Method and device for measuring content homogeneity in an internet media platform

Abstract

The application provides a kind of measurement method and device of content homogeneity in internet media platform, including determining the group attribute characteristics of target group;Based on the group attribute characteristics of the target group determined, deploy artificial intelligence agent cluster in the internet media platform environment;Run artificial intelligence agent to simulate the internet media platform use behavior of target group;During the process of artificial intelligence agent simulation use behavior, record the content information browsed by each artificial intelligence agent on the recommended page of internet media platform;Based on content information, calculate and output the content homogeneity when target group uses internet media platform.The application can realize high accuracy, high efficiency, protect user privacy and support cross-platform measurement for content homogeneity.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet media, in particular to a method and device for measuring content homogeneity in an Internet media platform. BACKGROUND

[0002] Currently, the way for users to obtain information in an Internet media platform is mainly platform pushing, and the Internet media platform can push personalized content to users according to their attribute characteristics through a recommendation algorithm. For different groups, the content recommended by the platform is also quite different. Due to the one-sided pursuit of attracting users' attention by the Internet media platform, the recommendation algorithm of the platform may recommend too many similar contents to the users, so that the users are trapped in the "information cocoon room" with high homogeneity of the contents they contact. Information cocoon room refers to the phenomenon that users are long-term limited to a single viewpoint or interest field in the information acquisition process due to factors such as algorithm recommendation, user preference selection and social network homogenization, resulting in the closedness of the information environment. The similarity of the content provided by the recommendation algorithm of the Internet media platform, i.e. the degree of information cocoon room, can be defined as "content homogeneity". In the field of communication, content homogeneity can be used as an index to measure the Internet media platform or user portrait, and has important significance in advertising marketing and social governance.

[0003] However, due to the dynamic, random and black-box nature of the platform recommendation algorithm, it is impossible to directly measure the recommendation algorithm itself. Therefore, previous studies mainly use the method of social investigation to measure the content homogeneity subjectively by asking different groups about the information they contact, but this method has great limitations in efficiency and accuracy. Another method is to contact real users or get official support from the platform inside to obtain the historical browsing records of the Internet platform account to calculate the content homogeneity, but this method faces the problem of personal privacy protection and cannot be applied to multiple Internet media platforms. Therefore, in order to calculate the content homogeneity, a new method is urgently needed, which is accurate, efficient, has no privacy risk and supports multiple platforms. SUMMARY

[0004] In view of the problems and deficiencies in the prior art, the present application provides a method and device for measuring content homogeneity in an Internet media platform, which solves the problems of low accuracy, low efficiency, privacy risk and non-cross-platform support of the existing methods, and realizes the measurement of content homogeneity with high accuracy, high efficiency, protection of user privacy and support of cross-platform.

[0005] The present application achieves the above-mentioned purposes through the following technical solutions:

[0006] A method for measuring content homogeneity in an Internet media platform, comprising:

[0007] determine the group attribute characteristics of the target group, take the activity time distribution Q(t) and the activity interval distribution P(t) as the active time attributes of the target group; wherein, t) as the active time attributes of the target group; wherein, , t is the time difference between two activities of the user, P is a probability density function, is a power-law distribution coefficient;

[0008] deploy a cluster of artificial agents in an Internet media platform environment based on the determined group attribute characteristics of the target group;

[0009] run the artificial agents to simulate the target group's Internet media platform usage behavior;

[0010] record the content information browsed by each artificial agent on the recommended page of the Internet media platform during the simulation of the usage behavior;

[0011] based on the content information, calculate and output the content homogeneity of the target group when using the Internet media platform; wherein, the content homogeneity of the artificial agent is calculated based on the content information returned by a single artificial agent; the content homogeneity of the target group is calculated based on all artificial agents.

[0012] According to the content homogeneity measurement method provided by the application, the content homogeneity of the artificial agent is calculated based on the content information returned by a single artificial agent i in the artificial agent running module, which includes the following steps: i

[0013] extract the text semantic vector of each piece of content information returned by the artificial agent i using a pre-trained language model;

[0014] cluster all semantic vectors using the DBSCAN method, and divide all content information into m semantic categories;

[0015] statistic the frequency of content occurrence in each semantic category, and calculate the proportion of each category ;

[0016] based on the probability distribution of each semantic category, calculate the semantic entropy as the content homogeneity of the artificial agent, and the calculation formula is:

[0017]

