CN114741595A - Method and device for pushing information - Google Patents

Abstract

The invention discloses a method and a device for pushing information, and relates to the technical field of computers. One embodiment of the method comprises: acquiring a user identifier set to be aligned, and performing identifier alignment processing on the user identifier set to be aligned and a business district user identifier set to obtain an aligned user identifier; obtaining moving track data of the aligned user identification according to service data, and determining a dense source of a visiting user based on the moving track data; respectively calculating the guest group portraits of the visiting users in each intensive source area according to the feature data of the visiting users; and comparing the similarity of the customer group image with the historical user images of the business circles to determine a target user, and pushing information of the target user. The embodiment can make up the defect of single characteristic dimension of user data, can better, more timely and more accurately depict the user customer group images, more accurately determines potential customers of a business circle, and is convenient for improving the success rate of information push.

Description

Method and device for information push

technical field

The present invention relates to the field of computer technology, and in particular, to a method and device for pushing information.

Background technique

In the current marketing scenarios of business circles (for example: advertisement placement, product recommendation, etc.), marketing mainly focuses on the repurchase of existing customers and the activation of potential new customers. At present, when mining potential new customers, crowd exploration is mostly carried out from local long-term and short-term user behavior data in the business district.

In the process of realizing the present invention, the inventor found that there are at least the following problems in the prior art:

The business district operator only conducts crowd exploration based on local long-term and short-term user behavior data in the business district, and lacks a detailed understanding of the customer flow outside the business district and the consumption of stores in the business district; The reason is that the business district cannot effectively use the data of other external agencies, so the data dimension is single and cannot effectively carry out marketing. The consumption characteristics of users in the customer group portraits are affected by market public opinion, and there are often large differences in the value of the characteristics at different times. The existing business circle insight methods do not consider the real-time characteristics of user data, so the obtained customer group portraits are inaccurate and cannot meet the real-time dynamic query in business analysis. When matching target new customers based on existing visiting users, the prior art only expands similar groups according to the characteristic attributes of a single user, and the obtained target new customers are often inaccurate, thereby affecting the marketing effect.

To sum up, when mining potential new customers in the current business district, there are situations in which the data dimension is single, the customer group portrait is not real-time, and the matching of target new customers is inaccurate. Therefore, the marketing effect of the business district is reduced.

SUMMARY OF THE INVENTION

In view of this, the embodiments of the present invention provide a method and device for information push, which can collect multi-party data to perform user portrait characterization and information push, make up for the defect of a single dimension of user data features in the business circle operation scenario, and can better, More real-time and more accurate depiction of user customer group portraits, more accurate identification of potential customers in the business district, and easy to improve the success rate of information push.

To achieve the above object, according to an aspect of the embodiments of the present invention, a method for pushing information is provided.

A method for pushing information, including:

Obtaining the user identification set to be aligned, and performing identification alignment processing on the user identification set to be aligned and the business circle user identification set to obtain the aligned user identification;

Acquire the movement trajectory data of the aligned user identifiers according to the business data, and determine the dense source of visiting users based on the movement trajectory data;

According to the characteristic data of the visiting users, calculate the customer group portraits of the visiting users in each dense source;

The similarity between the customer group portrait and the historical user portrait in the business district is compared to determine a target user, and information is pushed to the target user.

Optionally, the logo alignment process includes:

Take intersection processing, take union processing, left outer join processing or right outer join processing.

Optionally, performing identification alignment processing on the to-be-aligned user identification set and the business circle user identification set includes:

The user identification set to be aligned is sent to a third-party arbitration institution, so that the third-party arbitration institution performs identification alignment processing according to the user identification set to be aligned and the business district user identification set.

Optionally, before performing identification alignment processing on the user identification set to be aligned and the business circle user identification set, the method further includes:

Obtain the processing rules of the user IDs agreed with the business circle in advance, and the processing rules include the data format and encryption method of the user IDs;

The user identification set to be aligned is processed by using the processing rule, and the processed user identification set to be aligned is used as the user identification set to be aligned.

Optionally, determining the dense sources of visiting users based on the movement trajectory data includes:

Determine the residence and work place of each visiting user based on the movement trajectory data;

Use the spatial clustering algorithm to cluster residence and work, and obtain multiple clustered areas;

From the plurality of clustering regions, a specified number of clustering regions where the visiting users are most densely distributed in space are selected as the dense source places of the visiting users.

Optionally, compare the similarity between the customer group portraits and the historical user portraits of the business district to determine that the target users include:

Category coding is performed on the structured portrait data in the customer group portrait and the historical user portrait in the business district respectively, and the portrait representation vector is obtained through dimensionality reduction;

Extracting social network representations from the unstructured portrait data in the customer group portraits and the historical user portraits of the business district, respectively, to obtain a network representation vector;

splicing the portrait representation vector and the network representation vector together to obtain the feature vector of the customer group portrait and the historical user portrait of the business district respectively;

Based on the similarity between the feature vectors, the similarity between the customer group portrait and the historical user portrait in the business district is compared, and the visiting user corresponding to the customer group portrait whose similarity meets a preset threshold is determined as the target user.

Optionally, performing information push to the target user includes:

Acquire the first dense source of the target user, and return the first dense source to the business circle, so that the business circle pushes information according to the first dense source.

According to another aspect of the embodiments of the present invention, an apparatus for pushing information is provided.

