An account evaluation method, device and system

Abstract

The application discloses a kind of account evaluation method, device and system.Method includes: obtaining the prior feature of to be evaluated account, prior feature is used to characterize the content data output characteristic of to be evaluated account in self-media platform, according to prior feature carries out feature fusion, obtains the prior evaluation result of to be evaluated account, according to the posterior feature of to be evaluated account carries out feature fusion, obtains the posterior evaluation result of to be evaluated account, according to prior evaluation result and posterior evaluation result, obtains the prior-posterior linkage evaluation result of to be evaluated account, for self-media platform carries out content recommendation.The technical scheme can evaluate account by the ability of to be evaluated account in platform to recommend high-quality content reflected by influence feature and potential feature in posterior feature, so the evaluation result is more accurate.In addition, it is strongly dependent on artificial experience and artificial power, and the account evaluation efficiency can be significantly improved in the self-media platform with massive data.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an account evaluation method, apparatus and system. Background Technology

[0002] With the rapid development of the internet and the widespread adoption of mobile applications, the entire media era is undergoing rapid transformation, transitioning from the traditional media era dominated by print media, radio, and television to the new media era of mobile social networking. In this new media era, platforms that allow users to voice their opinions, share, express their opinions, and disseminate information are called self-media platforms. Self-media platforms typically display content through recommended information feeds.

[0003] For self-media platforms that display content through recommendation feeds, the accounts are a mixed bag, with varying levels of quality in their content. To ensure the healthy operation of these platforms, it is necessary to evaluate the accounts. Currently, the accuracy of account evaluation methods is insufficient; therefore, improvements are needed to enhance the accuracy of account evaluation. Summary of the Invention

[0004] This application provides an account evaluation method, apparatus, system, device, computer-readable storage medium, and computer program product to improve the efficiency of account evaluation and the accuracy of evaluation results.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] Firstly, this application provides an account evaluation method, including:

[0007] Obtain the prior features of the account to be evaluated, which are used to characterize the content output characteristics of the account to be evaluated on the self-media platform;

[0008] Based on the prior features, feature fusion is performed to obtain the prior evaluation result of the account to be evaluated;

[0009] The posterior features of the account to be evaluated are fused to obtain the posterior evaluation result of the account to be evaluated. The posterior features include influence features and potential features. The influence features are determined based on the follow and unfollow data of the account to be evaluated on the self-media platform. The potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated.

[0010] Based on the prior evaluation results and the posterior evaluation results, the prior and posterior linked evaluation results of the account to be evaluated are obtained, and the prior and posterior linked evaluation results are used by the self-media platform for content recommendation.

[0011] Secondly, this application provides an account evaluation device, comprising:

[0012] A priori feature acquisition unit is used to acquire priori features of the account to be evaluated, wherein the priori features are used to characterize the content output characteristics of the account to be evaluated on the self-media platform.

[0013] A prior feature fusion unit is used to perform feature fusion based on the prior features to obtain the prior evaluation result of the account to be evaluated.

[0014] The posterior feature fusion unit is used to perform feature fusion based on the posterior features of the account to be evaluated to obtain the posterior evaluation result of the account to be evaluated; the posterior features include influence features and potential features, wherein the influence features are determined based on the follow and unfollow operation data of the account to be evaluated on the self-media platform, and the potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated.

[0015] The evaluation result fusion unit is used to obtain the prior and subsequent evaluation results of the account to be evaluated based on the prior evaluation results and the subsequent evaluation results. The prior and subsequent evaluation results are used by the self-media platform for content recommendation.

[0016] Thirdly, this application provides an account evaluation system, including: an account evaluation device, a self-media account level database, a content database, and a statistical reporting interface server;

[0017] The account assessment device is used to perform the steps of the account assessment method described in the first aspect above;

[0018] The content database is used to provide the account evaluation device with metadata of the content produced by the account to be evaluated, so that the account evaluation device can obtain the prior characteristics of the account to be evaluated based on the metadata.

[0019] The statistical reporting interface server is used to provide the account evaluation device with statistical data of the account to be evaluated, so that the account evaluation device can obtain the posterior features of the account to be evaluated based on the statistical data.

[0020] The self-media account level database is used to receive the pre- and post-evaluation linkage evaluation results of the account to be evaluated, which are written to or updated by the account to be evaluated.

[0021] Fourthly, this application provides an account evaluation device, which includes a processor and a memory:

[0022] The memory is used to store program code and transmit the program code to the processor;

[0023] The processor is used to execute the steps of the account evaluation method provided in the first aspect according to the instructions in the program code.

[0024] Fifthly, this application provides a computer-readable storage medium for storing program code for executing any of the implementation methods of the account evaluation method provided in the first aspect above.

[0025] Sixthly, this application provides a computer program product including instructions that, when run on a computer, cause the computer to execute any of the implementation methods of the account evaluation method provided in the first aspect above.

[0026] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0027] The account evaluation method provided in this application first obtains the prior features of the account to be evaluated, performs feature fusion based on the prior features to obtain the prior evaluation result of the account to be evaluated, and performs feature fusion based on the posterior features of the account to be evaluated to obtain the posterior evaluation result of the account to be evaluated. The posterior features include influence features and potential features. Then, based on the prior evaluation result and the posterior evaluation result, a combined prior and posterior evaluation result of the account to be evaluated is obtained and used for content recommendation on the self-media platform. Because this method not only uses prior features as the evaluation basis, but also incorporates posterior features that can dynamically reflect the account's ability to recommend high-quality content into the evaluation basis, where the influence feature is determined based on the follow and unfollow operation data of the account to be evaluated on the self-media platform, and the potential feature is determined based on the trend changes of various interaction types of data corresponding to the account to be evaluated, it can effectively refer to the feedback of content consumers on the content produced by the account, and more accurately reflect the account's ability to recommend high-quality content on the platform in the final combined prior and posterior evaluation result. Furthermore, both the aforementioned prior and posterior features can be acquired in real time through automation, and the feature fusion and evaluation result fusion operations can also be completed without human intervention or experience. Therefore, this method can significantly improve the efficiency of account evaluation on self-media platforms with massive amounts of data. Attached Figure Description

[0028] Figure 1 A flowchart illustrating an account evaluation method provided in this application embodiment;

[0029] Figure 2 A schematic diagram of a tag for account-generated content provided in an embodiment of this application;

[0030] Figure 3 This is a schematic diagram of the structure of an account evaluation device provided in an embodiment of this application;

[0031] Figure 4 This is a schematic diagram of the structure of an account evaluation system provided in an embodiment of this application;

[0032] Figure 5 This is a schematic diagram of the structure of another account evaluation system provided in an embodiment of this application;

[0033] Figure 6 This is a schematic diagram of the server structure in an embodiment of this application;

[0034] Figure 7 This is a schematic diagram of the structure of a terminal device in an embodiment of this application. Detailed Implementation

[0035] Accounts that produce content for platform users to consume on self-media platforms are called self-media accounts. Platforms need to recommend as much fresh, high-quality content produced by self-media accounts as possible to users to maintain the platform's healthy operation. With the continuous creation of accounts and content production within the platform, self-media platforms now need to evaluate these accounts to better facilitate content recommendations in subsequent stages. Account evaluation methods can involve operators assessing accounts based on prior information. For example, operators might determine the account's level based on its industry reputation, performance on other platforms, and their personal operational experience, and then periodically review the accounts using a combination of manual inspections and data summarizing exposure from top accounts to update the account level. However, with a massive number of accounts, manual account evaluation is obviously inefficient and heavily reliant on the experience of manual operators. Furthermore, because account content publishing behavior is dynamic, the cycle for manually updating account evaluation results is often long. The above evaluation method, relying solely on prior information, cannot reflect these changes in the evaluation results in a timely and accurate manner, thus hindering the platform's recommendation of high-quality content to some extent.

[0036] This application provides an account evaluation method, apparatus, and system to improve the efficiency of account evaluation and the accuracy of evaluation results.

[0037] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] The account evaluation method provided in this application can be applied to devices with data transmission capabilities, such as servers or terminals. Specifically, terminal devices can include smartphones, desktop computers, laptops, tablets, smart voice interaction devices, smart home appliances, vehicle terminals, and aircraft. Servers can be independent physical servers, server clusters composed of multiple physical servers, or distributed systems. Furthermore, servers can also be cloud servers providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. In the method described below, the executing entity can be either a terminal device or a server.

[0039] The terminal devices mentioned above, as well as the user terminals involved in the methods described below, such as content consumption terminals and content production terminals, include, but are not limited to, mobile phones, computers, smart voice interaction devices, smart home appliances, vehicle terminals, and aircraft.

[0040] The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be connected directly or indirectly through wired or wireless communication methods, which is not limited herein.

[0041] Cloud computing refers to the delivery and usage model of IT infrastructure, meaning obtaining necessary resources in an on-demand and easily scalable manner through a network. In a broader sense, cloud computing also refers to the delivery and usage model of services, meaning obtaining necessary services in an on-demand and easily scalable manner through a network. These services can be IT and software related, internet-related, or other services. Cloud computing is a product of the development and integration of traditional computer and network technologies such as grid computing, distributed computing, parallel computing, utility computing, network storage technologies, virtualization, and load balancing.

[0042] In this application's embodiments, all user-related data is acquired with the user's authorization or consent. For example, before analyzing data based on a user's account, a pop-up window or notification message is used to inform the user of the subsequent application of the data. Upon receiving the user's "confirmation" or "agreement" response to the notification, it is determined that the user has authorized or consented to the use of the data in a manner consistent with the notification.

[0043] Figure 1 A flowchart of an account evaluation method provided in this application embodiment is shown below. Figure 1 As shown, the account evaluation method provided in this embodiment includes the following steps S101-S104:

[0044] S101, Obtain the prior features of the account to be evaluated.

[0045] In this embodiment, when obtaining the prior features of the account to be evaluated, the prior features of the account to be evaluated can be obtained by acquiring the original flow information of the content produced by the account to be evaluated and the metadata of the content produced by the account to be evaluated. The prior features are used to characterize the content output characteristics of the account to be evaluated on the self-media platform.

[0046] When obtaining the posterior features of an account to be evaluated, information such as the interactive behavior of content consumers when consuming content and the information of the account to be evaluated can be obtained from the statistical data of the account to be evaluated. Then, the posterior features of the account to be evaluated can be obtained from these. The posterior features represent the degree of contribution of the account to the self-media platform and the recognition of the account by platform users.

[0047] S102, perform feature fusion based on prior features to obtain the prior evaluation result of the account to be evaluated.

[0048] After obtaining the prior features of the account to be evaluated, the prior features can be fused according to the fusion method to obtain the prior evaluation result, which can be expressed in numerical form.