[0018] wherein, is the content homogeneity of the content information returned by the artificial agent i , m is the total number of semantic categories after clustering,​ The proportion of the number of content items in the first semantic category in the total number of content items, i.e., the number of content items in the first semantic category divided by the total number of content items returned by the artificial intelligence entity. j

[0019] According to the content homogeneity measurement method of the internet media platform provided by the application, the content homogeneity of the target group is calculated based on all the artificial intelligence entities i , and the method comprises the following steps:

[0020] For the content information returned by each artificial intelligence entity i , a content distribution vector is constructed according to the set of semantic categories calculated: , wherein m is the number of content categories;

[0021] An average vector of the content distribution of all artificial intelligence entities is calculated, and the calculation formula is:

[0022]

[0023] Based on and each , the Jensen-Shannon divergence between the content distribution of all artificial intelligence entities and the overall average distribution is calculated and the average value is taken as the content homogeneity index of the target group, and the calculation formula is:

[0024]

[0025] , wherein represents the content homogeneity of the target group, is the content distribution of the first artificial intelligence entity, i is the average of the content distribution of all artificial intelligence entities, represents the Jensen-Shannon divergence between the content distribution of the first artificial intelligence entity and the overall average content distribution, and the calculation formula is: i

[0026]

[0027] , wherein , .

[0028] According to the content homogeneity measurement method of the internet media platform provided by the application, the content homogeneity measurement method of the internet media platform further comprises:

[0029] Based on the attribute characteristics, some artificial intelligence entities are screened, and the content homogeneity of a subdivided group in the target group can be calculated:

[0030] ​​​Determine attribute feature set of subpopulation The attribute feature set Including artificial intelligence body geographical location attribute g and artificial intelligence body age attribute a;

[0031] Based on the attribute feature set From the deployed artificial intelligence agent cluster, the artificial intelligence agent meeting the sub-attribute feature is screened out, and the content homogeneity of the subpopulation is calculated based on the screened part of artificial intelligence agent , Including:

[0032] The content homogeneity of the subpopulation in a certain region;

[0033] The content homogeneity of the subpopulation with a specific age or age range;

[0034] The calculated content homogeneity calculation result is transmitted to the content homogeneity query module.

[0035] According to the measuring method of content homogeneity in the internet media platform provided by the application, the artificial intelligence agent cluster is deployed in the internet media platform environment, including:

[0036] Based on the group attribute features of the target group, the artificial intelligence agent generates the internet media platform account registration information, wherein the nickname is generated according to the language habit of the target group; The gender and age are set according to the proportion of the target group; The IP address is selected according to the address corresponding to the target group region; The content preference is determined by data mining the target group common interactive content; The initial attention user is selected from the account commonly followed by the target group;

[0037] Design and realize the artificial intelligence agent with the ability to simulate the target group internet media platform use behavior;

[0038] Develop account binding interface to bind the internet media platform account with the artificial intelligence agent;

[0039] Set the activity parameters of the artificial intelligence agent.

[0040] According to the measuring method of content homogeneity in the internet media platform provided by the application, when designing and realizing the artificial intelligence agent with the ability to simulate the target group internet media platform use behavior, including:

[0041] For web page simulation, browser automation control tools or browser plug-ins are used to realize the artificial intelligence agent simulating human internet media platform use behavior, and the browser automation control tools and browser plug-ins at least include Webdriver, puppeteer;

[0042] For mobile client simulation, a mobile terminal automation control tool is used to realize artificial agent simulation of human internet media platform use behavior, and the mobile terminal automation control tool at least includes Appium and Auto.js;

[0043] The artificial agent has a simulation mode and a recording mode: in the simulation mode, the target group internet media platform use behavior is simulated according to preset parameters; and in the recording mode, the behavior and platform content information of the artificial agent are recorded.

[0044] The behavior logic and running mode of the artificial agent are written into an executable program and executed in a running module, and the running state is monitored and managed.

[0045] According to the method for measuring content homogeneity in an internet media platform provided by the application, when setting the activity parameters of the artificial agent, the following steps are included:

[0046] The activity time parameter includes setting the activity time distribution Q(t) and the activity interval distribution P(t) according to the activity law of the target group.

[0047] The interaction rule parameter includes:

[0048] According to the access frequency of the target group to different content source pages, a content source page is selected based on a preset probability P(source).

[0049] According to the target group interested content theme, a keyword is set, and based on a preset probability P(click), the artificial agent randomly clicks a piece of content from the current page to enter a detail information page to perform a browsing behavior; if there is no information meeting the condition, the operation is skipped.