A device for pushing information, comprising:

a user identification aligning module, configured to obtain a user identification set to be aligned, and perform identification alignment processing on the user identification set to be aligned and the business circle user identification set to obtain an aligned user identification;

a user source determination module, configured to obtain the movement trajectory data of the aligned user identifiers according to the business data, and determine the dense source of visiting users based on the movement trajectory data;

The customer group portrait calculation module is used to calculate the customer group portrait of the visiting users in each dense source according to the characteristic data of the visiting users;

The target user determination module is used to compare the similarity between the customer group portrait and the historical user portrait of the business district to determine the target user, and push the information to the target user.

According to another aspect of the embodiments of the present invention, an electronic device for pushing information is provided.

An electronic device for information push, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, so that all The one or more processors implement the information push method provided by the embodiment of the present invention.

According to yet another aspect of the embodiments of the present invention, a computer-readable medium is provided.

A computer-readable medium on which a computer program is stored, and when the program is executed by a processor, implements the information push method provided by the embodiment of the present invention.

One embodiment of the above invention has the following advantages or beneficial effects: by acquiring the user identification set to be aligned, and performing identification alignment processing on the user identification set to be aligned and the business circle user identification set, the aligned user identification is obtained; the alignment user identification is obtained according to business data; Identify the movement trajectory data, and determine the intensive source of visiting users based on the movement trajectory data; according to the characteristic data of the visiting users, calculate the customer group portraits of the visiting users in each dense source area; The similarity comparison of the historical user portraits in the circle to determine the target users, and push the information of the target users, realize the user portrait characterization and information push through the collection of multi-party data, and make up for the single dimension of user data characteristics in the business circle operation scenario. Defects, realize the full mining of the source of unvisited users in the business district, so as to better push advertisements and other information in the scenario of new customer drainage in the business district; based on the spatiotemporal attributes of user trajectory information, obtain the precise and dense source of users , so that it can better, more real-time and more accurately describe the user group portrait, more accurately determine the potential customers of the business district, and improve the success rate of information push.

Further effects of the above non-conventional alternatives will be described below in conjunction with specific embodiments.

Description of drawings

The accompanying drawings are used for better understanding of the present invention and do not constitute an improper limitation of the present invention. in:

1 is a schematic diagram of main steps of a method for information push according to an embodiment of the present invention;

Fig. 2 is the implementation flow schematic diagram of the embodiment of the present invention;

3 is a schematic diagram of a customer group portrait construction and a new customer mining process according to an embodiment of the present invention;

4 is a schematic diagram of main modules of an apparatus for pushing information according to an embodiment of the present invention;

5 is an exemplary system architecture diagram to which an embodiment of the present invention may be applied;

FIG. 6 is a schematic structural diagram of a computer system suitable for implementing a terminal device or a server according to an embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, which include various details of the embodiments of the present invention to facilitate understanding and should be considered as exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted from the following description for clarity and conciseness.

In the current marketing scenario of business circles, marketing mainly focuses on the repurchase of existing customers and the activation of potential new customers. However, due to the fact that the local data is not only seriously insufficient in data and feature dimensions, but also always limited to the data generated within the current business district. For example, an operating company in a business district hopes to know the source of potential customers who have never visited the business district, the crowd portraits, etc., and through targeted analysis and drainage of the source of target new customers, to attract new customers for various formats. . In order to fully expand the marketing radiation scope of the business circle, it is necessary to fully mine the crowd characteristics data of the passenger flow outside the business circle, analyze the data characteristics and travel characteristics of various dimensions for a comprehensive comparison, lock the focus of the existing passenger flow and find potential The customer source information provided by the business insight new customer insight section provides data support. However, the operators of the business circle do not have the demographic data and consumption data of customers outside the business circle, and even the specific consumption data of the shops in the business circle are difficult to obtain due to the commercial secrets. Corresponding marketing has formed a great resistance. If the business district can access online passenger flow data, it can analyze the target new customers to attract new customers for various business formats, thereby increasing revenue.

The main defects in the current business district in the process of mining new customers are as follows:

1. Sparse feature dimensions: The data of the business district operators are only based on the local long-term and short-term user behavior data of the business district for crowd exploration, and they lack a detailed understanding of the passenger flow outside the business district and the consumption of stores in the business district; Due to the protection of interests and the privacy and security of personal data, the business district cannot effectively use the data of other external organizations, so the data dimension is single, and marketing cannot be effectively carried out.

2. Customer group portraits are not real-time: the consumption characteristics of users in the customer group portraits are affected by market public opinion, and there are often large differences in the value of the characteristics at different times. The existing business circle insight methods do not consider the real-time characteristics of user data, so the obtained customer group portraits are inaccurate and cannot meet the real-time dynamic query in business analysis.

3. Inaccurate matching of target new customers: when the existing technology matches target new customers based on existing visiting users, it only uses the TF-IDF method to expand similar groups according to the characteristic attributes of a single user, and the obtained target new customers are often inaccurate. Accurate, thereby affecting the marketing effect.

In order to solve the defect existing in the above-mentioned prior art, the present invention mainly adopts the following technical means:

1. Solve the problem of insufficient data dimension of business operators. By combining the data of one or more external partners, under the condition that the privacy of individual user data is protected and business secrets are not disclosed, it can safely realize multi-party and cross-domain data statistics query , complement the data characteristics of users inside and outside the business circle, and find the target new customer gathering places to achieve the purpose of assisting commercial marketing to expand the radiation area of the business circle through targeted drainage methods such as advertising, bus drainage, and business format optimization;

2. Based on the spatiotemporal attributes of user trajectory information, spatial clustering is performed through different users' long-term residence time periods to obtain the precise source of users; Data and long-term data on a daily and weekly basis, build customer group portraits of visiting users and target new customers, reflect the characteristics of customer groups dynamically in real time, and recommend target new customers in real time according to visiting users; business circles cooperate with social platforms , using the social graph network, through the Graph Embedding method of LINE, to extract the first-order and second-order similarity representations of the user's social graph as features to improve the matching accuracy of target new customers.