[0049] S103, perform feature fusion based on the posterior features of the account to be evaluated to obtain the posterior evaluation result of the account to be evaluated.

[0050] In this application embodiment, the posterior features include at least the influence features and potential features of the account to be evaluated.

[0051] The influence characteristic can be determined based on the follow and unfollow (i.e., unfollow) data of the account to be evaluated on the self-media platform. Since the influence characteristic is determined by the aforementioned follow and unfollow data, using the follow and unfollow operations of the account on the platform as the data guide, this influence characteristic can reflect the influence of the account to be evaluated on the operational effectiveness of the self-media platform and its influence on platform users. Because follow and unfollow operations are often time-sensitive, such as being performed at a certain time, these time-attributed operations can also dynamically reflect the influence of the account on the platform.

[0052] Among these, potential characteristics can be determined based on the changing trends of various interaction types for the account being evaluated. By analyzing these interaction data trends, the potential characteristics of the account being evaluated can be dynamically analyzed to reflect the potential of the content output by the account to attract platform users to interact. Compared to evaluating an account based solely on static interaction data at a certain moment, dynamic trends can more comprehensively and specifically highlight the real-time characteristics of the content published by the account that attract user interaction.

[0053] That is, in this embodiment, the influence feature is determined through time-related operational data, realizing a dynamic correlation between the influence feature and account evaluation. Additionally, the potential feature is determined through time-related methods, realizing a dynamic correlation between the potential feature and account evaluation. The influence and potential features obtained in this way can help improve the accuracy and reliability of the posterior evaluation results, thereby improving the accuracy and reliability of the prior- and posterior-linked evaluation results. Content recommendations based on the prior- and posterior-linked evaluation results not only benefit the platform and its users but also the use and development of the account to be evaluated. S104, Based on the prior and posterior evaluation results, the prior- and posterior-linked evaluation results of the account to be evaluated are obtained.

[0054] After obtaining the prior and posterior evaluation results, they can be fused according to the fusion method to obtain the combined prior and posterior evaluation results of the account to be evaluated. These combined prior and posterior evaluation results can be used for content recommendation on self-media platforms. The processes of obtaining the prior and posterior features of the account to be evaluated, calculating the prior, posterior, and combined prior and posterior evaluation results, can be automated, for example, implemented separately by computer programs, or by pre-training corresponding models and calling them. Alternatively, the above processes can be implemented within a single computer program or a single training model; this embodiment does not impose any limitations on this.

[0055] The account evaluation method provided in this application first obtains the prior features of the account to be evaluated, performs feature fusion based on the prior features to obtain the prior evaluation result of the account to be evaluated, and performs feature fusion based on the posterior features of the account to be evaluated to obtain the posterior evaluation result of the account to be evaluated. The posterior features include influence features and potential features. Then, based on the prior evaluation result and the posterior evaluation result, a combined prior and posterior evaluation result of the account to be evaluated is obtained and used for content recommendation on the self-media platform. Since this method not only uses prior features as the evaluation basis, but also incorporates posterior features that can dynamically reflect the account's ability to recommend high-quality content into the evaluation basis, it can effectively refer to the feedback of content consumers on the content produced by the account, thereby more accurately reflecting the account's ability to recommend high-quality content on the platform in the final combined prior and posterior evaluation result. In addition, the above-mentioned prior features and posterior features can be obtained in real time through automated means, and the feature fusion and evaluation result fusion operations can also be completed without human effort or experience. Therefore, this method can significantly improve the efficiency of account evaluation on self-media platforms with massive amounts of data.

[0056] In one embodiment, the posterior features of the account to be evaluated may include content interaction features in addition to influence features and potential features. Content interaction features can be determined based on the public content and interaction data of the account within a preset time range, reflecting the engagement of the works published by the account with users. When performing feature fusion on the posterior features of the account to be evaluated, the influence features, content interaction features, and potential features of the account can be obtained. These features are then fused according to a preset posterior feature fusion method to obtain the posterior evaluation result. The preset time range can be any length not less than one day; for example, it could be one month or one week. The length of the preset time range can be set according to actual needs and is not specifically limited here.

[0057] In this embodiment, the influence characteristics of the account to be evaluated can be reflected by the number of followers of the account, which indicates the account's influence on the self-media platform. By determining the influence characteristics of the account to be evaluated, self-media accounts with a larger total number of followers, more recent follower activity, and fewer recent unfollower activity can ultimately obtain better posterior evaluation results, thereby improving the platform's ability to recommend high-quality content.

[0058] The method for calculating the influence characteristics of the account to be evaluated can be as shown in formula (1):

[0059]

[0060] Among them, Ai This represents the latest total number of followers for the i-th account; These represent the number of objects in the first object set that follow the i-th account and the number of objects in the second object set that unfollow the i-th account, respectively. t0 represents the dates when object u follows and unfollows the i-th account, respectively; t0 is the current date. Let be the weights of the i-th account followed and unfollowed by object u, respectively, calculated using the formulas (1-1) and (1-2):

[0061]

[0062]

[0063] P represents the number of accounts in the sets of accounts that object u follows and unfollows, respectively; + P - These represent the number of accounts in the sets of accounts that have been followed and unfollowed, respectively.

[0064] In formula (1), δ and η are parameters that play a smoothing role, and can be δ = 1.0 and η = 10.0 respectively. α is a control parameter, and can be α = 2; where the values ​​of δ, η, and α can be other values ​​according to actual needs, and this embodiment does not impose any restrictions on this. When i takes different values, it means that the account to be evaluated has changed. In one example, a set of accounts to be evaluated can be established, and all accounts contained in this set are accounts to be evaluated. To facilitate the differentiation of different accounts, a corresponding identifier ordinal number is assigned to each account to be evaluated in this set, such as 1, 2, 3, etc. Thus, as an example, A i This represents the latest total number of followers for the i-th account in the set of accounts to be evaluated.

[0065] Specifically, when calculating the influence characteristic I of the account to be evaluated using the above formula (1), the latest total number of followers A of the account to be evaluated can be obtained based on the follow and unfollow operation data of the account on the self-media platform. i And the number of objects in the first set of objects to follow and the second set of objects to unfollow for the accounts to be evaluated. Calculate the attention weight of each object in the first object set for the account to be evaluated. For example, the attention weight of object u in the first object set for the i-th account. Then, based on the attention weight of each object in the first object set to the account to be evaluated, and the date on which each object in the first object set paid attention to the account to be evaluated... And the current date t0, calculate the first operation result corresponding to the object's attention behavior.

[0066] Similarly, calculate the unfollow weight for each object in the second object set on the account to be evaluated. For example, the unfollow weight of object u on the i-th account in the second object set. Then, based on the unfollow weight of each object in the second object set for the account to be evaluated, and the date on which each object in the second object set unfollowed the account to be evaluated... And the current date t0, calculate the second operation result corresponding to the object unfollowing behavior. In one alternative approach, the latest total number of followers A of the account to be evaluated is obtained from the account information to be evaluated. i At the same time, some filtering conditions can be added, such as only including accounts with an activity level greater than a preset condition in the latest total number of followers A. i In the middle, and / or, accounts that are considered "reposting" accounts or have low activity levels will not be included in the latest total number of accounts followed (A). i This improves the accuracy of the latest total number of followers data and reduces interference. When obtaining the first set of objects from the accounts to be evaluated, only objects that have been read or viewed two or more times within a preset time range can be included in the first set, thus avoiding accidental clicks on objects of the accounts to be evaluated. The preset time range can be any length of at least one day, such as one month or one week. The length of the preset time range can be set according to actual needs, and there are no specific restrictions here. The above method can improve the accuracy of calculating the influence characteristic I of the accounts to be evaluated.

[0067] Finally, the influence characteristic I of the account to be evaluated is obtained based on the latest total number of followers, the first calculation result, and the second calculation result. It should be noted that the set of objects mentioned above can be a set containing user identification information. For example, after a user registers an account on the platform, to distinguish different accounts, the username can be used as identification information, or a specific identification information can be assigned to each user account and applied one-to-one. As an example, the objects in the first set of objects represent user accounts that follow the account to be evaluated; the objects in the second set of objects identify user accounts that unfollow the account to be evaluated.

[0068] It should be noted that the above is only one example of how to obtain influence characteristics. In other implementation scenarios, the influence characteristics of the account to be evaluated can also be calculated based on changes in the number of daily active users (DAU) on the self-media platform after the account's works are published, or other factors such as user login time. Therefore, this application embodiment does not specifically limit the method of obtaining influence characteristics.

[0069] In this embodiment, the content interaction characteristics of the account to be evaluated can be reflected in the interaction results between the content consumer end (users) and the account after the account publishes content, such as comments, likes, and shares, demonstrating the performance of the work published by the account among users. By determining the content interaction characteristics of the account to be evaluated, accounts with a higher proportion of works with high comment rates, like rates, and share rates among self-media accounts can ultimately obtain better evaluation results, thereby improving the platform's ability to recommend high-quality content.

[0070] The calculation method for the content interaction characteristics of the account to be evaluated can be as shown in formula (2):

[0071]

[0072] Formula (2) can be used to determine the content interaction characteristics P of the account to be evaluated by using the interaction data of the content of the account to be evaluated in a public state within a preset time range. k zanCnt represents the number of pieces of content from the account to be evaluated that were publicly available on day k. k The number of user likes received by content in the publicly accessible section of the account to be evaluated on day k; shareCnt k The number of user shares of content in the publicly accessible state of the account to be evaluated on day k; Cnt is the total number of interactions between users and the account to be evaluated; δ and η are smoothing coefficients, which can be δ = 1.0 and η = 10.0; where the values ​​of δ and η can be other values ​​according to actual needs, and this embodiment does not impose any restrictions on them.

[0073] Specifically, when calculating the content interaction characteristics of the account to be evaluated using the above formula (2), the interaction data of the content of the account to be evaluated in a public state within a preset time range can be obtained from the interaction information of the consumed content on the content consumption end. The length of the preset time range can be any length not less than one day, such as one month or one week. The length of the preset time range can be set according to actual needs, and no specific restrictions are imposed here. The total number of interactions Cnt within the preset time range is extracted from the obtained interaction data. The number of interactions each day within the preset time range includes the number of user likes and user shares obtained by the account to be evaluated. For example, the number of interactions on the k-th day within the preset time range (including the number of likes zanCnt) k and the number of shares (shareCnt) k The number of pages (PV) of content that is publicly displayed on each day within the preset time frame of the account to be evaluated. k Then, based on the number of interactions and the amount of publicly accessible content on the same day within a preset time frame, the interaction calculation result for the account to be evaluated on that day is determined. The sum of the interaction calculation results for each day within the preset time range of the account to be evaluated is obtained. Finally, the content interaction feature P of the number to be evaluated is obtained by summing the calculated interaction results and the total number of interactions Cnt.

