Incremental cross-social-platform account matching method

Abstract

The invention provides an incremental cross-social-platform account matching method, and belongs to the technical field of cross-social-platform account matching. The method aims at solving the problem that a matched account of another social platform is found in a one-way mode based on account information of one social platform, accurate matching from a user ai in a set A of a social platform A to a corresponding account bj in a set B of a social platform B is achieved through a series of steps, and the set A lacks social information. However, the accounts in the set A belong to the same mechanism. The method comprises the following steps: firstly, acquiring two social user sets A and B to be matched by utilizing a network search technology, and then determining an initial account set P of a part of users in the set B as a reference point; and then, adding the account information of the association mechanism into the relational network as auxiliary information. On the basis, personal information, including user names, head portraits, geographic positions, personal brief introductions, social relations and other features, of the remaining users in the set B is extracted, and similarity analysis is conducted on the personal information and corresponding information of the users in the set A; once a new matched account pk is confirmed, the new matched account pk is added into the set P, and on this basis, the social contact between the remaining accounts in the set B and the set P is further improved. And then, the system continuously executes a similarity comparison iteration process until the growth rate of the newly-added matching account is reduced to be below a preset threshold value. According to the method, the problem that social relation information is insufficient possibly existing in the cross-social-platform account matching process is effectively solved, and the accuracy and efficiency of account matching between different social platforms are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cross-social platform account matching, and in particular to an incremental cross-social platform account matching method. Background Art

[0002] With the rapid development of the Internet, social platforms have become an important communication and interaction platform for contemporary people. While profoundly changing the way people communicate, they have accumulated a large amount of users' personal information, behavioral data and social relationships. These high-value data contain deep information that can be mined, providing a rich data foundation and a wide range of application scenarios for the development of technologies such as user profiling and behavioral modeling.

[0003] Based on this background, cross-social platform account matching technology has become one of the important research directions in the Internet field. This technology analyzes the attribute data, text content and association relationships of users on different social platforms, models and generates personalized user feature descriptions and behavior models, thereby achieving account matching and association identification. This technology has important practical application value in precision push, personalized services, recommendation systems, market research and public safety prevention and control.

[0004] However, the existing cross-social platform account matching technology mainly focuses on two-way matching, that is, finding corresponding accounts between different social platforms, and has not yet fully solved the problem of one-way discovery of matching accounts on another social platform based on account information on one social platform. Especially in the field of public safety prevention and control, one-way matching technology has significant practical significance, because newly matched accounts can often provide additional related information, further improving the depth and breadth of data analysis. Therefore, it is particularly necessary to develop a cross-social platform account matching technology that can achieve one-way matching in order to break through this limitation. Summary of the invention

[0005] The purpose of the present invention is to solve the problem of unidirectionally discovering matching accounts on another social platform based on account information on one social platform, and proposes a method for incremental account matching across social platforms. Through an iterative matching strategy, more confirmed account social relationships are obtained.

[0006] The technical solution proposed by the present invention includes the following parts:

[0007] S1. Collect user account information of various social platforms using network search technology;

[0008] S2. adding the account information of the associated institution as auxiliary information to the relationship network;

[0009] S3, analyze user account attributes and quantitatively evaluate the correlation across social platforms;

[0010] S4, integrating successfully matched accounts and their social relationships into the relationship network;

[0011] S5. The matching process is performed in multiple rounds until the growth rate of newly added matching accounts is lower than a preset threshold.

[0012] Furthermore, the step S1 comprises:

[0013] S11. For a set of personnel from a specific organization, first obtain their account information on platform A as set A;

[0014] S12, then use each of their account names on the A platform to search for account information with similar account names on the B platform as set B;

[0015] S13, the matching goal is to find accounts in set B that belong to users in set A, where set A lacks social information, but the accounts in set A all belong to the same organization;

[0016] S14. When obtaining account information, some clear accounts associated with set A and set B are found, and a definite account set P (P∈B) is obtained.