In the introduction of the embodiments of the present invention, the related technical terms and their definitions are as follows.

1. OPTICS algorithm: The full name of OPTICS algorithm is Ordering points to identify the clustering structure, and the goal is to cluster the data in space according to the density distribution. Its idea is very similar to DBSCAN, but DBSCAN is difficult to select input parameters, that is, DBSCAN is more sensitive to input parameters. The difference between OPTICS and DBSCAN is that the density-based clustering structure can show a special order. The clustering structure corresponding to this order contains the information of each level of clustering and is easy to analyze. For the neighborhood radius A slight change in , does not affect the clustering results. The OPTICS algorithm has many advantages:

1) There is no need to specify the number of clusters in advance, and clusters of arbitrary shapes can be found;

2) It is not sensitive to abnormal points, and can automatically identify abnormal points in the clustering process;

3) The clustering result does not depend on the traversal order of nodes;

4) It is not sensitive to the neighborhood radius parameter, and the clustering result is more stable.

2. TF-IDF: TF-IDF (Term Frequency-inverse Document Frequency) is a statistical analysis method for keywords, which is used to evaluate the importance of a word to a document set or a corpus. The importance of a word is proportional to the number of times it appears in the article and inversely proportional to the number of times it appears in the corpus. This calculation method can effectively avoid the influence of common words on keywords, and improve the correlation between keywords and articles. Among them, TF refers to the total number of times a word appears in the article. This indicator is usually defined as TF=(the number of times a word appears in the document/the total number of words in the document), which can prevent the result from being biased too much. Long documents (the same word in a long document usually has a higher word frequency than a short document). IDF reverses the document frequency. The fewer documents that contain a word, the greater the IDF value, indicating that the word has a strong ability to distinguish. IDF=loge (the total number of documents in the corpus / the number of documents containing the word + 1), +1 The reason is to avoid a denominator of 0. TF-IDF=TFxIDF, the larger the TF-IDF value, the greater the importance of the feature word to the text.

3. Cosine Similarity: Cosine Similarity evaluates the similarity of two vectors by calculating the cosine value of the angle between them. Draw the vectors into the vector space according to the coordinate values, obtain their included angle, and obtain the cosine value corresponding to the included angle. This cosine value can be used to characterize the similarity of the two vectors. The smaller the included angle, the closer the cosine value is to 1, and the closer their directions are, the more similar they are.

4. LINE: LINE (Larg-scale Information Network Embedding) was proposed by Jian Tang in 2015. This method proposes a node embedding algorithm that can be applied to large-scale networks of any edge type. By considering first-order proximity (local structure) and second-order proximity (global structure) for network embedding. Compared with the previous Graph/Network Embedding method, LINE has the following advantages: 1. It is suitable for any type of network. The arbitrary type here mainly refers to the weight and direction of the edge in this article: directed, undirected, and weighted , without weight. (LINE does not consider heterogeneous networks under different node types and edge types, which has certain limitations. There are subsequent studies on heterogeneous and homogeneous networks); 2.LINE proposes an edge-sampling (edge-sampling) ) algorithm to boost and optimize the objective function, thus overcoming the limitations of traditional stochastic gradient decent.

5. Homomorphic encryption: Homomorphic encryption (HE) is a type of encryption method with special natural properties. This concept was first proposed by Rivest et al. in the 1970s. Compared with general encryption algorithms, homomorphic encryption In addition to basic encryption operations, encryption can also implement various computing functions between ciphertexts, that is, computing after decryption can be equivalent to computing after decryption.

FIG. 1 is a schematic diagram of main steps of a method for pushing information according to an embodiment of the present invention. In this embodiment, the execution subject of the method for pushing information is a partner of the business circle. The partners here can be mobile APPs, online shopping websites, offline shopping platforms, online or offline travel product sales platforms, etc., which mainly depend on the business needs of the business district parties and the actual information that the partners can provide. As shown in FIG. 1 , the method for pushing information according to the embodiment of the present invention mainly includes the following steps S101 to S104.

Step S101: obtaining a user identification set to be aligned, and performing identification alignment processing on the user identification set to be aligned and the business circle user identification set to obtain an aligned user identification;

Step S102: obtaining the movement track data of the aligned user identifiers according to the business data, and determining the dense source of visiting users based on the movement track data;

Step S103: according to the characteristic data of the visiting users, respectively calculate the customer group portraits of the visiting users in each dense source;

Step S104: Compare the similarity between the customer group portrait and the historical user portrait in the business district to determine a target user, and push information to the target user.

According to an embodiment of the present invention, the identification alignment processing includes: intersection processing, union processing, left outer join processing or right outer join processing. The way of identification (ID) alignment is not fixed. There are also various methods based on security considerations, but the core is to take the intersection. If the actual business needs, it can also include the union, or the left outer join, the right outer join and so on.

