the user information push method (such as Figure 2The distribution engine server shown in [Figure X] can be communicatively connected to other electronic devices via wired or wireless connection. As an example, as Figure 2 shown, the execution entity can obtain the user information and historical mailing data of the target user from the human-machine collaboration module. Among them, the target user can be set according to actual needs. For example, it can be a user who has not signed up for the logistics service. Another example is that in order to improve the success rate of signing up, the target user here can also be an interested non-signed user, that is, a user with a relatively high signing probability. Here, Figure 2 the potential customer module in [Figure X] crawls potential customers on major websites through web scraping technology, and selects some high-quality potential customers for cultivation using recall and ranking algorithms. The human-machine collaboration module can cultivate leads for potential customers, supplement key customer information, and identify interested customers.

[0042] In some embodiments, the execution entity can extract the user characteristic data of the target user based on the historical mailing data of the target user. The historical mailing data here can be data related to historical mailed items. For example, it can include user basic information (such as name, gender, age, etc.), address information (such as the region where the sending address belongs, company name, company type, and industry), item information (such as item name, weight, quantity, etc.). The user characteristic data can be data representing the key characteristics of the target user. As an example, the execution entity can screen out the characteristic data highly relevant to potential customers from numerous data, such as screening according to preset attributes, so as to obtain the user characteristic data.

[0043] It should be noted that during the process of extracting characteristic data, there are often multiple types of data. For example, there are often certain fixed attributes in the user's basic data, such as province, city, industry, product, address, preference, etc.). The characteristics of such features are non-numerical and the values are not unique. In addition, there may also be numerical data (such as age, height, item weight and size, etc.) and time series data (such as the number of historical mailing orders, etc.). That is to say, the above user characteristic data can include at least one of the following: character type characteristic data, numerical type characteristic data, and time series characteristic data.

[0044] In some embodiments, in order to perform feature extraction from multiple perspectives and thus improve the accuracy of subsequent feature matching, for different types of data, the execution entity may adopt different processing methods. As an example, the execution entity may determine character data with a string length greater than a specified value as long string data. For such long string data, the Word2vec word vector model may be used to process the long string data to obtain a low-dimensional first character feature vector. Among them, Word2vec is a technology in the field of NLP (Natural Language Processing). It is a process of converting words into "computable" vectors. In addition, character data with a string length not greater than the specified value may be determined as short string data. For such short string data, the Embedding technology may be used to process the short string data to obtain a low-dimensional second character feature vector.

[0045] That is to say, for character feature data, in the embodiments of the present disclosure, it can be divided into two types. One is long string type feature data, and the other is short string type feature data. Since the long string type feature data has more feature values, if one-hot (one-hot encoding) is used, the dimension will become too high. Therefore, for such difficult-to-process character features, the word2vec method is used here for feature mapping to obtain its low-dimensional vector representation. Since the short string type feature has fewer feature values, in order to fully learn the representation of each dimension of the feature, the Embedding technology in TensorFlow (a deep learning platform) can be used here to map it to a low-dimensional vector representation.

[0046] Furthermore, for the above two different types of feature vectors, in order to facilitate subsequent matching processing, the execution entity may also splice the above first character feature vector and the above second character feature vector to obtain the character feature data of the target user. For example, the two feature vectors can be combined by using a residual connection method. Residual Connection can greatly improve the performance of the neural network. Many problems in deep learning, such as gradient disappearance and network degradation, can be solved by residual connection. Residual connection is generally commonly known as shortcut, and usually, the input signal and the output signal are directly added to ensure that the network obtains the ability of "skipping layers" during training. Specifically, when constructing a neural network, residual connection generally directly uses the output of the previous layer as one of the inputs of the subsequent layer.

[0047] In some alternative implementation manners, for the basic numerical type features, this embodiment also processes them in two cases. One is the numerical values with unique mapping relationships (such as height, age), and such numerical values usually do not have a magnitude relationship. Therefore, they can be processed according to the mapping relationship to obtain the corresponding feature vectors. The other is the numerical values representing magnitude relationships, such as the weight of the mailed item. Different weights result in different mailing fees. For such numerical data, the execution entity can normalize the attribute values of the same attribute. Based on the normalized numerical data, key information extraction is performed to obtain the numerical feature data of the target user. That is to say, for this type of data, standardization and normalization can be performed first. The normalization method is not limited. For example, each numerical value is divided by the average value, or (the difference between each numerical value and the minimum value) is divided by (the difference between the maximum value and the minimum value), etc. Then, a basic fully connected neural network can be used for feature extraction, and through the activation function (Relu), non-linear mapping is performed to extract the key information therein. And the final key information is retained as part of the final model (such as Figure 3 the matching model shown) to implement a more refined feature modeling solution.