[0074] The formula above for calculating the content interaction characteristics of the account being evaluated only illustrates how to calculate these characteristics by referring to two types of interaction behaviors: likes and shares. In practical applications, other types of interaction behaviors, such as favorites and comments, can also be included in the formula. Furthermore, in the formula above, the weighting of the number of likes and shares is 1:3. In practical applications, other weighting ratios can be set according to specific needs, such as 1:2, 2:1, etc. This is not a limitation; the above formula is merely an example formula for calculating content interaction characteristics.

[0075] In this embodiment, the potential characteristics of the account to be evaluated can be determined by observing the data trends of users consuming the content after the account publishes content. These trends include variations in exposure rate, like rate, comment rate, and share rate, among other interaction types. By identifying these potential characteristics, accounts with high comment, like, and share rates, and those showing a strong trend in user engagement time, can achieve better evaluation results, thereby enhancing the platform's ability to promote high-quality content. This embodiment requires obtaining the comprehensive potential characteristics of the account to be evaluated, taking into account various interaction types between users and the account.

[0076] When calculating the overall potential characteristics of an account to be evaluated, it is necessary to first determine the potential characteristics of the account in each interaction type. The following example, using the "like" interaction type as an example, illustrates how to obtain the potential characteristics of the target interaction type:

[0077]

[0078] Among them, PV zan_trend This indicates the potential characteristics of the account to be evaluated in the "like" interaction type, and calculates the average growth trend of the number of target interaction types between each two adjacent time periods. In this embodiment, a preset time range is used as the time range for considering the potential characteristics of the account to be evaluated. The preset time range is divided into multiple time segments according to the length of the time, which can be done by dividing the preset time range into equal segments or by different time segments having different lengths. cnt Indicates the number of segments. Zan right This indicates the number of target interaction types in two adjacent segments, categorized by the right-hand side of the timeline; Zan leftThis indicates the number of target interaction types in two adjacent segments, categorized by the segment to the left of the timeline; time right This indicates the number of days between the end of the time segment on the right side of the time axis and the start of the preset time range in two adjacent segments; time left This indicates the number of days between the end of the time segment on the left side of the time axis and the start of the preset time range in two adjacent segments.

[0079] Specifically, when calculating the comprehensive potential characteristics of an account to be evaluated, interaction data of the content produced by the account within a preset time range can be obtained from the interaction information of the consumed content on the content consumption side. The preset time range can be any length of at least one day, such as one month or one week, and its length can be set according to actual needs; no specific restrictions are imposed here. The obtained interaction data includes various interaction types, such as user likes, comments, and shares of consumed content, as well as user consumption time, etc., which are not limited here.

[0080] Taking the target interaction type as user likes on consumed content as an example, the trend calculation result of the number of likes on consumed content can be determined according to the above formula (3). Specifically, the preset time range is divided into multiple time segments. For example, the preset time range can be 1 month, and 1 month is divided into 4 segments, each time segment being 1 week. The number of likes on consumed content by users in each time segment is extracted from the obtained interaction data. For each two adjacent time segments, the difference Zan between the number of likes in the two time segments is calculated according to the time axis based on the time segment with the earlier time and the time segment with the later time. right -Zan left And the difference in number of days between the end date of the later time segment and the end date of the earlier time segment. right -time left Zan based on the difference in the number of likes right -Zan left and the difference in the number of days (time) right -time left Determine the trend of user likes on consumed content. Following the steps described above, calculate the trend of likes for consumed content from users in all adjacent time segments across multiple time segments. Then, sum these multiple trend calculations to obtain the overall trend of likes for consumed content. Then, based on the overall trend of user likes on consumed content and the total number of segments across multiple time periods, the potential PV (page view) of the account to be evaluated in terms of user likes on consumed content is calculated. zan_trend .

[0081] Similarly, when the target interaction types are user comments, sharing, and user consumption time, the method for calculating the potential characteristics of the account to be evaluated in user comments, user sharing, and user consumption time is the same as the method for calculating the potential characteristics of the account to be evaluated in user likes, and will not be repeated here. After calculating the potential characteristics of the account to be evaluated in each different interaction type, the comprehensive potential characteristics of the account to be evaluated are obtained according to the potential characteristics of the account to be evaluated in various different types and the weights corresponding to each type. The specific calculation method is shown in formula (4):

[0082] S growth =α*PV zan_trend +β*PV comm_trend +χ*PV share_trend +γ*PV time_trend (4)

[0083] Among them, S growth PV represents the overall potential characteristics of the account to be evaluated. zan_trend This indicates the potential of the account to be evaluated in terms of user engagement with the content consumed; PV (Page Views). comm_trend This indicates the potential characteristics of the account to be evaluated in terms of user comments on consumed content; PV share_trend This indicates the potential of the account to be evaluated in terms of user sharing of consumed content; PV (Page Views) time_trend This represents the potential characteristics of the account to be evaluated in terms of the duration of user consumption of content; α, β, χ, and γ are the weighting coefficients assigned to the potential characteristics of the four interaction types of likes, comments, shares, and consumption duration when obtaining the comprehensive potential characteristics of the account to be evaluated. The values ​​can be α = 0.15, β = 0.25, χ = 0.25, and γ = 0.35; where the values ​​of α, β, χ, and γ can be other values ​​according to actual needs, and this embodiment does not impose any restrictions on this.

[0084] Specifically, the comprehensive potential characteristics of the account to be evaluated are obtained based on the calculated potential characteristics of the account in terms of users' liking, commenting, sharing, and consumption time of the content, along with the weights corresponding to each interaction type. The weights corresponding to each interaction type can be adjusted according to actual needs; for example, the weight coefficients of interaction types with high reference value can be increased, while the weight coefficients of interaction types with lower reference value can be decreased.

[0085] In this embodiment, after calculating the influence characteristics, content interaction characteristics, and potential characteristics of the account to be evaluated, the influence characteristics, content interaction characteristics, and potential characteristics can be fused according to a preset posterior feature fusion method to obtain the posterior evaluation result.

[0086] The calculation method for obtaining the posterior evaluation result by fusing the influence characteristics, content interaction characteristics and potential characteristics of the account to be evaluated can be shown in formula (5):

[0087] S C =(μ+I) α (π+P) β (σ+S growth ) γ (5)

[0088] Among them, S C I represents the posterior evaluation result of the account to be evaluated; P represents the influence characteristics of the account to be evaluated; S represents the content interaction characteristics of the account to be evaluated; growth The α, β, and γ represent the potential characteristics of the account to be evaluated; α, β, and γ represent the control parameters of the influence characteristics, content interaction characteristics, and potential characteristics of the account to be evaluated, respectively, and can be used as the relative weight values ​​corresponding to the three posterior features, with values ​​of α = 0.5, β = 5, and γ = 0.5; μ, π, and σ represent the smoothing coefficients, with values ​​of μ = 0.04, π = 0.08, and σ = 0.0001; the values ​​of α, β, γ, μ, π, and σ can be taken as other values ​​according to actual needs, and this embodiment does not impose any restrictions on this.

[0089] Specifically, this can be based on the calculated influence characteristic I, content interaction characteristic P, and potential characteristic S. growth The posterior evaluation result S of the account to be evaluated is calculated. C .

[0090] In this embodiment, the posterior features of the account to be evaluated are obtained, including the influence features, content interaction features, and potential features of the account. The influence and potential features reflect the account's influence and contribution to the self-media platform, as well as the platform users' recognition of the account. The influence features, content interaction features, and potential features are fused according to a preset posterior feature fusion method to obtain the posterior evaluation result. This incorporates posterior features that dynamically reflect the account's ability to recommend high-quality content into the evaluation criteria for the account, effectively referencing the feedback from content consumers on the content produced by the account, thereby improving the accuracy of the self-media account evaluation.

[0091] In one embodiment, obtaining prior features of the account to be evaluated may include obtaining content production stability features, content verticality features, and content difference features of the account to be evaluated. Then, the content production stability features, content verticality features, and content difference features are fused according to a preset prior feature fusion method to obtain the prior evaluation result.

[0092] In this embodiment, the content production stability characteristics of the account to be evaluated can be reflected by the stability of the content produced and the recent activity level. By determining the content production stability characteristics of the account to be evaluated, accounts that consistently produce content, have a relatively stable amount of recent content production, and have high recent activity levels can ultimately obtain better evaluation results, thereby improving the platform's ability to recommend high-quality content.

[0093] The calculation method for the content production stability characteristics of the account to be evaluated can be shown in formula (7):

[0094]

[0095] Wherein, σ1 and σ2 represent the average of the mean and the average of the variance of the number of contents produced in each day within a preset time window, which divides the preset time range into multiple time windows; α and β represent the control parameters for the activity performance of the content produced by the account to be evaluated within the target time window, and the control parameters for the activity performance of the content produced within the preset time range can be α = 1 and β = 2; η represents the smoothing coefficient, which can be η = 10; where the values ​​of α, β, and η can be other values ​​according to actual needs, and this embodiment does not impose any restrictions on them.

[0096] Specifically, when calculating the content production stability characteristics of the account to be evaluated using the above formula (7), the preset time range can be divided into multiple time windows, where each time window has the same length. The length of the preset time range and the length of the time window can be any length not less than one day. For example, the preset time range can be one month, and each time window can be one week, etc. The length of the preset time range can be set according to actual needs, and no specific restrictions are imposed here. A target time window is determined among the multiple time windows. The amount of content produced by the account to be evaluated within the target time window is obtained from the original flow information of the content produced by the account to be evaluated. The mean and variance of the amount of content produced by the account to be evaluated within the target time window are calculated based on the amount of content produced by the account to be evaluated within the target time window, in units of days. Then, the mean and variance of the amount of content produced by the account to be evaluated within the other time windows are calculated using the same method as above. Based on the mean and variance of the amount of content produced by the account to be evaluated within the other time windows, the average value σ1 of all the calculated means and the average value σ2 of all the variances are obtained. Finally, the content production stability characteristic C of the account to be evaluated is obtained by calculating the average value σ1 of the mean number of contents produced by the account to be evaluated in multiple time windows and the average value σ2 of the variance of the number of contents produced by the account to be evaluated in multiple time windows.

[0097] In this embodiment, the content verticality characteristic of the account to be evaluated reflects the focus of the content produced by that account. Specifically, it indicates whether the content belongs to a concentrated area or a few fixed areas. Accounts that lack focus or whose content is simply copied from other accounts typically exhibit poor focus. By determining the content verticality characteristic of the account to be evaluated, accounts with better focus and a more concentrated content area can ultimately achieve better evaluation results, thereby improving the platform's ability to recommend high-quality content.