According to another embodiment of the present invention, performing identification alignment processing on the user identification set to be aligned and the business circle user identification set includes: sending the user identification set to be aligned to a third-party arbitration organization, so that the third party The arbitration institution performs identification alignment processing according to the user identification set to be aligned and the business district user identification set. Among them, the third-party arbitration institution is an independent third-party neutral security node. However, in the specific implementation process, in addition to summarizing the user IDs to a third-party arbitration institution for ID alignment, if the merchant and the business district trust each other safely, the partner can also send the user ID set to be aligned directly to the business district. For the operator, after the business circle side completes the logo alignment, the business circle side will send it to the partner, or vice versa, depending on which method meets the business and security compliance requirements. That is, the third-party arbitration institution can be a third-party neutral security node different from the merchant side and the business circle side, or it can be either the merchant side or the business circle side that meets the business and security compliance requirements.

According to yet another embodiment of the present invention, before performing the identification alignment process on the user identification set to be aligned and the business circle user identification set, the method further includes: acquiring a processing rule for user identification agreed with the business circle in advance, the processing rule Including the data format and encryption method of the user identification; using the processing rule to process the user identification set to be aligned, and using the processed user identification set to be aligned as the user identification set to be aligned. Before performing the identification alignment process, the user identification set to be aligned and the business circle user identification set need to be uniformly processed. Specifically, the business circle party and the partner must specify a unified user identification data format for identification ID alignment. Here, the ID is a unique identification. Aligned ID types can be different according to different marketing scenarios. In addition, a unified ID encryption method should also be specified. The ID cannot be transmitted in plain text, but must be cipher text, and the encryption method should be unified. The encryption method can be in various forms, such as MD5 or hash encryption, or a combination of various encryption methods, so as to avoid the encrypted ID from being decrypted, so as to better ensure the security and privacy of the data.

After the identity alignment is performed, for the aligned user identity, the partner queries and aligns the user's movement track information in the past year according to its own business data (eg, mobile phone signaling data, etc.). Among them, the signaling data is the data collected by the mobile operator on the current time and current location of the mobile terminal at a fixed frequency, and can be obtained after the mobile operator performs security processing and authorization. In addition, assuming that the partner is an e-commerce platform, the user's movement track information can be queried according to the delivery address in the order data of the aligned user, and so on. For different business scenarios, the alignment user's movement trajectory information can be obtained in different ways.

According to another embodiment of the present invention, determining the dense source of visiting users based on the movement trajectory data includes: determining the residence and work place of each visiting user based on the movement trajectory data; Clustering at the site and work place to obtain a plurality of clustering areas; from the plurality of clustering areas, select a specified number of clustering areas with the most dense distribution of visiting users in space as the dense source of visiting users . In an embodiment of the present invention, it is assumed that the alignment user's movement trajectory information is obtained through mobile phone signaling data. First, after obtaining the alignment user trajectory information, the partner judges the user's residence and work place according to the trajectory information. The logic is: the user's high-frequency residence point from 9:00 to 17:00 on weekdays is the work place, and the user's high-frequency residence point from 10:00 p.m. to 8:00 a.m. is the residence (dwelling point: sample trajectory information for a long time) the stop position). Then, complete the calculation of all the user's residence and work, and use the spatial aggregation algorithm to cluster the residence and work to obtain the clustering area. Finally, from the clustering results, a specified number (for example, 10) of the clustering areas with the most densely distributed visiting users in the space are screened out and displayed on the map, that is, the top ten densely sourced places of visiting users in the business district.

After that, after determining the dense source of visiting users, the customer group portrait of the visiting users in each dense source will be calculated according to the characteristic data of the visiting users. The dimensions of the customer group portrait include: the number of sample visits in a fixed period of time, the time distribution of visiting users (for example: counting the number of visiting samples in different periods), component analysis, social graph analysis, etc. The composition analysis can include gender , age, children, spending power, shopping preferences, and other structural features; social graph analysis mainly uses the relevant data information of social software, and the relationship with the relatives, friends and colleagues of the visiting users is extracted by using the LINE algorithm in GraphEmbedding. A representation that reflects the first-order and second-order similarity is used as the basis for matching the customer group portraits of visiting users with the target new customer group portraits in the later stage. At the same time, in the processing of structured features, different processing is adopted for the temporal attributes of user tags: for static features, their representations are stored in the content library to facilitate later use; for dynamic features, such as user preferences, interests, etc. change with time For features that change frequently, long-term representation vectors can be calculated in units of days and weeks, and short-term representation vectors can be calculated in units of hours and minutes, and the short-term representation vectors and long-term representation vectors can be weighted for later calculation and matching of target new customers. to achieve real-time user recommendation. The customer group portrait here does not judge the specific portrait or label of the individual.

According to another embodiment of the present invention, the similarity between the customer group portrait and the historical user portrait in the business district is compared to determine that the target users include:

Category coding is performed on the structured portrait data in the customer group portrait and the historical user portrait in the business district respectively, and the portrait representation vector is obtained through dimensionality reduction;

Extracting social network representations from the unstructured portrait data in the customer group portraits and the historical user portraits of the business district, respectively, to obtain a network representation vector;

splicing the portrait representation vector and the network representation vector together to obtain the feature vector of the customer group portrait and the historical user portrait of the business district respectively;

Based on the similarity between the feature vectors, the similarity between the customer group portrait and the historical user portrait in the business district is compared, and the visiting user corresponding to the customer group portrait whose similarity meets a preset threshold is determined as the target user.