[0048] It can be understood that time series features are often the key features for Figure 2 hot lead allocation. Here, by extracting the time series features, the historical mailing situation and interests of the user can be known. That is, key feature extraction is performed on the time series data to obtain the time series feature data of the target user. The time series feature data here can include the feature data in the first time period and the feature data in the second time period. Among them, the duration of the first time period can be greater than that of the second time period. That is to say, considering that the change of the customer state may be affected by multiple factors, in this embodiment, the long-term interest change and the short-term interest change of the customer can be considered simultaneously.

[0049] Step 102, according to the historical business data of multiple candidate operation objects, extract the object feature data of the multiple candidate operation objects.

[0050] In some embodiments, the execution entity can also extract the object feature data of multiple candidate operation objects according to the historical business data of the multiple candidate operation objects. Among them, the historical business data can be the data representing the historical business processed by the operation object, and the business type is not limited. The object feature data here can also include at least one of the following: character type feature data, numerical type feature data, and time series feature data. Such as Figure 2As shown in the figure, the data factory can provide portrait information of front-line logistics salespersons / station masters, and this part of the portrait can be obtained through the processing of historical data. Benefiting from the onlineization of logistics data, relatively complete historical business data of candidate employees (i.e., the objects to be operated) can be obtained. For example, the efficiency, effect, and preferences of each salesperson / station master in handling leads from various channels can be obtained, and the monthly achievement indicators and corresponding rewards of the salesperson can also be obtained.

[0051] As an example, the basic sales features can include sales type, war zone, area, province, city address, etc. The numerical type features can include daily lead volume, processing volume, abandonment volume, processing rate, business opportunity rate, signing rate, KPI (Key Performance Indicator), performance, etc. The time series features can include information such as the number, type, industry distribution, order volume distribution, revenue distribution of historical signed customers of the salesperson. It can be understood that through the time series feature data, the historical processing situation of the employee can be known, and it can also be identified whether the employee is enthusiastic about such hot leads (i.e., target users), so as to achieve precise matching.

[0052] It should be noted that for the extraction method of employee feature data, reference can be made to the extraction process of user feature data in step 101, which will not be elaborated here.

[0053] Step 103: Select a target operation object from multiple candidate operation objects based on the matching degree between the user feature data and the object feature data.

[0054] In some embodiments, the execution subject can determine the matching degree between the user feature data and each object feature data. Thus, based on the matching degree between the two, a target operation object can be selected from multiple candidate operation objects. The selection method here is also not limited. For example, in order to increase the probability of the user signing up for the logistics service, the candidate employee with the highest matching degree can be selected as the target employee.

[0055] In some embodiments, for the convenience of matching calculation, the execution subject can splice the three types of feature data to avoid feature loss. Specifically, key feature data can be extracted from the character type feature data, numerical type feature data, and time series feature data of the user feature data respectively. The key feature data of the three types of feature data are fused to be used for determining the matching degree. The matching degree calculation method here is also not limited. For example, common vector similarity algorithms such as Jaccard similarity can be used.

[0056] Optionally, the execution subject can also use a pre-trained matching model (such as Figure 3As shown, the matching degree between the target user and the operation object is obtained. Among them, the matching model can be used to analyze the similarity between the input feature data. Here, the execution subject can input the above user feature data and object feature data into the matching model. The matching model can extract key information from three types of feature data respectively to obtain a key vector containing important information. Then, feature fusion can be performed using the fully connected network and non-linear function mapping in the matching model. Finally, the cosine similarity can be used to calculate the matching degree between the current operation object and the user.

[0057] Here, the training samples of the matching model are obtained through the following method: Normalize each type of feature data in the obtained user feature data to obtain sample user feature data. The normalization method is not limited either. Among them, the dimension of each type of feature data in the sample user feature data is at least the sample size multiplied by the number of this type of feature data. Normalize each type of feature data in the obtained object feature data to obtain sample object feature data. Use the sample user feature data and the sample object feature data as training samples. That is to say, each training sample of the model can be composed of three parts: string type features, with a dimension of (sample size * the number of this type of feature data); numerical type features, with a dimension of (sample size * the number of this type of feature data); time series features, with a dimension of (sample size * the number of this type of feature data * time series length).