[0050] In the detail information page, the artificial agent judges whether to perform a like behavior based on a preset probability P(like) and judges whether to perform a follow content publisher behavior based on a preset probability P(follow).

[0051] The content source page includes a recommendation page, a content classification page and a search page.

[0052] The activity parameters of the artificial agent are saved as a configuration file in JSON or XML text format, and the configuration file is placed in the artificial agent running module.

[0053] According to the method for measuring content homogeneity in an internet media platform provided by the application, when running the artificial agent to simulate the target group internet media platform use behavior, the following steps are included:

[0054] ​Start the simulation mode of the artificial intelligence agent, so that the artificial intelligence agent continuously simulates the target group internet media platform use behavior;

[0055] The behavior of the artificial intelligence agent in the simulation mode includes:

[0056] Logging in the internet media platform account bound to the artificial intelligence agent;

[0057] The artificial intelligence agent enters the recommended page of the internet media platform, scans the page content according to the preset keyword list, and when the content title or abstract contains the keyword, the artificial intelligence agent randomly selects a content that meets the condition to click, and enters the detail information page of the content;

[0058] In the content classification page, the artificial intelligence agent selects a corresponding content category according to the preset content category list; After entering the category page, the artificial intelligence agent browses the content under the page and randomly clicks one of the contents to enter the detail information page;

[0059] The artificial intelligence agent enters the search page and inputs the preset keyword to search; In the search result page, the artificial intelligence agent randomly selects a result to click, and enters the detail information page of the content; Wherein, the selection of search keywords is also based on the analysis of the search behavior of the target group, to simulate the process of the target group actively searching for information;

[0060] When the artificial intelligence agent enters the detail information page, the like behavior and the attention content publisher behavior are performed according to the preset probability value;

[0061] The artificial intelligence agent will accept control commands from the artificial intelligence agent management module during operation, including:

[0062] Suspend the task of the artificial intelligence agent, abort the task of the artificial intelligence agent, restart the artificial intelligence agent, and modify the activity parameters of the artificial intelligence agent.

[0063] According to the content homogeneity measurement method provided by the application, when recording the content information browsed by the artificial intelligence agent on the recommended page, it includes:

[0064] After meeting the preset recording trigger condition, start the recording mode of the artificial intelligence agent;

[0065] The behavior of the artificial intelligence agent in the recording mode includes:

[0066] Browse the recommended page: after entering the recording mode, the artificial intelligence agent browses the recommended page according to its preset browsing rule;

[0067] Text information saving: during the process of browsing the recommended page, the artificial intelligence agent extracts and saves the text information contained in the content appearing in the page;

[0068] Result file generation and return: the saved text information is converted into a result file in JSON text format, and the result file is returned to the artificial intelligence agent management module through the content information upload channel between the artificial intelligence agent running module and the artificial intelligence agent management module.

[0069] A content homogeneity measuring device in an Internet media platform, comprising:

[0070] An artificial intelligence agent running module for deploying and executing a plurality of distributively independently running artificial intelligence agents, and providing a network access environment required for running for the artificial intelligence agents;

[0071] An artificial intelligence agent management module for receiving running data and behavior logs of the artificial intelligence agents in the artificial intelligence agent running module, and issuing scheduling commands to the artificial intelligence agent running module;

[0072] A content homogeneity calculation module for directional reading of data, and calculating a homogeneity index of content in the Internet media platform;

[0073] A content homogeneity query module for interactive query of the calculation result of the content homogeneity;

[0074] Wherein, the content information upload channel and the artificial intelligence agent control instruction transmission channel are provided between the plurality of artificial intelligence agent running modules and the plurality of artificial intelligence agent management modules;

[0075] The content homogeneity calculation module and the artificial intelligence agent management module are provided with a query instruction transmission channel and a data result return channel;

[0076] The content homogeneity query module and the content homogeneity calculation module are provided with a query instruction transmission channel and a content homogeneity result return channel.

[0077] It can be seen that, compared with the prior art, the content homogeneity measuring method in the Internet media platform provided by the present application has the following beneficial effects:

[0078] 1. The content homogeneity measuring method provided by the present application can realize autonomous, accurate and efficient evaluation of content homogeneity of different user groups in the Internet media platform;

[0079] 2. The present application uses artificial intelligence agents to replace the investigation of real users in the traditional method, and obtains the recommended content presented by the Internet media platform in the use process of the corresponding user group in a simulated form. The present application simulates the use behavior of real users in the Internet media platform through artificial intelligence agents, approaches the experience situation of real users in the Internet media platform, and thus realizes accurate, efficient and convenient measurement of content homogeneity.