Among them, the historical user portraits of the business district are characterized by feature extraction of historical users who have visited the business district. algorithm to achieve. The similar user mining algorithm is as follows: First, for structured portrait data (such as age, gender, spending power, etc.), perform category coding (such as: male is marked as 0001, female is marked as 1000), and through dimensionality reduction to get a more streamlined Portrait representation vector. Then, for unstructured relationships (such as social graphs), the LINE algorithm of Graph Embedding is used to extract social network representations to obtain network representation vectors, which are spliced together with portrait representation vectors. Finally, the similarity between samples is calculated according to the characterization vector of the samples obtained after splicing, and the samples that exceed a certain threshold are the target new customer group. The sample here is a user.

According to yet another embodiment of the present invention, pushing the information to the target user includes: acquiring a first dense source of the target user, and returning the first dense source to the business circle, so that the business The circle pushes information according to the first dense source. After the target user is determined, since the movement track of the target user is not limited to a certain business circle, the dense source of the target user can also be calculated according to the method of determining the dense source of the user in the aforementioned step S102, so as to better Determine the scope and content of information push in the business district.

In addition, in order to ensure the security and privacy of the data, when comparing the similarity between the customer group portrait and the historical user portrait in the business district to determine the target user, it can also be done through a third-party arbitration agency. And, after determining the first intensive source of target users, it can also be returned to a third-party arbitration institution for aggregation, and then sent to the business circle operator by the third-party arbitration institution for follow-up marketing planning. The partners here can be one or more, depending on the content needs of marketing. When constructing the customer group portrait, the user features come from the business circle’s own label features, e-commerce label features, social graph features of social platforms, etc. The feature vector data of all parties cannot be aggregated to one party to calculate the cosine similarity. Therefore, the homomorphic encryption technology is used to encrypt the results of different feature vectors of the three-party platforms through public key encryption and send them to a third-party arbitration institution. square.

FIG. 2 is a schematic diagram of an implementation flow of an embodiment of the present invention. As shown in FIG. 2 , in an embodiment of the present invention, it shows how to push information based on the business circle party, the cooperative party and the third-party arbitration institution. It mainly includes the following steps:

1. Both parties (business party and partner) agree on the processing rules for user identification IDs: At the beginning of the project, the business community and partners stipulate a unified user data ID format for ID alignment. The ID here is unique of identity. Aligned ID types can be different according to different marketing scenarios. In addition, a unified ID encryption method should also be specified. The ID cannot be transmitted in plaintext, but must be ciphertext, and the encryption method should be unified. The encryption method can have various forms, MD5 or hash encryption, or a combination of various encryption methods to avoid the encrypted ID from being decrypted;

2. Both parties (business district party and partner) prepare ID sets and feature sets: The business district party provides users who have visited in the past year based on consumption data of its own businesses (such as catering, accommodation, tourism, travel, etc.). ID. Similarly, the partner provides the access user ID for the past year based on its own business data;

3. ID alignment preparation: The operator of the business district can initiate an ID alignment task at this time and send it to the partner, so that the partner can choose to accept or reject the ID alignment task;

4. ID alignment processing: both parties (business circle party and partner) encrypt the IDs that need to be aligned and send them to a third-party arbitration institution (independent third-party neutral security node), and the arbitration institution performs ID alignment (for example: taking the intersection set ) operation, after the alignment is passed to the partner by the arbitration institution. In the specific implementation, in addition to summarizing the ID to the arbitration institution, in the case of mutual trust between the business district party and the partner, the partner can also send the ID directly to the business district party, and complete the ID alignment in the business district party. Afterwards, the business district party will send it to the partner, or vice versa, depending on which method meets the business and security compliance requirements. The way of ID alignment is not fixed. There are also many methods based on security considerations, but the core is to take the intersection. If the actual business needs, it can also include the union, or the left outer join, the right outer join and so on. The partners here can be mobile APPs, online shopping websites, offline shopping platforms, online or offline travel product sales platforms, etc., which mainly depend on the business needs of the business district parties and the actual information that the partners can provide;

5. Obtain the trajectory information of the alignment ID: After obtaining the encrypted alignment ID (that is, the data sample shared by both parties), the partner queries the movement trajectory information of the sample in the past year according to its own business data (such as mobile phone signaling, etc.). Among them, the signaling data is the data collected by the mobile operator at a fixed frequency of the current time and current location of the mobile terminal;

6. Calculate the dense source of visiting users: First, after obtaining the alignment sample trajectory information, the partner judges the sample residence and work place according to the trajectory information, and the judgment logic sample is in the high-frequency station from 9:00 to 17:00 on weekdays. The staying point is the work place, and the high-frequency residence point of the sample from 10 pm to 8 am is the residence (residency point: the location where the sample trajectory information stays for a long time). Then, complete the calculation of all the sample residence and work places, and use the spatial aggregation algorithm to cluster the residence and work places to obtain the clustering area. Finally, from the clustering results, the 10 cluster areas with the most dense distribution of visiting users in the space are screened out and displayed on the map, that is, the top ten densely sourced places of visiting users in the business district;

