Highly sensitive account evaluation method based on multi-dimensional model weighting

Abstract

The present invention relates to the technical field of social network technology, and specifically to a highly sensitive account evaluation method based on a multi-dimensional model weighting. When evaluating a website account, the present invention first obtains the number of social behaviors of the account to be evaluated on the website; then calculates the size of the initial sensitivity index of the account to be evaluated, calculates the influence of the account to be evaluated based on an improved HITS algorithm, establishes a final influence model of the account to be evaluated, adopts multi-dimensional weighting to calculate the sensitivity weight of the account to be evaluated, and finally calculates the final sensitivity of the account to be evaluated. When calculating the influence of the account to be evaluated, it is not necessary to iteratively calculate the user weight, so that the evaluation process is more efficient. In addition, when evaluating the website account, the present invention evaluates the sensitivity of the account to be evaluated based on the two dimensions of influence and activity of the account to be evaluated in the social network, so that the evaluation accuracy is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of social network technology, and in particular to a high-sensitivity account evaluation method based on multi-dimensional model weighting. Background Art

[0002] With the development of society, the internet has gradually become integrated into social life, bringing about tremendous changes in production and living habits throughout society. The internet has become an important part of people's lives. Leveraging China's continuous investment in internet infrastructure since 1989 and the new development perspectives and opportunities brought about by integration with the international internet, China's internet has experienced rapid growth since the beginning of the 21st century. The continuous growth of internet users has led to the formation of a massive online trading market. This has significantly increased demand for online resources, giving rise to a variety of assets, including online accounts.

[0003] Faced with the ever-expanding self-media account trading market, there's currently no clear method for accurately assessing the value of online accounts. This is partly due to the fact that self-media account transactions differ from physical asset transactions, making traditional methods for valuing online virtual assets inherently challenging and inadequate. Furthermore, due to a lack of case studies of self-media account transactions, it's difficult to construct a suitable valuation indicator system based on the factors influencing their value. Against this backdrop, exploring a suitable approach to self-media account valuation is crucial for self-media account trading.

[0004] In the prior art, the sensitivity of an account is mainly evaluated using a neural network model. For example, a Chinese invention patent (CN115391710A) discloses a method and device for evaluating the influence of a social media account based on a graph neural network. The method first constructs a knowledge graph and a graph neural network, replaces the feature vectors of the nodes in the graph neural network with the importance index of the knowledge graph, and aggregates the nodes into aggregated nodes. The attention mechanism and the importance index of the nodes are used to calculate the weights of the neighbors corresponding to the nodes; the centrality index of each node in the graph neural network is calculated, and the final prediction score of the node is calculated based on the centrality index and importance index of the node.

[0005] However, when evaluating the sensitivity of an account, the above scheme requires a large number of samples to train the neural network model, which will result in low evaluation efficiency. On the other hand, the above scheme considers fewer influencing factors during the evaluation, which will also affect the accuracy of the final sensitivity evaluation of the account.

[0006] Therefore, the existing technology urgently needs an account sensitivity assessment method that improves assessment accuracy and efficiency. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to address the shortcomings of the above technical solutions and provide a high-sensitivity account evaluation method based on multi-dimensional model weighting to improve the efficiency and accuracy of account sensitivity evaluation.

[0008] To achieve the above objectives, according to one aspect of the present invention, a method for evaluating highly sensitive accounts based on multi-dimensional model weighting is provided, comprising the following steps:

[0009] Step 1: Obtain the number of social behaviors of the account to be evaluated on the website;

[0010] Step 2: Calculate the initial sensitivity index of the account to be evaluated;

[0011] Step 3: Calculate the influence of the account to be evaluated based on the improved HITS algorithm;

[0012] Step 4: Establish the final influence model of the account to be evaluated;

[0013] Step 5: Use multi-dimensional weighting to calculate the sensitivity weight of the account to be evaluated;

[0014] Step 6: Calculate the final sensitivity of the account to be evaluated.

[0015] Preferably, the social behavior includes replying, commenting, liking, etc.