[0080] 3、 The application can be applied to any Internet media platform that uses recommendation algorithms to recommend personalized content to users, including but not limited to short video platforms, news platforms, video platforms, graphic sharing platforms, e-commerce platforms, etc., and has strong cross-platform capabilities.

[0081] The application will be further described in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0082] Figure 1 is a flowchart of an embodiment of the application of a content homogeneity measurement method in an Internet media platform.

[0083] Figure 2 is a logic flow block diagram of the running of the artificial intelligence agent in the simulation mode in an embodiment of the application of a content homogeneity measurement method in an Internet media platform.

[0084] Figure 3 is a logic flow block diagram of the running of the artificial intelligence agent in the recording mode in an embodiment of the application of a content homogeneity measurement method in an Internet media platform.

[0085] Figure 4 is a schematic diagram of an embodiment of the application of a content homogeneity measurement device.

[0086] BRIEF DESCRIPTION OF DRAWINGS 1、Artificial intelligence agent running module; 2、Artificial intelligence agent management module; 3、Content homogeneity calculation module; 4、Content homogeneity query module; 5、Strategy scheduling unit; 6、Data storage unit. DETAILED DESCRIPTION

[0087] In order to make the purpose, technical scheme and advantages of the application clearer, the technical scheme in the application will be described clearly and completely below in combination with the drawings in the application. Obviously, the described embodiments are part of the embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0088] In this paper, "embodiment" means that the specific features, structures or characteristics described in combination with the embodiment can be included in at least one embodiment of the application. The phrase appears at various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment to other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0089] Reference Figure 1The embodiment provides a content homogeneity measurement method in an Internet media platform, which comprises the following steps:

[0090] In step S1, group attribute characteristics of a target group are determined, and activity time distribution Q(t) and activity interval distribution P(t) are taken as active time attributes of the target group. t) as active time attributes of the target group. , t is the time difference between two activities of a user, P is a probability density function, is a power-law distribution coefficient.

[0091] In step S2, based on the determined group attribute characteristics of the target group, a cluster of artificial agents is deployed in an Internet media platform environment.

[0092] In step S3, the artificial agents simulate the target group Internet media platform use behavior.

[0093] In step S4, during the simulation of the use behavior of the artificial agents, the content information browsed by each artificial agent on the recommended page of the Internet media platform is recorded.

[0094] In step S5, based on the content information, the content homogeneity of the target group when using the Internet media platform is calculated and output; wherein the content homogeneity of the artificial agent is calculated based on the content information returned by a single artificial agent; and the content homogeneity of the target group is calculated based on all artificial agents.

[0095] Specifically, the artificial agents are used to replace the investigation of real users, so that the recommended content presented by the Internet media platform received by the corresponding user group in the use process is obtained in a simulated form, and then used for calculating the content homogeneity.

[0096] The target group refers to a collection of user individuals similar in population characteristic attributes (geographical location attributes, gender attributes, age attributes, active time attributes), content preference attributes, etc.

[0097] The artificial agent refers to an automatic program that can realize the access of the Internet media platform through programming technology and can simulate the Internet media platform use behavior of the target user group in a real environment.

[0098] The artificial agent has two functions, in addition to the function of simulating the Internet media platform use behavior of the target user group, it can also record the content information recommended by the Internet media platform in the page.

[0099] The content information refers to the information in the recommended content of the Internet media platform, including text information, picture information, content tag information, comment information, etc.

[0100] The application can calculate the content homogeneity of different target groups by binding the artificial agent cluster of different group attribute characteristics.

[0101] In the step S1, when determining the attribute characteristics of the target user group, the following are included:

[0102] The demographic attribute characteristics of the target group include gender attribute, age attribute, geographical location attribute, and active time attribute.

[0103] The content preference attribute characteristics of the target group include content category attribute and keyword attribute.

[0104] Further, when determining the demographic attribute characteristics of the target group, the following are included:

[0105] The geographical location attribute can be determined according to geographical regions and city levels, wherein the geographical regions are divided into provincial administrative regions, and the overseas geographical locations are divided into countries or regions.

[0106] The city level division includes first-tier cities, second-tier cities, third-tier cities, and non-tier cities.