7. Depicting the portraits of visiting user groups: After obtaining the top ten densely sourced places of visiting users in the business district, calculate the portraits of the sample groups of all origins. The dimensions of group portraits include: the number of sample visits in a fixed period of time, the time distribution of visiting users (such as counting the number of visiting samples in different periods), component analysis, social graph analysis, etc. The composition analysis can include gender, age , the presence or absence of children, spending power, shopping preferences, and other structural features; social graph analysis mainly uses the relevant data information of social software, and the relationship with the relatives, friends and colleagues of the visited users is extracted by using the LINE algorithm in Graph Embedding. The representations reflecting the first-order and second-order similarity are used as the basis for matching the customer group portraits of visiting users with the target new customer group portraits in the later stage. At the same time, in the processing of structured features, different processing is adopted for the temporal attributes of user tags: for static features, their representations are stored in the content library to facilitate later use; for dynamic features, such as user preferences, interests, etc. change with time For features that change frequently, long-term representation vectors can be calculated in units of days and weeks, and short-term representation vectors can be calculated in units of hours and minutes, and the short-term representation vectors and long-term representation vectors can be weighted for later calculation and matching of target new customers. to achieve real-time user recommendation. The group portrait here does not judge the specific portrait or label of the individual;

8. Acquire the target new customer group: The partner will obtain the user portraits of the visiting users in the business circle and the historical user portraits of the business circle through the similar user mining algorithm to obtain the user IDs of the non-visiting users with a high similarity to the visiting users. , which is the target new customer ID. The similar user mining algorithm is as follows: First, for structured portrait data (such as age, gender, spending power, etc.), perform category coding (such as: male is marked as 0001, female is marked as 1000), and through dimensionality reduction to get a more streamlined Portrait representation vector. Then, for the unstructured relationship (such as social graph), the LINE algorithm of Graph Embedding is used to extract the social network representation, obtain the network representation vector, and splicing it together with the portrait representation vector. Finally, the similarity between the samples is calculated according to the characterization vector of the samples, and the samples that exceed a certain threshold are the target new customer group.

9. Obtain the dense source of the target new guest group: To obtain the target new guest group, use the method as in step 6 to calculate the top ten dense sources of the target new guest group;

10. Return the result: return it to a third-party arbitration institution for summary, and then the arbitration institution will send it to the business circle operator for follow-up marketing planning. The partners here can be one or more, depending on the content of marketing;

11. The actual marketing activities are carried out by the business district side.

FIG. 3 is a schematic diagram of a process of constructing a portrait of a customer group and mining new customers according to an embodiment of the present invention. As shown in Figure 3, after the business district party and the partner perform user feature extraction and dimensionality reduction, group portraits are obtained. User i and user j shown in the figure both represent group portraits. After that, an intermediate result is obtained by comparing the group portraits by similarity, and then the intermediate result is homomorphically encrypted and sent to a third-party arbitration institution. During the whole process, neither the third-party arbitration institution nor the business circle party or the partner can directly obtain the original data of the other party, but the obtained data after encryption or privacy processing, thus ensuring the security of the data. sex and privacy.

FIG. 4 is a schematic diagram of main modules of an apparatus for pushing information according to an embodiment of the present invention. As shown in FIG. 4 , the information push apparatus 400 according to the embodiment of the present invention mainly includes a user identification alignment module 401 , a user source determination module 402 , a customer group portrait calculation module 403 and a target user determination module 404 .

The user identification alignment module 401 is used to obtain a user identification set to be aligned, and perform identification alignment processing on the user identification set to be aligned and the business circle user identification set to obtain an aligned user identification;

A user source determination module 402, configured to obtain the movement trajectory data of the aligned user identifiers according to the business data, and determine the dense source of visiting users based on the movement trajectory data;

The customer group portrait calculation module 403 is used to calculate the customer group portraits of the visiting users of each dense source respectively according to the characteristic data of the visiting users;

The target user determination module 404 is configured to compare the similarity between the portrait of the customer group and the historical user portrait of the business district to determine a target user, and push information to the target user.

According to an embodiment of the present invention, the identification alignment processing includes: intersection processing, union processing, left outer join processing or right outer join processing.

According to another embodiment of the present invention, the user identification aligning module 401 may be further configured to: send the set of user identifications to be aligned to a third-party arbitration organization, so that the third-party arbitration organization can make the set of user identifications to be aligned according to the set of user identifications to be aligned. Perform logo alignment processing with the business circle user logo set.

According to another embodiment of the present invention, the apparatus 400 for pushing information further includes an identity set preprocessing module (not shown in the figure), which is used for:

Before performing the identification alignment process on the user identification set to be aligned and the business circle user identification set, obtain the processing rule of the user identification agreed with the business circle in advance, and the processing rule includes the data format and encryption method of the user identification;

The user identification set to be aligned is processed by using the processing rule, and the processed user identification set to be aligned is used as the user identification set to be aligned.

According to yet another embodiment of the present invention, the user origin determination module 402 may also be used to:

Determine the residence and work place of each visiting user based on the movement trajectory data;

Use the spatial clustering algorithm to cluster residence and work, and obtain multiple clustered areas;

From the plurality of clustering regions, a specified number of clustering regions where the visiting users are most densely distributed in space are selected as the dense source places of the visiting users.

According to yet another embodiment of the present invention, the target user determination module 604 can also be used for:

Category coding is performed on the structured portrait data in the customer group portrait and the historical user portrait in the business district respectively, and the portrait representation vector is obtained through dimensionality reduction;

Extracting social network representations from the unstructured portrait data in the customer group portraits and the historical user portraits of the business district, respectively, to obtain a network representation vector;

splicing the portrait representation vector and the network representation vector together to obtain the feature vector of the customer group portrait and the historical user portrait of the business district respectively;

Based on the similarity between the feature vectors, the similarity between the customer group portrait and the historical user portrait in the business district is compared, and the visiting user corresponding to the customer group portrait whose similarity meets a preset threshold is determined as the target user.