[0016] Furthermore, the replies include replies from users to fans or friends, as well as replies from other users to the user of the evaluated account; the comments include comments from users to fans or friends, as well as comments from other users to the user of the evaluated account; the likes include likes from users to fans or friends, as well as likes from other users to the user of the evaluated account;

[0017] Preferably, the initial sensitivity of the account to be evaluated on the website is related to three types of behavior: replying, commenting, and liking. The specific calculation formula is as follows:

[0018]

[0019] Where S is the initial sensitivity index of the account to be evaluated, R s , C s , L s The number of times users reply, comment, and like posts for sensitive content.

[0020] Based on the shortcomings of the existing technology, the present invention improves the HITS algorithm. By adjusting the evaluation logic, it is assumed that the main reason why a reply content attracts attention is the user influence of the source content itself and the neighbor publishers in the dissemination network, and has almost nothing to do with the indirect forwarders or the original author. In other words, the influence of the account to be evaluated does not form an iterative relationship during the content dissemination process. Therefore, when the improved HITS algorithm is used to calculate the user weight relationship, iterative calculation of user weights is no longer required. The improved HITS algorithm that omits iterative calculation is not only more reasonable, but also greatly reduces the algorithm's convergence time. Finally, the optimized formula is:

[0021]

[0022] Among them, a i is the influence of the account to be evaluated, user j is the fan of user i, F j represents the number of fans of user j, n j is the number of friends of user j;

[0023] It is worth emphasizing that the influence of the account to be evaluated in step 3 reflects the influence of the account to be evaluated from the perspective of the quality of fans;

[0024] Preferably, the final influence model of the account to be evaluated proposed in the present invention is composed of the number of user fans of the account to be evaluated and the weighted quality of the fans:

[0025] W i =N i +μa i

[0026] Among them, W i Represents the final influence of the account to be evaluated, N i represents the number of followers of user i; the parameter μ is used to adjust the weight of the two parts, here it is set to 0.5, and ai is the influence of the account to be evaluated;

[0027] It is worth emphasizing that, through step 4, when evaluating the sensitivity of the account to be evaluated, a multi-dimensional evaluation of the quality and number of fans is adopted.

[0028] Preferably, the sensitivity weight of the account to be evaluated integrates the user influence and user activity characteristics, and is composed of the weighted user influence and user activity; the formula is as follows:

[0029] W=W i +λ·W a

[0030] Where W is the sensitivity weight of the account to be evaluated, Wi is the activity of the account to be evaluated, which is obtained from the social behavior statistics of the account to be evaluated, and W aThe parameter λ is used to adjust the balance between the weights of the two parts to represent the activity of the account to be evaluated.

[0031] Preferably, the final calculation formula for the sensitivity of the account to be evaluated is as follows:

[0032] S c =S·W

[0033] Among them, S is the initial sensitivity index of the account to be evaluated, S c is the final sensitivity of the account to be evaluated, and W is the sensitivity weight of the account to be evaluated.

[0034] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a data processing program is stored. The data processing program is executed by a processor to implement the above-mentioned high-sensitivity account evaluation method based on multi-dimensional model weighting.

[0035] Based on the above technical solution, the highly sensitive account assessment method based on multi-dimensional model weighting provided by this application has the following technical effects:

[0036] When evaluating website accounts, the present invention improves the HITS algorithm and adjusts the evaluation logic. It assumes that the reason a reply attracts attention is mainly due to the user influence of the neighboring publishers in the dissemination network, and has almost nothing to do with indirect forwarders or the original author. The influence of the account to be evaluated does not form an iterative relationship during the content dissemination process. Therefore, when the improved HITS algorithm is used to calculate the user weight relationship, it is no longer necessary to iteratively calculate the user weight, making the evaluation process more efficient.

[0037] In addition, when evaluating website accounts, the present invention evaluates the sensitivity of the account to be evaluated based on the two dimensions of influence and activity of the account to be evaluated in the social network, thereby making the evaluation more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. 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 any creative work.

[0039] Figure 1 A flow chart of a method for evaluating highly sensitive accounts based on a multi-dimensional model weighting provided in an embodiment of the present application;

[0040] Figure 2 Schematic diagram of traditional HITS algorithm calculation. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0042] The following first describes the concepts involved in this application with reference to the accompanying drawings. It should be noted that the following description of each concept is only for the purpose of making the content of this application easier to understand and does not limit the scope of protection of this application.

[0043] Example 1: Figure 1 As shown, a highly sensitive account evaluation method based on multi-dimensional model weighting includes the following steps:

[0044] Step 1: Obtain the number of social behaviors of the account to be evaluated on the website;

[0045] Specifically, the social behaviors include replying, commenting, liking, etc.