[0107] The age attribute is determined according to the approximate age range of the target group.

[0108] The gender attribute includes male and female.

[0109] The active time attribute includes active time distribution and active interval distribution.

[0110] The content preference attribute includes two dimensions of content category attribute and keyword attribute, wherein the content category attribute includes current affairs, sports, entertainment, technology, military, games, and health.

[0111] The keyword includes interest field keyword, geographical location keyword, and gender keyword.

[0112] In the step S2, when deploying the artificial agent cluster based on the group attribute characteristics of the target group, the following are included:

[0113] Based on the group attribute characteristics of the target group, the artificial agent generates the internet media platform account registration information, including nickname, gender, age, IP address, content preference, and initial attention user, wherein the nickname is generated according to the language habit of the target group; the gender and age are set according to the proportion of the target group; the IP address is selected according to the address corresponding to the region of the target group; the content preference is determined by data mining the common interactive content of the target group; and the initial attention user is selected from the account commonly followed by the target group.

[0114] Design and implement an artificial intelligence agent with the ability to simulate the target group's internet media platform usage behavior;

[0115] Develop an account binding interface to bind the internet media platform account with the artificial intelligence agent;

[0116] Set the activity parameters of the artificial intelligence agent.

[0117] Further, in the design and implementation of an artificial intelligence agent with the ability to simulate the target group's internet media platform usage behavior, including:

[0118] For web page simulation, use browser automation control tools or browser plugins to simulate human internet media platform usage behavior by artificial intelligence agent, browser automation control tools and browser plugins include but not limited to Webdriver, puppeteer;

[0119] For mobile client simulation, use mobile terminal automation control tools to simulate human internet media platform usage behavior by artificial intelligence agent, mobile terminal automation control tools include but not limited to Appium, Auto.js;

[0120] The artificial intelligence agent has simulation mode and recording mode: in simulation mode, it simulates the target group's internet media platform usage behavior according to preset parameters; in recording mode, it records its own behavior and platform content information;

[0121] Use Java, JavaScript and Python programming languages to write the behavior logic and running mode of the artificial intelligence agent as executable programs, place them in the running module for execution, and monitor and manage the running state.

[0122] Further, in setting the activity parameters of the artificial intelligence agent, including:

[0123] Activity time parameters, including: setting activity time distribution Q(t) and activity interval distribution P( t) according to the target group's active law;

[0124] Interaction rule parameters, including:

[0125] According to the target group's access frequency to different content source pages, select content source pages based on preset probability P(source);

[0126] According to the target group's interest content theme, set keywords, and based on preset probability P(click), randomly click a piece of content from the current page to enter the detail information page to perform browsing behavior according to whether the content title or label contains the specified keywords; if there is no information meeting the conditions, skip this operation;

[0127] In the detail information page, the artificial intelligence entity determines whether to perform the like behavior based on a preset probability P(like) and determines whether to perform the content publisher following behavior based on a preset probability P(follow);

[0128] The content source page includes a recommendation page, a content classification page, and a search page.

[0129] The activity parameters of the artificial intelligence entity are saved as a configuration file in a text format such as JSON or XML, and the file is placed in the artificial intelligence entity running module.

[0130] In the above step S3, when running the artificial intelligence entity to simulate the target group internet media platform usage behavior, the following is included:

[0131] The simulation mode of the artificial intelligence entity is started, and the artificial intelligence entity continuously simulates the target group internet media platform usage behavior;

[0132] The behavior of the artificial intelligence entity in the simulation mode includes:

[0133] An internet media platform account bound to the artificial intelligence entity is logged in;

[0134] The artificial intelligence entity enters the recommendation page of the internet media platform, scans the page content according to a preset keyword list, and when a keyword is found in the content title or abstract, the artificial intelligence entity randomly selects a piece of content that meets the condition to click and enter the detail information page of the content;

[0135] In the content classification page, the artificial intelligence entity selects a corresponding content category according to a preset content category list; after entering the category page, the artificial intelligence entity browses the content under the page and randomly clicks one piece of content to enter the detail information page;

[0136] The artificial intelligence entity enters the search page and inputs a preset keyword to search; in the search result page, the artificial intelligence entity randomly selects a result to click and enter the detail information page of the content; wherein, the selection of the search keyword is also based on the analysis of the target group search behavior to simulate the process of the target group actively searching for information;

[0137] When the artificial intelligence entity enters the detail information page, the like behavior and the content publisher following behavior are performed according to a preset probability value;

[0138] During the running process of the artificial intelligence entity, control commands from the artificial intelligence entity management module are accepted, including:

[0139] The tasks of the artificial intelligence entity are paused, the tasks of the artificial intelligence entity are suspended, the artificial intelligence entity is restarted, and the activity parameters of the artificial intelligence entity are modified.