[0098] The calculation method for the content verticality feature of the account to be evaluated can be shown in formula (8):

[0099]

[0100] Where H represents the verticality characteristic of the content of the account to be evaluated; i represents the i-th vertical category in the set of vertical categories involved in the content produced by the account to be evaluated; n represents the total number of vertical categories in the set of vertical categories; P i This represents the percentage of content produced by the account to be evaluated in the i-th vertical category within a preset time frame.

[0101] Specifically, when calculating the content verticality feature of the account to be evaluated using the above formula (8), the correspondence between the content produced by the account to be evaluated and the vertical category within a preset time range can be obtained from the original flow information of the content produced by the account to be evaluated. The length of the preset time range can be any length not less than one day, such as one month or one week. The length of the preset time range can be set according to actual needs, and no specific restrictions are imposed here. Then, the set of vertical categories involved in the content produced by the account to be evaluated within the preset time range is determined according to the correspondence between the content and the vertical category. For the target vertical category in the set of vertical categories, the number of content produced by the account to be evaluated in the target vertical category within the preset time range is obtained according to the correspondence. The ratio of the number of content in the target vertical category to the number of content produced by the account to be evaluated within the preset time range is calculated as the content proportion of the account to be evaluated in the target vertical category within the preset time range. The content proportion of the account to be evaluated in each vertical category in the set of vertical categories is calculated in the same way as above. Finally, the calculation is performed as shown in formula (8) based on the content proportion of the account to be evaluated in different vertical categories in the set of vertical categories within the preset time range to obtain the content verticality feature H of the account to be evaluated within the preset time range.

[0102] In this embodiment, the content difference characteristics of the accounts to be evaluated can reflect the differences between the content published by the accounts to be evaluated, as well as the degree of matching between the content published by the accounts to be evaluated and the account's positioning. By determining the content difference characteristics of the accounts to be evaluated, accounts with a high degree of matching between the content published by the self-media accounts and the account's positioning can ultimately obtain better evaluation results, thereby improving the ability to recommend high-quality content.

[0103] The calculation method for the content difference characteristics of the account to be evaluated can be shown in formula (9):

[0104]

[0105] Where T represents the content difference characteristics of the account to be evaluated; the larger the value, the greater the difference between the content produced by the account to be evaluated; n represents the number of pieces of content produced by the account to be evaluated within a preset time range; i represents the i-th piece of content ordered chronologically within the preset time range; IntersectionSize i,i+1 UnionSize represents the size of the intersection of the tags of the i-th and (i+1)-th content items. i,i+1 This represents the size of the union of the tags of the i-th content and the (i+1)-th content.

[0106] Specifically, when calculating the content difference characteristics of the account to be evaluated using the above formula (9), the tags of the content produced by the account to be evaluated within a preset time range can be obtained from the original flow information of the content produced by the account to be evaluated. The length of the preset time range can be any length not less than one day, such as one month or one week. The length of the preset time range can be set according to actual needs, and no specific restrictions are imposed here. The specific principle is as follows: Figure 2 As shown in the diagram, the tags represent the labels of the content produced by the account being evaluated. The tags of the content produced by the account within a preset time range are sorted chronologically by production time to characterize the similarity between adjacent content tags produced by the account, thus describing the differences between the content produced by the account. Then, based on the production time of the content, the tag intersection size (IntersectionSize) is calculated for adjacent content tags produced by the account within the preset time range. i,i+1 Union Size of the Labels i,i+1 The similarity metric between two adjacent pieces of content is calculated based on the size of the intersection and union of their tags. The same method is used to calculate the similarity metric for all adjacent pieces of content produced by the account being evaluated. Finally, the content difference feature T of the account to be evaluated within the preset time range is calculated based on the similarity metric of all two adjacent pieces of content produced by the account to be evaluated within the preset time range.

[0107] In this embodiment, after calculating the content production stability characteristics, content verticality characteristics, and content difference characteristics of the account to be evaluated, the content production stability characteristics, content verticality characteristics, and content difference characteristics can be fused according to a preset prior fusion method to obtain the prior evaluation results.

[0108] The calculation method for obtaining the posterior evaluation result by fusing the content production stability characteristics, content verticality characteristics, and content difference characteristics can be shown in formula (10):

[0109]

[0110] Among them, S BThe prior evaluation result of the account to be evaluated is represented by α; C represents the content production stability characteristic of the account to be evaluated; H represents the content verticality characteristic of the account to be evaluated; T represents the content difference characteristic of the account to be evaluated; β, χ, and γ are the control parameters for the content production stability characteristic, content verticality characteristic, and content difference characteristic of the account to be evaluated, respectively, and can be used as the relative weight values ​​corresponding to the three prior characteristics, with values ​​of β = 2, χ = 1, and γ = 0.5; μ, π, and σ are the smoothing coefficients, with values ​​of μ = 0.04, π = 0.08, and σ = 0.0001; where the values ​​of β, χ, γ, μ, π, and α can be taken as other values ​​according to actual needs, and this embodiment does not impose any restrictions on this.

[0111] Specifically, the posterior evaluation result S of the account to be evaluated can be calculated based on the calculated content production stability characteristic C, content verticality characteristic H, and content difference characteristic T. B .

[0112] In this embodiment, prior features of the account to be evaluated are obtained, including content production stability features, content verticality features, and content difference features. The content production stability features, content verticality features, and content difference features are then fused according to a preset prior feature fusion method to obtain a prior evaluation result. This allows prior features that reflect the content output characteristics of the account on the self-media platform to be incorporated into the evaluation criteria for the account, effectively referencing the content producer's operation of the account, thereby improving the accuracy of self-media account evaluation.

[0113] In one embodiment, after obtaining the prior evaluation result and the posterior evaluation result of the account to be evaluated, the prior evaluation result and the posterior evaluation result can be fused to obtain the evaluation score of the account to be evaluated. Then, according to the evaluation scores of each account in the self-media platform to which the account to be evaluated belongs, the evaluation score of the account to be evaluated is sorted to obtain the rating result of the account to be evaluated. The rating result of the account to be evaluated is used as the prior and posterior linked evaluation result of the account to be evaluated.

[0114] In this embodiment, after obtaining the prior evaluation result and the posterior evaluation result of the account to be evaluated, the prior evaluation result and the posterior evaluation result can be fused according to a preset prior- and posterior fusion method to obtain the evaluation score of the account to be evaluated.

[0115] The method for calculating the evaluation score of the account to be evaluated by fusing the prior evaluation results and the posterior evaluation results can be shown in formula (11):

[0116] S final =α*S B +(1-α)*SC (11)

[0117] Among them, S final S represents the assessment score of the account to be evaluated. B S represents the prior evaluation result; C α represents the posterior evaluation result; α represents the weighting coefficient of the prior evaluation result.

[0118] Specifically, it can be based on the obtained prior evaluation results S B and posterior evaluation results S C The evaluation score S of the account to be evaluated is calculated. final The weighting coefficients of the prior and posterior evaluation results can be adjusted according to the actual needs of the application. For example, when evaluating an account, in order to make the evaluation more objective and accurate, the posterior evaluation result can be used as the primary reference and the prior evaluation result as the secondary reference. In this case, the weighting coefficient of the posterior evaluation result can be increased and the weighting coefficient of the prior evaluation result can be decreased. For example, if the weighting coefficient of the posterior evaluation result is 0.7, then the weighting coefficient of the prior evaluation result is 0.3. The value of the weighting coefficient can be other values ​​according to actual needs, and this embodiment does not impose any restrictions on this.

[0119] After obtaining the evaluation score of the account to be evaluated, the scores can be sorted based on the evaluation scores of other accounts on the same self-media platform. Specifically, accounts can be ranked from highest to lowest score within the same self-media platform. Then, based on these rankings, accounts can be divided into several level ranges, with accounts in different ranges corresponding to different levels. The account to be evaluated is then assigned a level corresponding to its ranking range, and this level assignment serves as the pre- and post-evaluation evaluation result. For example, based on the rankings of accounts on the same self-media platform, they can be divided into three level ranges: high-quality, average, and low-quality. The high-quality range corresponds to the top 10% of accounts, and accounts in this range are classified as high-quality. The average range corresponds to the middle 50% of accounts, and accounts in this range are classified as average. The low-quality range corresponds to the bottom 40% of accounts, and accounts in this range are classified as low-quality. When dividing accounts into different tiers based on their rankings on the same self-media platform, the number of tiers and the ranking percentage of each tier can be set according to different application scenarios and actual needs.

[0120] In this embodiment, different rating rules can be set according to the different fields to which the account belongs, and the rating rules corresponding to each field can be used as the basis for rating the account to be evaluated. When rating the account to be evaluated, the field to which the account belongs can be determined first, and then the rating rules corresponding to that field can be used as the basis for rating the account to be evaluated. For example, accounts belonging to the same field in the same self-media platform can be ranked according to their evaluation scores and divided into three rating ranges: high-quality, average, and low-quality. For the technology field, the rating rules corresponding to this field can be set as follows: the high-quality range corresponds to the top 20% of accounts, and accounts in this range are rated as high-quality accounts; the average range corresponds to the middle 40% of accounts, and accounts in this range are rated as average accounts; the low-quality range corresponds to the bottom 40% of accounts, and accounts in this range are rated as low-quality accounts. For the entertainment sector, the ranking rules can be set as follows: the premium ranking range corresponds to the top 5% of accounts, and accounts within this range are classified as premium accounts; the average ranking range corresponds to the middle 45% of accounts, and accounts within this range are classified as average accounts; and the low-quality ranking range corresponds to the bottom 50% of accounts, and accounts within this range are classified as low-quality accounts.

[0121] In one optional approach, after obtaining the prior assessment results and the posterior assessment results of the account to be assessed, the prior assessment results of the account to be assessed can be sorted according to the prior assessment results of each account on the self-media platform to which the account to be assessed belongs, to obtain the first rating result of the account to be assessed. And according to the posterior assessment results of each account on the self-media platform, the posterior assessment results of the account to be assessed can be sorted to obtain the second rating result of the account to be assessed. Then, the first rating result and the second rating result are fused according to a preset fusion method to obtain the fused rating result as the prior and posterior linked assessment result of the account to be assessed. Specifically, based on the prior evaluation results of each account on the self-media platform to which the account to be evaluated belongs, the accounts are ranked to obtain the prior evaluation result ranking of the account to be evaluated as the first grading result. Based on the subsequent evaluation results of each account on the self-media platform to which the account to be evaluated belongs, the accounts are ranked to obtain the subsequent evaluation result ranking of the account to be evaluated as the second grading result. Then, different weight coefficients are assigned to the prior evaluation result ranking and the subsequent evaluation result ranking of the account to be evaluated. The final ranking of the account to be evaluated is then calculated based on the weight coefficients. The account to be evaluated is graded according to the grade range in which the final ranking of the account to be evaluated is located, and this grading result is used as the combined prior and subsequent evaluation results of the account to be evaluated.