[0046] Furthermore, the replies include replies from users to fans or friends, as well as replies from other users to the user of the evaluated account; the comments include comments from users to fans or friends, as well as comments from other users to the user of the evaluated account; the likes include likes from users to fans or friends, as well as likes from other users to the user of the evaluated account;

[0047] Step 2: Calculate the initial sensitivity index of the account to be evaluated;

[0048] Specifically, the initial sensitivity of the account to be evaluated on the website is related to three types of behavior: replying, commenting, and liking. The specific calculation formula is as follows:

[0049]

[0050] Where S is the initial sensitivity index of the account to be evaluated, R s , C s , L s The number of times users reply, comment, and like posts for sensitive content.

[0051] Step 3: Calculate the influence of the account to be evaluated based on the improved HITS algorithm;

[0052] Traditionally, in the existing technology, the HITS algorithm is used to calculate the influence of an account. The algorithm is applied to calculate the weight of web pages on the Internet. Specifically, Figure 2As shown in the figure, when a user connects to web page B through web page A, and there is a link to web page C in web page B, if there is a strong topic correlation between web pages A, B, and C, then the account to be evaluated will have a high probability of browsing web pages B and C through web page A. In this case, web page A will have an impact on web pages B and C at the same time. The HITS algorithm works by evaluating the influence of user accounts based on the user's click behavior on web pages. Among them, the evaluation is mainly carried out through two parameters: center value and authority value. The center value represents the aggregation ability of the node in the network, and the authority value represents the information collection ability of the node in the network. The center value and authority value are a mutually reinforcing dependency. A high center value web page will point to many high authority value web pages, and a high authority value web page will be pointed to by many high center value web pages. The above figure shows the user account relationship.

[0053] When applying the HITS algorithm to a website, the authority value reflects the user's authority and can represent their influence. The center value reflects the information disseminator and can represent the user's communication power. However, the interaction method and frequency between users will affect user relationships. When calculating user influence using the HITS algorithm, the user evaluation weights must be iteratively calculated until the results are relatively stable and convergent. However, the ranking order of users usually changes after each iteration, so reaching convergence is a particularly slow process.

[0054] In view of the above shortcomings, this embodiment improves the HITS algorithm by adjusting the evaluation logic. It assumes that the main reason why a reply content attracts attention is the user influence of the neighboring publishers in the dissemination network, and has almost nothing to do with the indirect forwarders or the original author. In other words, the influence of the account to be evaluated does not form an iterative relationship during the content dissemination process. Therefore, when using the improved HITS algorithm to calculate the user weight relationship, iterative calculation of user weight is no longer required. The improved HITS algorithm that omits iterative calculation is not only more reasonable, but also greatly reduces the algorithm's convergence time. Finally, the optimized formula is:

[0055]

[0056] Among them, a i is the influence of the account to be evaluated, user j is the fan of user i, F j represents the number of fans of user j, n j is the number of friends of user j;

[0057] It is worth emphasizing that the influence of the account to be evaluated in step 3 reflects the influence of the account to be evaluated from the perspective of the quality of fans;

[0058] Step 4: Establish the final influence model of the account to be evaluated;

[0059] The HITS algorithm, which omits the iteration process, unifies the hub weight and authority weight of a node into a single authority weight, where the authority weight is obtained by summing the in-degree and out-degree values ​​of all neighboring nodes pointing to the node. Therefore, the final influence model of the account to be evaluated proposed in this embodiment is composed of the number of user followers of the account to be evaluated and the weighted quality of the followers:

[0060] W i =N i +μa i

[0061] Among them, W i Represents the final influence of the account to be evaluated, N i represents the number of followers of user i; the parameter μ is used to adjust the weight of the two parts, here it is set to 0.5, and ai is the influence of the account to be evaluated;

[0062] It is worth emphasizing that, through step 4, when evaluating the sensitivity of the account to be evaluated, a multi-dimensional evaluation of the quality and number of fans is adopted.