According to yet another embodiment of the present invention, the target user determination module 604 may also be used for:

Acquire the first dense source of the target user, and return the first dense source to the business circle, so that the business circle pushes information according to the first dense source.

According to the technical solution of the embodiment of the present invention, the user identification set to be aligned is obtained by obtaining the user identification set to be aligned, and the user identification set to be aligned and the business circle user identification set are subjected to identification alignment processing to obtain the aligned user identification; according to the business data, the movement trajectory data of the aligned user identification is obtained. , and based on the movement trajectory data to determine the dense source of visiting users; according to the characteristic data of the visiting users, calculate the customer group portraits of the visiting users in each dense source place; Compare the similarity to determine the target user, and push the target user information, which realizes the collection of multi-party data for user portrait characterization and information push, which makes up for the single dimension of user data in the business circle operation scenario. The source of unvisited users in the business district is fully explored, so as to better push advertisements and other information in the scenario of new customer drainage in the business district; , More real-time and more accurate characterization of user customer group portraits, more accurate identification of potential customers in the business district, and easy to improve the success rate of information push.

This technical solution first uses the time attribute of the trajectory to screen out the precise residence as the source of the user; then uses the spatial attribute, adopts the OPTICS density clustering algorithm and the similar user mining algorithm to realize the target new customer source through the existing passenger flow analysis The exploration and mining of potential target customers, to solve the problem of "who to promote, where to promote, how to promote, and when to promote", greatly improve the efficiency of drainage planning and marketing. When building the customer group portrait, the user features come from the business circle’s own label features, e-commerce label features, and social graph features of social platforms. The feature vector data of each party cannot be aggregated to one party to calculate the cosine similarity. Therefore, the homomorphic encryption technology is used to encrypt the results of different feature vectors of the three-party platforms through public key encryption and send them to a third-party arbitration institution. .

FIG. 5 shows an exemplary system architecture 500 to which the method for pushing information or the apparatus for pushing information according to an embodiment of the present invention can be applied.

As shown in FIG. 5 , the system architecture 500 may include terminal devices 501 , 502 , and 503 , a network 504 and a server 505 . The network 504 is a medium used to provide a communication link between the terminal devices 501 , 502 , 503 and the server 505 . Network 504 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

The user can use the terminal devices 501, 502, 503 to interact with the server 505 through the network 504 to receive or send messages and the like. Various communication client applications may be installed on the terminal devices 501 , 502 and 503 , such as shopping applications, web browser applications, search applications, advertising push tools, social platform software, etc. (just examples).

The terminal devices 501, 502, 503 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

The server 505 may be a server that provides various services, for example, a background management server that provides support for shopping websites browsed by the terminal devices 501 , 502 , and 503 (just an example). The background management server can obtain the user identification set to be aligned on the received information push request and other data, and perform identification alignment processing on the user identification set to be aligned and the business circle user identification set to obtain the aligned user identification; obtain the user identification according to the business data. Align the movement trajectory data of the user identification, and determine the dense source of visiting users based on the movement trajectory data; according to the characteristic data of the visiting users, calculate the customer group portraits of the visiting users of each dense source respectively; The similarity between the customer group portrait and the historical user portrait in the business district is compared to determine the target user, and the target user is processed such as information push, and the processing results (such as information push results, product information push results - only as an example) feedback to the terminal device.

It should be noted that the method for pushing information provided in the embodiment of the present invention is generally performed by the server 505 , and accordingly, the apparatus for pushing information is generally set in the server 505 .

It should be understood that the numbers of terminal devices, networks and servers in FIG. 5 are only illustrative. There can be any number of terminal devices, networks and servers according to implementation needs.

Referring to FIG. 6 below, it shows a schematic structural diagram of a computer system 600 suitable for implementing a terminal device or a server according to an embodiment of the present invention. The terminal device or server shown in FIG. 6 is only an example, and should not impose any limitations on the functions and scope of use of the embodiments of the present invention.

As shown in FIG. 6, a computer system 600 includes a central processing unit (CPU) 601, which can be loaded into a random access memory (RAM) 603 according to a program stored in a read only memory (ROM) 602 or a program from a storage section 608 Instead, various appropriate actions and processes are performed. In the RAM 603, various programs and data necessary for the operation of the system 600 are also stored. The CPU 601 , the ROM 602 , and the RAM 603 are connected to each other through a bus 604 . An input/output (I/O) interface 605 is also connected to bus 604 .

The following components are connected to the I/O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc. ; and a communication section 609 including a network interface card such as a LAN card, a modem, and the like. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I/O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 610 as needed so that a computer program read therefrom is installed into the storage section 608 as needed.

In particular, the processes described above with reference to the flowcharts may be implemented as computer software programs in accordance with the disclosed embodiments of the present invention. For example, embodiments disclosed herein include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the method illustrated in the flowchart. In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 609 and/or installed from the removable medium 611 . When the computer program is executed by the central processing unit (CPU) 601, the above-described functions defined in the system of the present invention are executed.

It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In the present invention, a 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 the present invention, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more logical functions for implementing the specified functions executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams or flowchart illustrations, and combinations of blocks in the block diagrams or flowchart illustrations, can be implemented in special purpose hardware-based systems that perform the specified functions or operations, or can be implemented using A combination of dedicated hardware and computer instructions is implemented.