[0058] Here, a unique distribution model and strategy (i.e., the matching model) are adopted, which can make the leads and sales in a highly compatible state, promoting customers to sign up for the express delivery service of our logistics. The matching model can adopt various machine learning model structures. It should be noted that considering the particularity of the data in the logistics industry itself, referring to models such as the DeepFM model in the recommendation system, the lead assignment model (matching model) based on deep learning in this embodiment is designed. Since this solution is mainly for specific business, when designing the model, business rules and algorithm models will be integrated to form a unified and business-compatible high-quality lead assignment system. The matching model can be composed of multiple parts, such as Figure 3 As shown, it can be used to process features highly relevant to potential customers, such as the historical consignment item features, industry, region, target order volume, etc. of potential customers at the B-end (leads), as well as the basic features and historical sequence data of sales / station masters.

[0059] Step 104, Push the user information of the target user to the terminal of the target operation object so that the target operation object serves the target user.

[0060] In some embodiments, the executing entity may push the user information of the target user to the terminal device of the target operation object. After obtaining the user information, the target operation object can actively contact the target user, thereby serving the target user. For example Figure 2 As shown in Figure 2 , the distribution engine can use the sales portrait and the lead information for recommendation, recommend high-quality leads to the sales staff, and thus promote signing contracts.

[0061] From the above description, it can be seen that the user information pushing method according to some embodiments of the present disclosure can design a brand-new lead distribution scheme by using the method of big data. It can collect available B-side leads from all over the country, select high-score leads for cultivation by using the recall ranking algorithm, and screen high-quality leads by using the method of manual cultivation. That is, it can obtain the historical mailing data of users by using big data, so as to extract user feature data. At the same time, it can obtain the historical business data of employees (i.e., operation objects), and thus extract employee feature data. By calculating the matching degree between the two types of feature data, suitable employees can be selected. Furthermore, the user information can be pushed to the matching sales employees, which is beneficial to improving the user service experience and satisfaction, and thus helps to increase the probability of successful user signing. And using big data can greatly increase the number of potential customer mining, so as to achieve continuous new customer signing.

[0062] The innovation points of this solution are as follows:

[0063] (1) This solution refers to the modeling method of mainstream recommendation systems, and designs a recommendation-based lead allocation algorithm for B-side leads according to the characteristics of the logistics industry. This algorithm has a refined modeling scheme, integrates multiple time series models, can handle complex business scenarios, and solves the problem of low efficiency of traditional manual lead reporting.

[0064] (2) The solution designed in this solution realizes lead matching, target, and sample design, and can be reused in other matching and similarity calculation scenarios.

[0065] (3) Based on the model designed in this solution, this patent has conducted effect tests in actual logistics scenarios and achieved good results. Compared with traditional channels, it can bring more than 500 additional customer signings per month, increasing logistics revenue.

[0066] (4) The recommendation-based lead allocation algorithm designed in this solution integrates deep learning algorithms and is deeply compatible with the lead recommendation business in the logistics scenario, which is a general solution in the industry.

[0067] Continuing to refer to Figure 4 , as an implementation of the method shown above Figure 1 , some embodiments of a user information pushing device are provided in the present disclosure. These device embodiments are related toFigure 1 correspond to the method embodiments shown. The user information push device can be specifically applied to various electronic devices.

[0068] Such as Figure 4 As shown, the user information push device 400 in some embodiments may include: a user feature extraction unit 401 configured to extract user feature data of a target user according to the historical mailing data of the target user, where the user feature data includes at least one of the following: character type feature data, numerical type feature data, and time series feature data; an object feature extraction unit 402 configured to extract object feature data of a plurality of candidate operation objects according to the historical service data of the plurality of candidate operation objects; a selection unit 403 configured to select a target operation object from the plurality of candidate operation objects based on the matching degree between the user feature data and the object feature data; and a push unit 404 configured to push the user information of the target user to the terminal of the target operation object so that the target operation object serves the target user.