[0122] In this embodiment, after obtaining the pre- and post-approval linkage evaluation results of the account to be evaluated, account recommendations can be made based on the pre- and post-approval linkage evaluation results of each account on the self-media platform to which the account to be evaluated belongs. This involves recommending content produced by the recommended accounts. For example, accounts with pre- and post-approval linkage evaluation results classified as "high-quality accounts" and their content can be recommended to content consumers. Optionally, content pools for different fields can be constructed based on the different fields to which the content produced by the account to be evaluated belongs. First, the account evaluation result filtering conditions corresponding to each field's content pool are determined. That is, based on whether the account evaluation results of accounts within each field meet preset conditions, it is determined whether the content posted by that account can enter the content pool of the corresponding field. Based on the pre- and post-approval linkage evaluation results of each account on the self-media platform and the account evaluation result filtering conditions corresponding to each field, content pools for each field are constructed, and then the content in each field's content pool is recommended to content consumers. For example, the account evaluation result filtering condition for the technology field's content pool is that the account evaluation results are in the top 10%. Therefore, content posted by accounts that meet this condition can enter the technology field's content pool. The content pool in the entertainment field uses accounts whose evaluation results rank in the top 5% as the selection criteria. Content posted by accounts meeting this criterion can then enter the content pool in the technology field. In this embodiment, account recommendations can be made solely based on the sequential evaluation results of accounts within the self-media platform to which the account to be evaluated belongs, or content pools can be constructed for each field, and content from those pools can be recommended. This embodiment allows content from high-quality self-media accounts to be prioritized for entry into the content pool, thus distributing it to platform users in advance and improving platform operational efficiency. It also concentrates recommended traffic on high-quality content creators who genuinely contribute to the operation of the self-media platform, reducing wasted traffic and enhancing incentives for high-quality content creators.

[0123] As mentioned earlier, the processes of obtaining prior and posterior features of the account to be evaluated, calculating the prior evaluation result, posterior evaluation result, and the combined prior and posterior evaluation result can be automated. For example, they can be implemented separately through computer programs, or corresponding training models can be pre-trained and invoked. Based on unsupervised analysis and modeling methods, manual annotation of account sample data is unnecessary, resulting in lower modeling costs and higher development efficiency. The modeling method for obtaining posterior features can be better integrated with the platform's consumer-side goals, reducing manual intervention. Furthermore, this application's embodiments, by evaluating platform accounts, can more agilely, quickly, and promptly identify accounts with abnormal ratings, reducing the cost of account screening and differentiation, and improving processing timeliness. Specifically, accounts with abnormal ratings have lower posterior evaluation results on the consumer side due to decreased content quality, causing their rating to drop promptly, their ranking to decrease, and preventing them from receiving recommended traffic, thus avoiding traffic waste.

[0124] In this embodiment, the evaluation score of the account to be evaluated is obtained by fusing the prior evaluation results and the posterior evaluation results. Based on the evaluation scores of each account on the self-media platform to which the account to be evaluated belongs, the evaluation scores of the account to be evaluated are sorted, and the rating result of the account to be evaluated is obtained as the prior and posterior linked evaluation result of the account to be evaluated. This realizes that when evaluating an account, not only are prior features used as the evaluation basis, but also posterior features that can dynamically reflect the account's ability to recommend high-quality content are incorporated into the evaluation basis. This can effectively refer to the feedback of content consumers on the content produced by the account, so that the final prior and posterior linked evaluation result more accurately reflects the account to be evaluated's ability to promote high-quality content on the platform.

[0125] In one embodiment, after obtaining the prior and posterior linked evaluation results of the account to be evaluated by fusing the prior and posterior evaluation results, the steps of obtaining the prior and posterior features of the account to be evaluated can be repeated at a preset frequency. Then, the latest prior and posterior evaluation results are obtained and fused to obtain the latest prior and posterior linked evaluation results of the account to be evaluated. At the same time, the latest obtained prior and posterior linked evaluation results of the account to be evaluated are updated to the self-media account level database, so that the prior and posterior linked evaluation results of the account to be evaluated can be updated in a timely manner.

[0126] In this embodiment, by performing the steps of obtaining the prior and posterior features of the account to be evaluated at a preset frequency, the prior and posterior linkage evaluation results of the account to be evaluated can be updated in a timely manner, thereby improving the platform's accuracy in recommending high-quality content.

[0127] Based on the above-described account evaluation method, this application also provides an account evaluation device. Figure 3This is a schematic diagram of the structure of an account evaluation device provided in an embodiment of this application, as shown below. Figure 3 As shown, it includes:

[0128] The prior feature acquisition unit 310 is used to acquire the prior features of the account to be evaluated, wherein the prior features are used to characterize the content output characteristics of the account to be evaluated on the self-media platform.

[0129] The prior feature fusion unit 320 is used to perform feature fusion based on the prior features to obtain the prior evaluation result of the account to be evaluated;

[0130] The posterior feature fusion unit 330 is used to perform feature fusion based on the posterior features of the account to be evaluated to obtain the posterior evaluation result of the account to be evaluated; the posterior features include influence features and potential features, wherein the influence features are determined based on the follow operation data and unfollow operation data of the account to be evaluated on the self-media platform, and the potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated.

[0131] The evaluation result fusion unit 340 is used to obtain the prior and subsequent evaluation results of the account to be evaluated based on the prior evaluation results and the subsequent evaluation results. The prior and subsequent evaluation results are used by the self-media platform for content recommendation.

[0132] This application provides an account evaluation device. First, a priori feature acquisition unit acquires priori features of the account to be evaluated. A priori feature fusion unit performs feature fusion based on these priori features to obtain a priori evaluation result for the account. Then, a posterior feature fusion unit performs feature fusion based on the posterior features of the account to be evaluated to obtain a posterior evaluation result for the account. The posterior features include influence features and potential features. Finally, an evaluation result fusion unit obtains a combined priori and posterior evaluation result for the account, which is used for content recommendation on a self-media platform. Because this device not only uses priori features as the evaluation basis but also incorporates posterior features that dynamically reflect the account's ability to recommend high-quality content, it can effectively reference the feedback from content consumers on the content produced by the account. Therefore, the final combined priori and posterior evaluation result more accurately reflects the account's ability to recommend high-quality content on the platform. Furthermore, both the priori and posterior features can be acquired automatically in real time, and the feature fusion and evaluation result fusion operations can be completed without human intervention or experience. Therefore, this device can significantly improve the efficiency of account evaluation on self-media platforms with massive amounts of data.

[0133] Optionally, the posterior feature fusion unit 330 is specifically used for:

[0134] The influence feature, the content interaction feature, and the potential feature are fused according to a preset posterior feature fusion method to obtain the posterior evaluation result.

[0135] Optionally, the posterior feature fusion unit 330 is specifically used for:

[0136] Based on the follow and unfollow data of the account to be evaluated, the latest total number of followers of the account to be evaluated is obtained;

[0137] Obtain a first set of objects that are following the account to be evaluated and a second set of objects that have unfollowed the account to be evaluated;

[0138] Based on the following weight of each user in the first object set for the account to be evaluated, the date on which each user in the first object set followed the account to be evaluated, and the current date, a first calculation result corresponding to the following behavior is obtained; and based on the following weight of each user in the second object set for unfollowing the account to be evaluated, the date on which each user in the second object set unfollowed the account to be evaluated, and the current date, a second calculation result corresponding to the unfollowing behavior is obtained.

[0139] The influence characteristics of the account to be evaluated are calculated based on the latest total number of followers, the first calculation result, and the second calculation result.

[0140] Optionally, the posterior feature fusion unit 330 is specifically used for:

[0141] Obtain interaction data of the content of the account to be evaluated that is in a public state within a preset time range;

[0142] Extract the total number of interactions and the number of interactions for each day within the preset time range from the interaction data;

[0143] Obtain the number of publicly accessible contents of the account to be evaluated for each day within the preset time range;

[0144] The interaction calculation result of the account to be evaluated on that day is obtained based on the number of interactions and the number of contents in public state within the preset time range.

[0145] The sum of the interaction calculation results of the account to be evaluated for each day within the preset time range is obtained.

[0146] The content interaction characteristics of the account to be evaluated are obtained by summing the interaction calculation results and the total number of interactions.

[0147] Optionally, the posterior feature fusion unit 330 is specifically used for:

[0148] Obtain interaction data of content produced by the account to be evaluated within a preset time range; the interaction data includes data of various different interaction types;

[0149] The preset time range is divided into multiple time segments, and the number of different interaction types in each time segment is extracted from the interaction data;

[0150] For the target interaction type among the various different interaction types, for each two adjacent time segments, the quantity trend calculation result of the target interaction type in the two time segments is obtained based on the number of the target interaction type in the earlier time segment and the later time segment, as well as the difference in the number of days between the end date of the later time segment and the end date of the earlier time segment.

[0151] The overall quantity trend calculation result of the target interaction type is obtained by summing the quantity trend calculation results of all two adjacent time segments in the multiple time segments;

[0152] Based on the overall trend of the target interaction type and the total number of segments in the multiple time segments, the potential characteristics of the account to be evaluated in the target interaction type are obtained.

[0153] The comprehensive potential characteristics of the account to be evaluated are obtained based on the potential characteristics of the account in various interaction types and the weights corresponding to each of the various interaction types.

[0154] In this embodiment, the posterior features of the account to be evaluated are obtained, including the influence features, content interaction features, and potential features of the account. The influence and potential features reflect the account's influence and contribution to the self-media platform, as well as the platform users' recognition of the account. The influence features, content interaction features, and potential features are fused according to a preset posterior feature fusion method to obtain the posterior evaluation result. This incorporates posterior features that dynamically reflect the account's ability to recommend high-quality content into the evaluation criteria for the account, effectively referencing the feedback from content consumers on the content produced by the account, thereby improving the accuracy of the self-media account evaluation.

[0155] Optionally, the prior feature fusion unit 320 is specifically used for:

[0156] The content production stability feature, the content verticality feature, and the content difference feature are fused according to a preset prior feature fusion method to obtain the prior evaluation result.

[0157] Optionally, the prior feature acquisition unit 310 is specifically used for:

[0158] Divide the preset time range into multiple time windows;

[0159] For the target time window within the multiple time windows, the mean and variance of the number of contents produced by the account to be evaluated within the target time window in days are obtained based on the number of contents produced by the account to be evaluated.