[0063] Step 5: Use multi-dimensional weighting to calculate the sensitivity weight of the account to be evaluated;

[0064] Specifically, the sensitivity weight of the account to be evaluated combines the user influence and user activity characteristics, and is composed of the weighted user influence and user activity; the formula is as follows:

[0065] W=W i +λ·W a

[0066] Where W is the sensitivity weight of the account to be evaluated, Wi is the activity of the account to be evaluated, which is obtained from the social behavior statistics of the account to be evaluated, and W a The parameter λ is used to adjust the balance between the weights of the two parts to represent the activity of the account to be evaluated.

[0067] Step 6: Calculate the final sensitivity of the account to be evaluated;

[0068] Specifically, the final calculation formula for the sensitivity of the account to be evaluated is as follows:

[0069] S c =S·W

[0070] Among them, S is the initial sensitivity index of the account to be evaluated, S c is the final sensitivity of the account to be evaluated, and W is the sensitivity weight of the account to be evaluated.

[0071] Embodiment 2: This embodiment includes a computer-readable storage medium on which a data processing program is stored. The data processing program is executed by a processor to implement a high-sensitivity account evaluation method based on multi-dimensional model weighting in embodiment 1.

[0072] It will be understood by those skilled in the art that the embodiments herein may be provided as methods, devices (equipment), or computer program products. Therefore, this document may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Including but not limited to RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device, or any other medium that can be used to store desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and may include any information delivery medium.

[0073] This document is described with reference to flowcharts and / or block diagrams of methods, apparatus (devices) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0074] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0075] The embodiments and / or implementation methods described above are only used to illustrate the preferred embodiments and / or implementation methods for realizing the technology of the present invention, and do not impose any form of limitation on the implementation methods of the technology of the present invention. Any person skilled in the art may make slight changes or modifications to other equivalent embodiments without departing from the scope of the technical means disclosed in the content of the present invention, but they should still be regarded as technologies or embodiments that are essentially the same as the present invention.

Claims

1. A highly sensitive account assessment method based on multi-dimensional model weighting, characterized by: The following steps are involved: Step 1: Obtain the number of social behaviors of the account to be evaluated on the website; Step 2: Calculate the initial sensitivity index of the account to be evaluated. The initial sensitivity of the account to be evaluated on the website is related to three types of behavior: replying, commenting, and liking. The specific calculation formula is as follows: Where S is the initial sensitivity index of the account to be evaluated, R s , C s , L s The number of times users reply, comment, and like posts containing sensitive content; Step 3: Calculate the influence of the account to be evaluated based on the improved HITS algorithm; The step 3 is specifically as follows: by adjusting the evaluation logic, the influence formula of the account to be evaluated is obtained as follows: in, is the influence of the account to be evaluated, user j is the fan of user i, F j represents the number of fans of user j, n j is the number of friends of user j; Step 4: Establish the final influence model of the account to be evaluated; The final influence model of the account to be evaluated is composed of the number of user fans of the account to be evaluated and the weighted quality of the fans: Among them, W i Represents the final influence of the account to be evaluated, N i Represents the number of fans of user i; the parameter μ is used to adjust the weights of the two parts. The influence of the account to be evaluated; Step 5: Use multi-dimensional weighting to calculate the sensitivity weight of the account to be evaluated; In step 5, the sensitivity weight of the account to be evaluated combines the user influence and user activity characteristics, and is composed of the weighted user influence and user activity; the formula is as follows: Among them, W is the sensitivity weight of the account to be evaluated, W i is the final influence of the account to be evaluated, The parameter λ is used to adjust the balance between the weights of the two parts. Step 6: Calculate the final sensitivity of the account to be evaluated; In step 6, the final sensitivity of the account to be evaluated S C The calculation formula is: S c =S·W Among them, S is the initial sensitivity index of the account to be evaluated, and W is the sensitivity weight of the account to be evaluated.

2. The method for evaluating highly sensitive accounts based on multi-dimensional model weighting according to claim 1 is characterized in that: The social behaviors include replying, commenting, and liking.

3. The highly sensitive account evaluation method based on multi-dimensional model weighting according to claim 2 is characterized in that: The replies include replies from users to fans or friends, and also include replies from other users to the evaluated account users. The comments include comments from users to fans or friends, and also include comments from other users to the evaluated account users. The likes include likes from users to fans or friends, and also include likes from other users to the evaluated account users.

4. The method for evaluating highly sensitive accounts based on multi-dimensional model weighting according to claim 1, characterized in that: The value of μ is 0.5.

Images