[0069] In some embodiments, the user feature extraction unit 401 may further be configured to determine character type data with a string length greater than a specified value as long string type data; process the long string type data using a Word2vec word vector model to obtain a low-dimensional first character type feature vector; and determine character type data with a string length not greater than the specified value as short string type data; process the short string type data using an embedding technique to obtain a low-dimensional second character type feature vector.

[0070] In some embodiments, the user feature extraction unit 401 may further be configured to splice the first character type feature vector and the second character type feature vector to obtain the character type feature data of the target user.

[0071] In some embodiments, the user feature extraction unit 401 may further be configured to normalize the attribute values of the same attribute for numerical type data representing size relationships; perform key information extraction based on the normalized numerical type data to obtain the numerical type feature data of the target user.

[0072] In some embodiments, the user feature extraction unit 401 may further be configured to perform key feature extraction on time series data to obtain the time series feature data of the target user, where the time series feature data includes feature data in a first time period and feature data in a second time period, and the duration of the first time period is greater than the duration of the second time period.

[0073] In some embodiments, the device 400 may further include a feature fusion unit (not shown in the figure), configured to respectively extract key feature data from the character-based feature data, numerical feature data, and time series feature data of the user feature data; fuse the key feature data of the three types of feature data for determining the matching degree.

[0074] In some embodiments, the device 400 may further include a matching degree determination unit (not shown in the figure), configured to input the user feature data and the object feature data into a pre-trained matching model to obtain the matching degree between the target user and the operating object, where the matching model is used to analyze the similarity between the input feature data.

[0075] In some embodiments, the device 400 may further include a training sample generation unit (not shown in the figure), configured to perform normalization processing on each type of feature data in the acquired user feature data to obtain sample user feature data, where the dimension of each type of feature data in the sample user feature data is the number of samples multiplied by the number of this type of feature data; perform normalization processing on each type of feature data in the acquired object feature data to obtain sample object feature data; use the sample user feature data and the sample object feature data as training samples.

[0076] It can be understood that the various units described in the user information push device 400 correspond to the respective steps in the user information push method described in the reference Figure 1 Therefore, the operations, features, and beneficial effects described above for the method also apply to the user information push device 400 and the units included therein, and will not be elaborated herein.

[0077] Next, refer to Figure 5 , which shows a schematic structural diagram of an electronic device 500 suitable for implementing some embodiments of the present disclosure. Figure 5 The electronic device shown is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present disclosure.

[0078] As Figure 5 shown, the electronic device 500 may include a processing device 501 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 502 or the program loaded from the storage device 508 into the random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0079] Typically, the following devices can be connected to the I / O interface 505: an input device 506 including, for example, a touch display screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, etc.; an output device 507 including, for example, a speaker, a vibrator, etc.; a storage device 508 including, for example, a memory card, etc.; and a communication device 509. The communication device 509 can allow the electronic device 500 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 5 the electronic device 500 with various devices is shown, it should be understood that it is not required to implement or have all the shown devices. More or fewer devices can be alternatively implemented or had. Figure 5 Each block shown in

[0080] Specifically, according to some embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such some embodiments, the computer program can be downloaded and installed from the network through the communication device 509, or installed from the storage device 508, or installed from the ROM 502. When the computer program is executed by the processing device 501, the above functions defined in the methods of some embodiments of the present disclosure are executed.

[0081] It should be noted that the computer-readable media described in some embodiments of the present disclosure may be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0082] In some embodiments, the client and the server can communicate using any currently known or future-developed network protocol such as HTTP (Hyper Text Transfer Protocol), and can be interconnected with digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include local area networks ("LANs"), wide area networks ("WANs"), the Internet (e.g., the Internet), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0083] The above computer-readable medium may be included in the above electronic device; or may exist separately without being assembled into the electronic device. The above computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: extract user characteristic data of a target user based on the historical mailing data of the target user, where the user characteristic data includes at least one of the following: character-type characteristic data, numerical-type characteristic data, and time-series characteristic data; extract object characteristic data of a plurality of candidate operation objects based on the historical service data of the plurality of candidate operation objects; select a target operation object from the plurality of candidate operation objects based on the matching degree between the user characteristic data and the object characteristic data; and push the user information of the target user to the terminal of the target operation object so that the target operation object serves the target user.

[0084] In addition, computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., by connecting through the Internet using an Internet service provider).