[0160] The content production stability characteristics of the account to be evaluated are obtained by calculating the mean and variance of the amount of content produced by the account in the multiple time windows.

[0161] Optionally, the prior feature acquisition unit 310 is specifically used for:

[0162] Obtain the correspondence between the content produced by the account to be evaluated and the vertical category within a preset time range;

[0163] Based on the correspondence, determine the set of vertical categories involved in the content produced by the account to be evaluated within the preset time range;

[0164] For a target vertical category in the set of vertical categories, the number of contents of the target vertical category produced by the account to be evaluated within the preset time range is obtained according to the correspondence.

[0165] The ratio of the amount of content in the target vertical category to the amount of content produced by the account to be evaluated within the preset time range is calculated as the proportion of content produced by the account to be evaluated in the target vertical category within the preset time range;

[0166] The content verticality feature of the account to be evaluated within the preset time range is obtained based on the content proportion of the account to be evaluated in different vertical categories within the vertical category set.

[0167] Optionally, the prior feature acquisition unit 310 is specifically used for:

[0168] Obtain the tags for the content produced by the account to be evaluated within a preset time range;

[0169] Based on the chronological order of content production time, calculate the tag intersection size and tag union size of two adjacent pieces of content produced by the account to be evaluated within the preset time range;

[0170] The similarity metric between the two adjacent contents is obtained by calculating the size of the intersection and the size of the union of the tags.

[0171] Based on the similarity metric of all two adjacent pieces of content produced by the account to be evaluated within the preset time range, the content difference characteristics of the account to be evaluated within the preset time range are obtained.

[0172] In this embodiment, prior features of the account to be evaluated are obtained, including content production stability features, content verticality features, and content difference features. The content production stability features, content verticality features, and content difference features are then fused according to a preset prior feature fusion method to obtain a prior evaluation result. This allows prior features that reflect the content output characteristics of the account on the self-media platform to be incorporated into the evaluation criteria for the account, effectively referencing the content producer's operation of the account, thereby improving the accuracy of self-media account evaluation.

[0173] Optionally, the evaluation result fusion unit 340 is specifically used for:

[0174] The prior evaluation results and the posterior evaluation results are fused to obtain the evaluation score of the account to be evaluated;

[0175] Based on the evaluation scores of each account on the self-media platform to which the account to be evaluated belongs, the evaluation scores of the account to be evaluated are sorted to obtain the rating result of the account to be evaluated as the pre- and post-evaluation linkage evaluation result of the account to be evaluated.

[0176] or,

[0177] Based on the prior evaluation results of each account in the self-media platform to which the account to be evaluated belongs, the prior evaluation results of the account to be evaluated are sorted to obtain a first rating result for the account to be evaluated; and based on the subsequent evaluation results of each account in the self-media platform, the subsequent evaluation results of the account to be evaluated are sorted to obtain a second rating result for the account to be evaluated.

[0178] The first rating result and the second rating result are merged to obtain the merged rating result, which is used as the sequential linkage evaluation result of the account to be evaluated.

[0179] Optionally, the device further includes:

[0180] The recommendation unit is used to recommend accounts based on the prior evaluation results of each account in the self-media platform to which the account to be evaluated belongs, so as to recommend the content produced by the recommended account.

[0181] And / or,

[0182] The filtering unit is used to construct a content pool for the target field based on the pre- and post-approval linkage evaluation results of each account in the self-media platform and the filtering conditions of the account evaluation results corresponding to the target field.

[0183] Recommend content from the content pool in the target domain.

[0184] Optionally, the device further includes:

[0185] The first determining unit is used to determine the domain to which the account to be evaluated belongs;

[0186] The second determining unit is used to determine the level setting rules of the account to be evaluated based on the field to which the account to be evaluated belongs, and to use the level setting rules as the basis for classifying the account to be evaluated.

[0187] In this embodiment, the evaluation score of the account to be evaluated is obtained by fusing the prior evaluation results and the posterior evaluation results. Based on the evaluation scores of each account on the self-media platform to which the account to be evaluated belongs, the evaluation scores of the account to be evaluated are sorted, and the rating result of the account to be evaluated is obtained as the prior and posterior linked evaluation result of the account to be evaluated. This realizes that when evaluating an account, not only are prior features used as the evaluation basis, but also posterior features that can dynamically reflect the account's ability to recommend high-quality content are incorporated into the evaluation basis. This can effectively refer to the feedback of content consumers on the content produced by the account, so that the final prior and posterior linked evaluation result more accurately reflects the account to be evaluated's ability to promote high-quality content on the platform.

[0188] Optionally, the device further includes:

[0189] The update unit is used to perform the steps of obtaining the prior and posterior features of the account to be evaluated at a preset frequency; and to update the latest obtained prior and posterior linkage evaluation results of the account to be evaluated to the self-media account level database.

[0190] In this embodiment, by performing the steps of obtaining the prior and posterior features of the account to be evaluated at a preset frequency, the prior and posterior linkage evaluation results of the account to be evaluated can be updated in a timely manner, thereby improving the platform's accuracy in recommending high-quality content.

[0191] Based on the above-described account evaluation method, this application also provides an account evaluation system. Figure 4 This is a schematic diagram of the structure of an account evaluation system provided in an embodiment of this application, such as... Figure 4 As shown, the account evaluation system includes: account evaluation device 410, content database 420, statistical reporting interface server 430, and self-media account level database 440.

[0192] The account evaluation device 410 is used to obtain metadata of the content produced by the account to be evaluated from the content database 420, obtain prior features of the account to be evaluated based on the metadata, perform feature fusion based on the prior features to obtain a prior evaluation result of the account to be evaluated, receive statistical data of the account to be evaluated from the statistical reporting server 430, obtain posterior features of the account to be evaluated based on the statistical data, perform feature fusion based on the posterior features of the account to be evaluated to obtain a posterior evaluation result of the account to be evaluated, wherein the posterior features include influence features and potential features, wherein the influence features are determined based on the follow and unfollow operation data of the account to be evaluated on the self-media platform, and the potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated; perform feature fusion based on the prior features to obtain a prior evaluation result of the account to be evaluated, and perform feature fusion based on the posterior features to obtain a posterior evaluation result of the account to be evaluated; and obtain a prior- and posterior-linked evaluation result of the account to be evaluated based on the prior evaluation result and the posterior evaluation result, which is used by the self-media platform for content recommendation.

[0193] The content database 420 is used to provide the account evaluation device with metadata of the content produced by the account to be evaluated, so that the account evaluation device can obtain the prior characteristics of the account to be evaluated based on the metadata.

[0194] The statistical reporting interface server 430 is used to provide the account evaluation device with statistical data of the account to be evaluated, so that the account evaluation device can obtain the posterior features of the account to be evaluated based on the statistical data.

[0195] The self-media account level database 440 is used to receive the prior and subsequent evaluation results of the account to be evaluated, which are written to or updated by the account to be evaluated.

[0196] Figure 5 This is a schematic diagram of the structure of another account evaluation system provided in an embodiment of this application, as shown below. Figure 5 As shown in the embodiment of this application, an account evaluation system may further include: an account evaluation device 501, a content database 502, a statistical reporting interface server 503, a self-media account level database 504, an upstream and downstream content interface server 505, a dispatch center 506, a manual review system 507, a deduplication server 508, a recommendation and distribution system 509, and a content distribution export server 510.

[0197] The account evaluation device 501 is used to obtain metadata of the content produced by the account to be evaluated from the content database 502, obtain prior features of the account to be evaluated based on the metadata, perform feature fusion based on the prior features to obtain a prior evaluation result of the account to be evaluated, receive statistical data of the account to be evaluated from the statistical reporting interface server 503, obtain posterior features of the account to be evaluated based on the statistical data, perform feature fusion based on the posterior features of the account to be evaluated to obtain a posterior evaluation result of the account to be evaluated, wherein the posterior features include influence features and potential features, wherein the influence features are determined based on the follow and unfollow operation data of the account to be evaluated on the self-media platform, and the potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated; perform feature fusion based on the prior features to obtain a prior evaluation result of the account to be evaluated, and perform feature fusion based on the posterior features to obtain a posterior evaluation result of the account to be evaluated; fuse the prior evaluation result and the posterior evaluation result to obtain a prior- and subsequent-prior-linked evaluation result of the account to be evaluated; the prior- and subsequent-prior-linked evaluation result is used by the self-media platform for content recommendation. It also communicates with the dispatch center 506 to complete the level marking of the account and the dynamic update of the account level; it communicates with the manual review system 507 to complete the manual review of the initial level, and the final level result is written into the self-media account level database 504.

[0198] The content database 502 is used to provide the account evaluation device 501 with metadata of the content produced by the account to be evaluated, so that the account evaluation device 501 can obtain the prior characteristics of the account to be evaluated based on the metadata; it is used to store the metadata of the content produced by the content production end, including content size, cover image link, title, publication time, account author, source channel, entry time, and the classification of the content during the manual review process (including first, second, and third level classifications and tag information, for example, an article explaining a certain brand of mobile phone, the first level classification is technology, the second level classification is smartphone, the third level classification is domestic mobile phone, and the tag information is brand name and mobile phone model. These tag information are used to characterize the similarity of adjacent content produced by the account), etc.; it receives and saves the review results and status returned by the manual review system 507; and it is used to receive and save the deduplication results of the deduplication server 508.

[0199] The statistical reporting interface server 503 is used to provide the account evaluation device 501 with statistical data of the account to be evaluated, so that the account evaluation device can obtain the posterior features of the account to be evaluated based on the statistical data; it is also used to receive statistical data reports from the content consumption end, providing data support for subsequent statistical analysis and identification; and to obtain the flow information of the content produced by the account to be evaluated from the uplink and downlink content interface server 505, collecting information in the process of self-media account production.

[0200] The self-media account level database 504 is used to receive the prior and subsequent linkage evaluation results of the account to be evaluated, which are written or updated by the account to be evaluated.

[0201] The uplink and downlink content interface server 505 is used to communicate with the content production end, obtain the content produced by the content production end, including the title, publisher, summary, cover image, publication time, etc., and store the obtained content in the content database 502; send the production flow information of each account's content to the statistics reporting interface server 503, including the production time and content type; and save the tagging information of the content produced by the content production end, including category, tags, etc., with the title as extended information, in the content database 502.

[0202] The scheduling center 506 is responsible for the entire scheduling process of content flow. It receives content produced by the content production end through the uplink and downlink content interface server 505, retrieves the content's metadata from the content database and updates it; the scheduling deduplication server 508 marks and filters duplicate content; the scheduling account evaluation device 501 evaluates each account and writes or updates the account evaluation results to the self-media account level database 504; and the manual review system 507 is called to perform manual review.