The units or modules involved in the embodiments of the present invention may be implemented in a software manner, and may also be implemented in a hardware manner. The described unit or module can also be set in the processor, for example, it can be described as: a processor includes a user identification alignment module, a user source determination module, a customer group portrait calculation module and a target user determination module. The names of these units or modules do not constitute a limitation on the units or modules themselves, for example, the user ID alignment module can also be described as "used to obtain the set of user IDs to be aligned, and The user identification set to be aligned and the business circle user identification set are subjected to identification alignment processing to obtain a module for aligning user identifications.”

As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or may exist alone without being assembled into the device. The above-mentioned computer-readable medium carries one or more programs, and when the above-mentioned one or more programs are executed by a device, the device includes: acquiring a user identification set to be aligned, and combining the user identification set to be aligned with the quotient. Circle the user identification set to carry out identification alignment processing to obtain the aligned user identification; obtain the movement trajectory data of the aligned user identification according to the business data, and determine the dense source of visiting users based on the movement trajectory data; according to the characteristic data of the visiting users , respectively calculate the customer group portraits of the visiting users in each dense source area; compare the similarity between the customer group portraits and the historical user portraits of the business district to determine the target users, and push the information to the target users.

According to the technical solution of the embodiment of the present invention, the user identification set to be aligned is obtained by obtaining the user identification set to be aligned, and the user identification set to be aligned and the business circle user identification set are subjected to identification alignment processing to obtain the aligned user identification; according to the business data, the movement trajectory data of the aligned user identification is obtained. , and determine the intensive source of visiting users based on the movement trajectory data; according to the characteristic data of the visiting users, calculate the customer group portraits of the visiting users in each dense source place; Compare the similarity to determine the target user, and push the target user information, which realizes the collection of multi-party data for user portrait characterization and information push, which makes up for the single dimension of user data in the business circle operation scenario. The source of unvisited users in the business district is fully explored, so as to better push advertisements and other information in the scenario of new customer drainage in the business district; , More real-time and more accurate portrayal of user customer group portraits, more accurate identification of potential customers in the business district, and easy to improve the success rate of information push.

The above-mentioned specific embodiments do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. A method of information push, comprising:

acquiring a user identifier set to be aligned, and performing identifier alignment processing on the user identifier set to be aligned and a business district user identifier set to obtain an aligned user identifier;

obtaining moving track data of the aligned user identification according to service data, and determining a dense source of a visiting user based on the moving track data;

respectively calculating the guest group portraits of the visiting users of each dense source area according to the feature data of the visiting users;

and comparing the similarity of the customer group image with the historical user image of the business district to determine a target user, and pushing information of the target user.

2. The method of claim 1, wherein the identity alignment process comprises:

taking intersection processing, taking union processing, left outer connection processing or right outer connection processing.

3. The method of claim 1, wherein performing identifier alignment processing on the to-be-aligned user identifier set and a business district user identifier set comprises:

and sending the user identifier set to be aligned to a third party arbitration mechanism so that the third party arbitration mechanism carries out identifier alignment processing according to the user identifier set to be aligned and the business district user identifier set.

4. The method according to claim 1, further comprising, before performing identifier alignment processing on the to-be-aligned user identifier set and the business district user identifier set:

acquiring a processing rule of a user identifier agreed with a business circle in advance, wherein the processing rule comprises a data format and an encryption mode of the user identifier;

and processing the user identifier set to be aligned by using the processing rule, and taking the processed user identifier set to be aligned as the user identifier set to be aligned.

5. The method of claim 1, wherein determining a dense source of visiting users based on the movement trajectory data comprises:

determining a place of residence and a place of work for each visiting user based on the movement trajectory data;

clustering the residential areas and the workplaces by adopting a space clustering algorithm to obtain a plurality of clustering areas;

and selecting a specified number of clustering regions with the visiting users distributed most densely in the space from the plurality of clustering regions as dense sources of the visiting users.

6. The method of claim 1, wherein comparing the guest group imagery to a similarity of historical user imagery of a business district to determine a target user comprises:

respectively carrying out category coding on the guest group images and the structured image data in the historical user images of the business circles, and obtaining image characterization vectors through dimension reduction;

respectively carrying out social network representation extraction on the client group image and unstructured image data in the historical user images of the business circles to obtain network representation vectors;

splicing the portrait representation vector and the network representation vector together to respectively obtain the characteristic vectors of the customer group portrait and the historical user portrait of the business circle;

and based on the similarity between the feature vectors, comparing the similarity of the customer group portrait with the historical user portrait of the business circle, and determining the visiting user corresponding to the customer group portrait with the similarity meeting a preset threshold as the target user.

7. The method of claim 1, wherein pushing information to the target user comprises:

and acquiring a first dense source area of the target user, and returning the first dense source area to a business district so that the business district pushes information according to the first dense source area.

8. An information pushing apparatus, comprising:

the system comprises a user identifier alignment module, a service area user identifier alignment module and a service area user identifier alignment module, wherein the user identifier alignment module is used for acquiring a user identifier set to be aligned and performing identifier alignment processing on the user identifier set to be aligned and a business area user identifier set to obtain an aligned user identifier;

the user source determining module is used for acquiring the moving track data aligned with the user identifier according to the service data and determining the dense source of the visiting user based on the moving track data;

the guest group image calculation module is used for calculating the guest group images of the visiting users in each dense source area according to the feature data of the visiting users;

and the target user determining module is used for comparing the similarity of the customer group image and the historical user image of the business circle to determine a target user and pushing information of the target user.

9. An electronic device for pushing information, comprising:

one or more processors;

a storage device to store one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-7.

10. A computer-readable medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-7.

Images