[0017] Furthermore, the step S2 comprises:

[0018] S21. Affiliated institutions are institutions of higher or lower levels or brother institutions that are closely related to the current institution;

[0019] S22, auxiliary information is to add the account social information of key personnel of the associated organization to the relationship network, so as to provide more social relationship information for the matching task;

[0020] S23. When the relationship network is initialized, it contains the initial social relationship information of the determined account. The relationship network is represented by G = (V, E, W), where V is the vertex set, E is the edge set, and W is the edge weight function. The formula of W is expressed as

[0021]

[0022] Furthermore, the step S3 comprises:

[0023] S31, extracting the following features of the account, namely, the account name, personal image identifier, geographic location information, and user self-description text;

[0024] S32, collecting the interaction patterns established by users in the network, including follow lists, fan groups, and two-way follow situations;

[0025] S33. Based on the data points obtained above, the consistency or similarity of the corresponding data points on different social platforms is compared by an algorithm to express the potential association between the accounts in a quantitative form;

[0026] S34. Introduce a similarity function S to measure the similarity between two accounts a(a∈A) and b(b∈B) under different features. Suppose we have n features, f i (a) represents the i-th feature of account a, f i (b) represents the i-th feature of account b, then the similarity can be expressed as:

[0027]

[0028] where w i is the weight of the ith feature, sim(·,·) is a similarity measurement function defined for a specific feature. For example, for strings (such as user names, occupations, and affiliated institutions), the edit distance can be used, for geographic locations, the distance between geographic coordinates can be used, and for personal image identification, a deep learning model can be used for measurement.

[0029] Furthermore, the step S4 comprises:

[0030] S41, obtain new confirmed account number P k social relationship information;

[0031] S42, the new confirmed account number P obtained after matching k Add to set P, and then update the remaining accounts in set B and determine the relationship network of account set P

[0032] G←Update(G,V,W,P old ,P new ) (3).

[0033] Furthermore, the step S5 comprises:

[0034] S51, re-evaluating and updating the social relationship attributes of the relevant accounts according to the updated social relationship network G;

[0035] S52. After each update, continue to perform multiple rounds of matching on the remaining accounts in set A and set B to continuously improve the social relationship network until the growth rate of newly added matching accounts is lower than the set threshold ε, and the matching is stopped. The growth rate is expressed as

[0036]

[0037] The present invention proposes a method for incremental account matching across social platforms, aiming to solve the technical problem that it is difficult to effectively discover and match accounts on other social platforms with account information on a single social platform. This method gradually determines the target account through iterative optimization, and uses the obtained account information to enrich the social relationship network of the account to be matched, thereby significantly improving the accuracy and efficiency of cross-platform account matching, and has strong practicality and innovation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 This is an incremental cross-social platform account matching method and steps;

[0040] Figure 2 The impact of incremental social information after iteration on the in-and-out degree of candidate accounts;

[0041] Figure 3 Iterative vertical incremental matching process of social information across social platform accounts. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0043] A cross-social platform account incremental matching method is designed to solve the technical problem that it is difficult to effectively discover and match accounts on other social platforms with account information on a single social platform. Figure 1 As shown, it includes the following four steps:

[0044] S1. Collect user account information of various social platforms using network search technology;

[0045] S2. adding the account information of the associated institution as auxiliary information to the relationship network;

[0046] S3. Analyze user account attributes and quantitatively evaluate the correlation across social platforms;

[0047] S4, integrating successfully matched accounts and their social relationships into the relationship network;

[0048] S5. The matching process is performed in multiple rounds until the growth rate of the newly added matching accounts drops below a set threshold.

[0049] In step S1, the user account information of each social platform is collected using network search technology. The specific process is as follows:

[0050] In this example, for a set of personnel from a specific organization, we first obtain their account information on platform A, as set A; then use each of their platform A account names to search for account information with similar account names on platform B, as set B; the matching goal is to find accounts in set B that belong to users in set A, where set A lacks social information but the accounts in set A all belong to the same organization; when obtaining account information, we find some clear associated accounts between set A and set B, and obtain a definite account set P (P∈B).

[0051] In step S2, the account information of the associated institution is added to the relationship network as auxiliary information. The specific process is as follows:

[0052] The associated organization is a subordinate organization or a brother organization that is closely related to the current organization. The auxiliary information is to add the account social information of the key personnel of the associated organization to the relationship network, so as to provide more social relationship information for the matching task. The relationship network contains the initial social relationship information of the determined account when it is initialized. The relationship network is represented by G = (V, E, W), where V is the vertex set, E is the edge set, and W is the edge weight function. The formula of W is expressed as

[0053]

[0054] In step S3, multiple attribute dimensions of each user account are collected and analyzed, such as account name, personal image identifier, geographic location information, and user self-description text; the interaction patterns established by users in the network are collected, including follow lists, fan groups, and two-way attention situations;

[0055] Based on the data points obtained above, the consistency or similarity of the corresponding data points on different social platforms is compared through algorithms to express the potential association between accounts in a quantitative form; a similarity function S is introduced to measure the similarity between two accounts a (a∈A) and b (b∈B) under different features. Assume that we have n features, f i (a) represents the i-th feature of account a, f i (b) represents the i-th feature of account b, then the similarity can be expressed as:

[0056]

[0057] where w iis the weight of the ith feature, sim(·,·) is a similarity measurement function defined for a specific feature. For example, for strings (such as user names, occupations, and affiliated institutions), the edit distance can be used, for geographic locations, the distance between geographic coordinates can be used, and for personal image identification, a deep learning model can be used for measurement.