[0203] The manual review system 507 is used to tag content; read the original information of the content in the content database 502 to ensure that the pushed content complies with the scope permitted by local laws and policies; and write the results of the manual review into the content database 502 through the dispatch center 506.

[0204] The deduplication server 508 is used for deduplication of titles, cover images, body text, video fingerprints, and audio fingerprints; it communicates with the dispatch center 506 and, when encountering the same content, uses the content from the account with the higher account level ranking.

[0205] The recommendation and distribution system 509 is used to obtain the recommended and distributed content from the content database 502 and send the recommended and distributed content to the content distribution export server 510;

[0206] The content distribution export server 510 is used to obtain the recommended content sent by the recommendation distribution system 509 and distribute the content to the content consumer.

[0207] All the actions performed by the devices in this application can be found in the steps outlined in the account assessment method described above.

[0208] This application provides an account evaluation system that obtains metadata about the content produced by an account to be evaluated from a content database using an account evaluation device. Based on the metadata, it obtains prior features of the account to be evaluated. It also receives statistical data about the account to be evaluated from a statistical reporting server and obtains posterior features based on the statistical data. The system performs feature fusion based on the prior features to obtain a prior evaluation result, and performs feature fusion based on the posterior features to obtain a posterior evaluation result. Finally, it fuses the prior and posterior evaluation results to obtain a combined prior and posterior evaluation result, which is then written to or updated in a self-media account level database. Because this system not only uses prior features as the evaluation basis but also incorporates posterior features that dynamically reflect the account's ability to recommend high-quality content, it can effectively reference the feedback from content consumers on the content produced by the account. Therefore, the final combined prior and posterior evaluation result more accurately reflects the account's ability to recommend high-quality content on the platform. Furthermore, both the aforementioned prior and posterior features can be acquired in real time through automation, and the feature fusion and evaluation result fusion operations can also be completed without human intervention or experience. Therefore, this system can significantly improve the efficiency of account evaluation on self-media platforms with massive amounts of data.

[0209] Regarding the account assessment device described in the account assessment system above, the structure of implementing the above account assessment device will be described below in terms of server form and terminal device form respectively.

[0210] Figure 6This is a schematic diagram of a server structure provided in an embodiment of this application. The server 600 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 622 (e.g., one or more processors) and memory 632, and one or more storage media 630 (e.g., one or more mass storage devices) for storing application programs 642 or data 644. The memory 632 and storage media 630 can be temporary or persistent storage. The program stored in the storage media 630 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the server. Furthermore, the CPU 622 may be configured to communicate with the storage media 630 and execute the series of instruction operations in the storage media 630 on the server 600.

[0211] Server 600 may also include one or more power supplies 626, one or more wired or wireless network interfaces 650, one or more input / output interfaces 658, and / or one or more operating systems 641, such as Windows Server. TM Mac OS X TM Unix TM Linux TM FreeBSD TM etc.

[0212] The steps performed by the server in the above embodiments can be based on this Figure 6 The server structure shown.

[0213] CPU 622 is used to perform the following steps:

[0214] Obtain the prior features of the account to be evaluated, which are used to characterize the content output characteristics of the account to be evaluated on the self-media platform;

[0215] Based on the prior features, feature fusion is performed to obtain the prior evaluation result of the account to be evaluated;

[0216] The posterior features of the account to be evaluated are fused to obtain the posterior evaluation result of the account to be evaluated. The posterior features include influence features and potential features. The influence features are determined based on the follow and unfollow data of the account to be evaluated on the self-media platform. The potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated.

[0217] Based on the prior assessment results and the posterior assessment results, the prior and posterior linked assessment results of the account to be assessed are obtained, and the prior and posterior linked assessment results are used for content creation on the self-media platform.

[0218] This application also provides an account evaluation device, such as... Figure 7 As shown, for ease of explanation, only the parts related to the embodiments of this application are shown. For specific technical details not disclosed, please refer to the method section of the embodiments of this application. The terminal can be any terminal device including mobile phones, tablets, personal digital assistants (PDAs), point-of-sale (POS) terminals, in-vehicle computers, etc. Taking a mobile phone as an example:

[0219] Figure 7 This is a block diagram illustrating a portion of the structure of a mobile phone related to the terminal provided in the embodiments of this application. (Reference) Figure 7 The mobile phone includes: a radio frequency (RF) circuit 710, a memory 720, an input unit 730, a display unit 740, a sensor 750, an audio circuit 760, a wireless fidelity (WiFi) module 770, a processor 780, and a power supply 790, among other components. Those skilled in the art will understand that... Figure 7 The mobile phone structure shown does not constitute a limitation on the mobile phone and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0220] The following is combined Figure 7 A detailed introduction to each component of a mobile phone:

[0221] The RF circuit 710 can be used for receiving and transmitting signals during information transmission or calls. Specifically, it receives downlink information from the base station and processes it with the processor 780; additionally, it transmits uplink data to the base station. Typically, the RF circuit 710 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier (LNA), a duplexer, etc. Furthermore, the RF circuit 710 can also communicate wirelessly with networks and other devices. The aforementioned wireless communications may use any communication standard or protocol, including but not limited to Global System for Mobile Communications (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, and Short Messaging Service (SMS).

[0222] The memory 720 can be used to store software programs and modules. The processor 780 executes various mobile phone functions and data processing by running the software programs and modules stored in the memory 720. The memory 720 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 720 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0223] The input unit 730 can be used to receive input numerical or character information, and to generate key signal inputs related to user settings and function control of the mobile phone. Specifically, the input unit 730 may include a touch panel 731 and other input devices 732. The touch panel 731, also known as a touch screen, can collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch panel 731), and drive the corresponding connected devices according to a pre-set program. Optionally, the touch panel 731 may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch position and the signal generated by the touch operation, and transmits the signal to the touch controller; the touch controller receives touch information from the touch detection device, converts it into touch point coordinates, and sends it to the processor 780, and can also receive and execute commands sent by the processor 780. In addition, the touch panel 731 can be implemented using various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel 731, the input unit 730 may also include other input devices 732. Specifically, other input devices 732 may include, but are not limited to, one or more of the following: physical keyboard, function keys (such as volume control buttons, power buttons, etc.), trackball, mouse, joystick, etc.

[0224] The display unit 740 can be used to display information input by the user or information provided to the user, as well as various menus of the mobile phone. The display unit 740 may include a display panel 741, which may optionally be configured as a Liquid Crystal Display (LCD), Organic Light-Emitting Diode (OLED), or similar display panel 741. Further, a touch panel 731 may cover the display panel 741. When the touch panel 731 detects a touch operation on or near it, it transmits the information to the processor 780 to determine the type of touch event. Subsequently, the processor 780 provides corresponding visual output on the display panel 741 based on the type of touch event. Although in Figure 7 In this embodiment, the touch panel 731 and the display panel 741 are two separate components to realize the input and output functions of the mobile phone. However, in some embodiments, the touch panel 731 and the display panel 741 can be integrated to realize the input and output functions of the mobile phone.

[0225] The mobile phone may also include at least one sensor 750, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor. The ambient light sensor can adjust the brightness of the display panel 741 according to the ambient light level, and the proximity sensor can turn off the display panel 741 and / or backlight when the phone is moved to the ear. As a type of motion sensor, an accelerometer sensor can detect the magnitude of acceleration in various directions (generally three axes). When stationary, it can detect the magnitude and direction of gravity and can be used for applications that recognize the phone's posture (such as landscape / portrait switching, related games, magnetometer posture calibration), vibration recognition-related functions (such as pedometer, taps), etc. Other sensors that may be configured in the mobile phone, such as gyroscopes, barometers, hygrometers, thermometers, and infrared sensors, will not be described in detail here.

[0226] Audio circuit 760, speaker 761, and microphone 762 provide an audio interface between the user and the mobile phone. Audio circuit 760 converts received audio data into electrical signals and transmits them to speaker 761, where speaker 761 converts them into sound signals for output. On the other hand, microphone 762 converts collected sound signals into electrical signals, which are received by audio circuit 760, converted into audio data, and then processed by processor 780 before being transmitted via RF circuit 710 to, for example, another mobile phone, or the audio data can be output to memory 720 for further processing.

[0227] WiFi is a short-range wireless transmission technology. Through the WiFi module 770, mobile phones can help users send and receive emails, browse web pages, and access streaming media, providing users with wireless broadband internet access. Although Figure 7 The WiFi module 770 is shown, but it is understood that it is not an essential component of a mobile phone and can be omitted as needed without changing the essence of the invention.

[0228] The processor 780 is the control center of the mobile phone, connecting various parts of the phone through various interfaces and lines. It executes software programs and / or modules stored in the memory 720, and calls data stored in the memory 720 to perform various functions and process data, thereby collecting overall data and information from the phone. Optionally, the processor 780 may include one or more processing units; preferably, the processor 780 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 780.

[0229] The mobile phone also includes a power supply 790 (such as a battery) that supplies power to various components. Preferably, the power supply can be logically connected to the processor 780 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0230] Although not shown, mobile phones may also include a camera, Bluetooth module, etc., which will not be described in detail here.

[0231] In this embodiment of the application, the processor 780 included in the terminal also has the following functions:

[0232] Obtain the prior features of the account to be evaluated, which are used to characterize the content output characteristics of the account to be evaluated on the self-media platform;

[0233] Based on the prior features, feature fusion is performed to obtain the prior evaluation result of the account to be evaluated;

[0234] The posterior features of the account to be evaluated are fused to obtain the posterior evaluation result of the account to be evaluated. The posterior features include influence features and potential features. The influence features are determined based on the follow and unfollow data of the account to be evaluated on the self-media platform. The potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated.

[0235] Based on the prior evaluation results and the posterior evaluation results, the prior and posterior linked evaluation results of the account to be evaluated are obtained, and the prior and posterior linked evaluation results are used by the self-media platform for content recommendation.

[0236] This application also provides a computer-readable storage medium for storing program code that executes any one of the implementation methods of the account evaluation method described in the foregoing embodiments.

[0237] This application also provides a computer program product including instructions that, when run on a computer, cause the computer to execute any one of the implementation methods of the account evaluation method described in the foregoing embodiments.

[0238] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0239] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0240] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0241] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0242] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0243] The above technical solution evaluates accounts by leveraging posterior features to assess their ability to recommend high-quality content on the platform, resulting in more accurate evaluations. Furthermore, it eliminates the heavy reliance on human experience and manpower, significantly improving account evaluation efficiency on self-media platforms with massive amounts of data.