[0085] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0086] The units described in some embodiments of the present disclosure can be implemented in software or in hardware. The described units can also be provided in a processor. For example, it can be described as: a processor includes a user feature extraction unit, an object feature extraction unit, a selection unit, and a push unit. Among them, the names of these units do not constitute a limitation to the unit itself in some cases. For example, the user feature extraction unit can also be described as "a unit that extracts user feature data of a target user according to the historical mailing data of the target user".

[0087] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, without limitation, the exemplary types of hardware logic components that can be used include: Field Programmable Gate Array (FPGA), Application Specific Integrated Circuit (ASIC), Application Specific Standard Product (ASSP), System on Chip (SOC), Complex Programmable Logic Device (CPLD), and so on.

[0088] Some embodiments of the present disclosure also provide a computer program product, including a computer program which, when executed by a processor, implements any one of the above user information push methods.

[0089] The above description is only some preferred embodiments of the present disclosure and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features with similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A user information pushing method, comprising: extracting user feature data of the target user according to the historical mailing data of the target user, wherein the user feature data includes at least one of the following: character type feature data, numerical type feature data, and time series feature data; extracting object feature data of the multiple candidate operation objects according to the historical business data of the multiple candidate operation objects; selecting a target operation object from the multiple candidate operation objects based on the matching degree between the user feature data and the object feature data; pushing the user information of the target user to the terminal of the target operation object, so that the target operation object serves the target user.

2. The user information push method according to claim 1, wherein, The extracting user feature data of the target user according to the historical mailing data of the target user includes: determining character type data with a string length greater than a specified value as long string type data; processing the long string type data by using a Word2vec word vector model to obtain a low-dimensional first character type feature vector; and determining character type data with a string length not greater than the specified value as short string type data; processing the short string type data by using an embedding technique to obtain a low-dimensional second character type feature vector.

3. The user information pushing method according to claim 2, wherein, The extracting user feature data of the target user according to the historical mailing data of the target user further includes: concatenating the first character type feature vector and the second character type feature vector to obtain the character type feature data of the target user.

4. The user information push method according to claim 1, wherein, The extracting user feature data of the target user according to the historical mailing data of the target user further includes: for numerical type data representing a size relationship, normalizing the attribute values of the same attribute; extracting key information based on the normalized numerical type data to obtain the numerical type feature data of the target user.

5. The user information push method according to claim 1, wherein, The extracting user feature data of the target user according to the historical mailing data of the target user further includes: extracting key features from time series data to obtain the time series feature data of the target user, wherein the time series feature data includes feature data in a first time period and feature data in a second time period, and the duration of the first time period is greater than the duration of the second time period.

6. The user information pushing method according to claim 1, wherein, The method further includes: respectively extracting key feature data from the character type feature data, numerical type feature data, and time series feature data of the user feature data; fusing the key feature data of the three types of feature data for determining the matching degree.

7. The user information pushing method according to any one of claims 1-6, wherein, The method further includes: inputting the user feature data and the object feature data into a pre-trained matching model to obtain the matching degree between the target user and the operation object, wherein the matching model is used to analyze the similarity between the input feature data.

8. The user information push method according to claim 7, wherein The training samples of the matching model are obtained by the following method: Normalize each type of feature data in the obtained user feature data to obtain sample user feature data, where the dimension of each type of feature data in the sample user feature data is at least the product of the sample size and the number of this type of feature data; Normalize each type of feature data in the obtained object feature data to obtain sample object feature data; Use the sample user feature data and the sample object feature data as training samples.

9. A user information push device, comprising: A user feature extraction unit configured to extract user feature data of a target user according to historical mailing data of the target user, where the user feature data includes at least one of the following: character type feature data, numerical type feature data, and time series feature data; An object feature extraction unit configured to extract object feature data of multiple candidate operation objects according to historical service data of the multiple candidate operation objects; A selection unit configured to select a target operation object from the multiple candidate operation objects based on the matching degree between the user feature data and the object feature data; A push unit configured to push user information of the target user to a terminal of the target operation object, so that the target operation object serves the target user.

10. An electronic device, comprising: One or more processors; A storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the user information push method according to any one of claims 1-8.

11. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, the user information push method according to any one of claims 1-8 is implemented.

12. A computer program product, comprising a computer program which, when executed by a processor, implements the user information push method according to any one of claims 1-8.

Images