[0058] In step S4, the successfully matched accounts and their social relationships are integrated into the relationship network, such as Figure 2 As shown, the specific process is as follows:

[0059] Get the new confirmed account P k The social relationship information of k Add to set P, and then update the remaining accounts in set B and determine the relationship network of account set P

[0060] G←Update(G,V,W,P old ,P new )(3).

[0061] In step S5, the matching process is performed in multiple rounds until the growth rate of newly added matching accounts is lower than a preset threshold, such as Figure 3 As shown, the specific process is as follows:

[0062] According to the updated social relationship network G, the social relationship characteristics of the relevant accounts are updated; after each update, multiple rounds of matching are continued for the remaining accounts in set A and set B to continuously improve the social relationship network until the growth rate of newly added matching accounts is lower than the set threshold ε, and the matching is stopped. The growth rate is expressed as

[0063]

[0064] Those skilled in the art may make modifications and combinations based on the principles and disclosures of the present invention without departing from the essence, which still fall within the protection scope of the present invention, and the embodiments are only provided for reference to understand the invention.

Claims

1. An incremental cross-social platform account matching method, characterized in that: The steps include: Use web search technology to collect user account information on various social platforms; Add the account information of the associated institution as auxiliary information to the relationship network; Analyze user account attributes and quantify the correlation across social platforms; Integrate successfully matched accounts and their social relationships into the relationship network; The matching process is performed in multiple rounds until the growth rate of newly matched accounts falls below a preset threshold.

2. The method according to claim 1, characterized in that The use of network search technology to collect user account information of various social platforms includes: For a set of personnel from a specific organization, first obtain their account information on platform A as set A; Then use each of their account names on platform A to search for account information with similar account names on platform B, as set B; The matching goal is to find accounts in set B that belong to users in set A, where set A lacks social information but all accounts in set A belong to the same organization; When obtaining account information, some clear accounts associated with set A and set B are found, and a definite account set P (P∈B) is obtained.

3. The method according to claim 1, characterized in that The step of adding the account information of the associated institution as auxiliary information to the relationship network includes: Affiliated institutions are institutions of higher or lower levels or brother institutions that are closely related to the current institution; Auxiliary information is to add the account social information of key personnel of the associated organization to the relationship network, thereby providing more social relationship information for the matching task; When the relationship network is initialized, it contains the initial social relationship information of the determined account. The relationship network is represented by G = (V, E, W), where V is the vertex set, E is the edge set, and W is the edge weight function. The formula of W is expressed as 4. The method according to claim 1, characterized in that The analysis of user account attributes and quantitative evaluation of cross-social platform correlations include: Collect and analyze multiple attribute dimensions of each user account, such as account name, personal image identifier, geolocation information, and user self-description text; Collect the interaction patterns established by users in the network, including follow lists, fan groups, and two-way attention; Based on the data points obtained above, the consistency or similarity of the corresponding data points on different social platforms is compared through algorithms to express the potential relationship between the accounts in a quantitative form; A similarity function S is introduced to measure the similarity between two accounts a(a∈A) and b(b∈B) under different features. Assume we have n features, f i (a) represents the i-th feature of account a, f i (b) represents the i-th feature of account b, then the similarity can be expressed as: where w i is the weight of the ith feature, sim(·,·) is a similarity measurement function defined for a specific feature. For example, for strings (such as user names, occupations, and affiliated institutions), the edit distance can be used, for geographic locations, the distance between geographic coordinates can be used, and for personal image identification, a deep learning model can be used for measurement.

5. The method according to claim 1, characterized in that The integration of successfully matched accounts and their social relationships into the relationship network includes: Get the new confirmed account P k social relationship information; The new confirmed account number P obtained after matching k Add to set P, and then update the remaining accounts in set B and determine the relationship network of account set P G←Update(G,V,W,P old ,P new )(3)。 6. The method according to claim 1, characterized in that The matching process is performed in multiple rounds until the growth rate of newly added matching accounts is lower than a preset threshold, including: Re-evaluate and update the social relationship attributes of the relevant accounts according to the updated social relationship network G; After each update, the remaining accounts in set A and set B are matched for multiple rounds to continuously improve the social relationship network until the growth rate of newly added matching accounts is lower than the set threshold ε, and the matching is stopped. The growth rate is expressed as

Images