[0244] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. An account evaluation method, characterized in that, include: Obtain the prior features of the account to be evaluated, which are used to characterize the content output characteristics of the account to be evaluated on the self-media platform; The prior features, including content production stability, content verticality, and content difference, are fused to obtain the prior evaluation result of the account to be evaluated; the prior evaluation result is calculated as follows: ； in, The prior evaluation result is represented by C; the content production stability characteristic is represented by H; the content verticality characteristic is represented by T; and the content difference characteristic is represented by T. , , These are the control parameters for the content production stability feature, the content verticality feature, and the content difference feature, respectively. These are the smoothing coefficients; The posterior features of the account to be evaluated, including influence features, content interaction features, and potential features, are fused to obtain the posterior evaluation result of the account. The influence features are determined based on follow and unfollow data for the account on the self-media platform; the content interaction features are determined based on the content and interaction data of the account in public within a preset time range; and the potential features are determined based on the trend changes of various interaction types corresponding to the account. The posterior evaluation result is calculated as follows: ； in, This represents the posterior evaluation result; P represents the influence feature; P represents the content interaction feature. This indicates the potential characteristics; , , These represent the control parameters for the influence feature, the content interaction feature, and the potential feature, respectively. These represent the smoothing coefficients; Based on the prior evaluation results and the posterior evaluation results, the prior and posterior linked evaluation results of the account to be evaluated are obtained, and the prior and posterior linked evaluation results are used by the self-media platform for content recommendation.

2. The method according to claim 1, characterized in that, Influence characteristics are determined based on follow and unfollow data of the account to be evaluated on self-media platforms, including: Based on the follow and unfollow data of the account to be evaluated, the latest total number of followers of the account to be evaluated is obtained; Obtain a first set of objects that are following the account to be evaluated and a second set of objects that have unfollowed the account to be evaluated; Based on the following weight of each object in the first object set to the account to be evaluated, the date on which each object in the first object set followed the account to be evaluated, and the current date, a first calculation result corresponding to the following operation is obtained; and based on the unfollowing weight of each object in the second object set to the account to be evaluated, the date on which each object in the second object set unfollowed the account to be evaluated, and the current date, a second calculation result corresponding to the unfollowing operation is obtained. The influence characteristics of the account to be evaluated are calculated based on the latest total number of followers, the first calculation result, and the second calculation result.

3. The method according to claim 1, characterized in that, Content interaction characteristics are determined based on interaction data of content from the account to be evaluated that is in a public state within a preset time range, including: Obtain interaction data of the content of the account to be evaluated that is in a public state within a preset time range; Extract the total number of interactions and the number of interactions for each day within the preset time range from the interaction data; Obtain the number of publicly accessible contents of the account to be evaluated for each day within the preset time range; The interaction calculation result of the account to be evaluated on that day is obtained based on the number of interactions and the number of contents in public state within the preset time range. The sum of the interaction calculation results of the account to be evaluated for each day within the preset time range is obtained. The content interaction characteristics of the account to be evaluated are obtained by summing the results of the interaction calculations and the total number of interactions.

4. The method according to claim 1, characterized in that, Potential characteristics are determined based on the trend changes in data from various different interaction types corresponding to the account to be evaluated, including: Obtain interaction data of content produced by the account to be evaluated within a preset time range; the interaction data includes data of various different interaction types; The preset time range is divided into multiple time segments, and the number of different interaction types in each time segment is extracted from the interaction data; For the target interaction type among the various different interaction types, for each two adjacent time segments, the quantity trend calculation result of the target interaction type in the two time segments is obtained based on the number of the target interaction type in the earlier time segment and the later time segment, as well as the difference in the number of days between the end date of the later time segment and the end date of the earlier time segment. The quantitative trend calculation results of the target interaction type in all two adjacent time segments in the multiple time segments are summed to obtain the overall quantitative trend calculation result of the target interaction type. Based on the overall trend of the target interaction type and the total number of segments in the multiple time segments, the potential characteristics of the account to be evaluated in the target interaction type are obtained; Based on the potential characteristics of the account to be evaluated in various interaction types and the weights corresponding to each of the various interaction types, the comprehensive potential characteristics of the account to be evaluated are obtained.

5. The method according to claim 1, characterized in that, Obtain the content production stability characteristics of the account to be evaluated, including: Divide the preset time range into multiple time windows; For the target time window within the multiple time windows, the mean and variance of the number of contents produced by the account to be evaluated within the target time window are obtained based on the number of contents produced by the account to be evaluated within the target time window, in units of days. The content production stability characteristics of the account to be evaluated are obtained by calculating the mean and variance of the amount of content produced by the account in the multiple time windows.

6. The method according to claim 1, characterized in that, Obtaining the content verticality characteristics of the account to be evaluated includes: Obtain the correspondence between the content produced by the account to be evaluated and the vertical category within a preset time range; Based on the correspondence, determine the set of vertical categories involved in the content produced by the account to be evaluated within the preset time range; For a target vertical category in the set of vertical categories, the number of contents of the target vertical category produced by the account to be evaluated within the preset time range is obtained according to the correspondence. The ratio of the amount of content in the target vertical category to the amount of content produced by the account to be evaluated within the preset time range is used as the proportion of content produced by the account to be evaluated in the target vertical category within the preset time range; The content verticality feature of the account to be evaluated within the preset time range is obtained based on the content proportion of the account to be evaluated in different vertical categories within the vertical category set.

7. The method according to claim 1, characterized in that, Obtain the content difference characteristics of the account to be evaluated, including: Obtain the tags for the content produced by the account to be evaluated within a preset time range; Based on the chronological order of content production time, calculate the tag intersection size and tag union size of two adjacent pieces of content produced by the account to be evaluated within the preset time range; The similarity metric between the two adjacent contents is obtained by calculating the size of the intersection and the size of the union of the tags. Based on the similarity metric of all two adjacent pieces of content produced by the account to be evaluated within the preset time range, the content difference characteristics of the account to be evaluated within the preset time range are obtained.

8. The method according to any one of claims 1-7, characterized in that, The step of obtaining the prior and subsequent linked evaluation results of the account to be evaluated based on the prior evaluation results and the subsequent evaluation results includes: The prior evaluation results and the posterior evaluation results are fused to obtain the evaluation score of the account to be evaluated; Based on the evaluation scores of each account on the self-media platform to which the account to be evaluated belongs, the evaluation scores of the account to be evaluated are sorted to obtain the rating result of the account to be evaluated as the pre- and post-evaluation linkage evaluation result of the account to be evaluated. or, Based on the prior evaluation results of each account in the self-media platform to which the account to be evaluated belongs, the prior evaluation results of the account to be evaluated are sorted to obtain a first rating result for the account to be evaluated; and based on the subsequent evaluation results of each account in the self-media platform, the subsequent evaluation results of the account to be evaluated are sorted to obtain a second rating result for the account to be evaluated. The first rating result and the second rating result are merged to obtain the merged rating result, which is used as the sequential linkage evaluation result of the account to be evaluated.

9. The method according to claim 8, characterized in that, Also includes: Based on the sequential evaluation results of each account in the self-media platform to which the account to be evaluated belongs, the account is recommended to recommend the content produced by the recommended account. And / or, Based on the prior and subsequent evaluation results of each account in the self-media platform and the filtering conditions of the account evaluation results corresponding to the target field, a content pool for the target field is constructed. Recommend content from the content pool in the target domain.

10. The method according to claim 8, characterized in that, Also includes: Determine the domain to which the account to be evaluated belongs; The rating rules for the account to be evaluated are determined based on the domain to which the account to be evaluated belongs, and the rating rules are used as the basis for rating the account to be evaluated.

11. The method according to claim 1, characterized in that, After obtaining the prior and posterior linked evaluation results of the account to be evaluated based on the prior evaluation results and the posterior evaluation results, the method further includes: The steps of obtaining the prior and posterior features of the account to be evaluated are performed at a preset frequency. The latest obtained pre- and post-evaluation results of the accounts to be evaluated will be updated in the self-media account level database.

12. An account evaluation device, characterized in that, include: A priori feature acquisition unit is used to acquire priori features of the account to be evaluated, wherein the priori features are used to characterize the content output characteristics of the account to be evaluated on the self-media platform. The prior feature fusion unit is used to fuse the prior features, including content production stability features, content verticality features, and content difference features, to obtain the prior evaluation result of the account to be evaluated; the calculation method of the prior evaluation result is as follows: ； in, The prior evaluation result is represented by C; the content production stability characteristic is represented by H; the content verticality characteristic is represented by T; and the content difference characteristic is represented by T. , , These are the control parameters for the content production stability feature, the content verticality feature, and the content difference feature, respectively. These are the smoothing coefficients; The posterior feature fusion unit is used to fuse the influence features, content interaction features, and potential features of the posterior features of the account to be evaluated to obtain the posterior evaluation result of the account to be evaluated. The influence features are determined based on the follow and unfollow data of the account to be evaluated on the self-media platform. The content interaction features are determined based on the content and interaction data of the account to be evaluated in a public state within a preset time range. The potential features are determined based on the trend changes of data of various different interaction types corresponding to the account to be evaluated. The posterior evaluation result is calculated as follows: ； in, This represents the posterior evaluation result; P represents the influence feature; P represents the content interaction feature. This indicates the potential characteristics; , , These represent the control parameters for the influence feature, the content interaction feature, and the potential feature, respectively. These represent the smoothing coefficients; The evaluation result fusion unit is used to obtain the prior and subsequent evaluation results of the account to be evaluated based on the prior evaluation results and the subsequent evaluation results. The prior and subsequent evaluation results are used by the self-media platform for content recommendation.

13. An account evaluation system, characterized in that, include: Account evaluation equipment, self-media account level database, content database, and statistical reporting interface server; The account assessment device is used to perform the steps of the account assessment method according to any one of claims 1-11; The content database is used to provide the account evaluation device with metadata of the content produced by the account to be evaluated, so that the account evaluation device can obtain the prior characteristics of the account to be evaluated based on the metadata. The statistical reporting interface server is used to provide the account evaluation device with statistical data of the account to be evaluated, so that the account evaluation device can obtain the posterior features of the account to be evaluated based on the statistical data. The self-media account level database is used to receive the pre- and post-evaluation linkage evaluation results of the account to be evaluated, which are written to or updated by the account to be evaluated.

14. An account evaluation device, characterized in that, Including the processor and memory: The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the steps of the account assessment method according to any one of claims 1-11 according to the instructions in the program code.

15. A computer-readable storage medium, characterized in that, Used to store program code, which, when executed by a processor, implements the steps of the account assessment method according to any one of claims 1-11.

16. A computer program product, characterized in that, Includes instructions that, when run on a computer, cause the computer to perform the steps of the account assessment method according to any one of claims 1-11.

Images