[0140] In the above step S4, when recording the content information browsed by the artificial intelligence agent on the recommendation page, the following is included:

[0141] After the preset recording trigger condition is met, the recording mode of the artificial intelligence agent is started;

[0142] The behavior of the artificial intelligence agent in the recording mode includes:

[0143] Browsing the recommendation page: after entering the recording mode, the artificial intelligence agent browses the recommendation page according to its preset browsing rules;

[0144] Text information saving: during the browsing of the recommendation page, the artificial intelligence agent extracts and saves the text information contained in the content appearing on the page;

[0145] Result file generation and return: the saved text information is converted into a result file in a text format such as JSON, and the result file is returned to the artificial intelligence agent management module through a content information upload channel between the artificial intelligence agent running module and the artificial intelligence agent management module.

[0146] In the above step S5,

[0147] In the calculation and output of the content homogeneity of the target group when using the Internet media platform, the following is included:

[0148] Based on the content information returned by a single artificial intelligence agent in the artificial intelligence agent running module, the content homogeneity of the artificial intelligence agent is calculated;

[0149] Based on all artificial intelligence agents, the content homogeneity of the target group is calculated;

[0150] Optionally, based on the attribute characteristics, some artificial intelligence agents are screened, and the content homogeneity of a sub-group in the target group can be calculated;

[0151] The content homogeneity calculation method based on a single artificial intelligence agent is: the content information returned by the artificial intelligence agent is encoded into a semantic vector, and a clustering algorithm is used to divide the content into multiple semantic categories, then the proportion of each category is calculated, and the semantic entropy is calculated based on the category distribution, which is taken as the content homogeneity of the artificial intelligence agent;

[0152] The content homogeneity calculation method of the target group is: based on the semantic category distribution of each artificial intelligence agent, the content distribution difference between it and the average distribution of the group is calculated, and the average value of the difference results of all artificial intelligence agents is taken as the content homogeneity of the target group.

[0153] Specifically, based on the content information returned by a single artificial intelligence agent in the artificial intelligence agent running module, i the content homogeneity of the artificial intelligence agent is calculatedi Content homogeneity includes the following steps:

[0154] For artificial intelligence entities i For each piece of content returned, a text semantic vector is extracted using a pre-trained language model;

[0155] Cluster all semantic vectors using the DBSCAN method, dividing all content information into... m One semantic category;

[0156] Analyze the frequency of content occurrences in each semantic category and calculate the proportion of each category. ;

[0157] Based on the probability distribution of each semantic category, semantic entropy is calculated as the content homogeneity of the AI ​​agent. The calculation formula is as follows:

[0158]

[0159] in, For artificial intelligence entities i The returned content information is homogeneous. m This represents the total number of categories after semantic clustering. For the first j The proportion of content items in a semantic category to the total number of content items, that is, the number of content items in that category divided by the total number of content items returned by the AI ​​agent.

[0160] Specifically, based on all artificial intelligence agents i Calculating the content homogeneity of the target group includes the following steps:

[0161] For each AI agent i The returned content information is used to construct its content distribution vector based on the calculated semantic category set: ,in m Number of content categories;

[0162] Calculate the average vector of the content distribution of all AI agents. The calculation formula is:

[0163]

[0164] based on With each The Jensen-Shannon divergence between the content distribution of all AI agents and the overall average distribution is calculated, and the average value is taken as the content homogeneity index of the target group. The calculation formula is as follows:

[0165]

[0166] wherein, content homogeneity of the target group, content distribution of the i-th AI, i content distribution of the i-th AI, average of all AI content distributions, Jensen-Shannon divergence between the content distribution of the i-th AI and the overall average content distribution, calculated as: i

[0167]

[0168] wherein, , .

[0169] Specifically, based on the attribute characteristics, a part of the AI is screened, and the content homogeneity of the sub-group in the target group can be calculated:

[0170] determining the attribute characteristic set of the sub-group The attribute characteristic set of the sub-group includes the geographic location attribute g of the AI and the age attribute a of the AI;

[0171] Based on the attribute characteristic set, the AI that meets the attribute characteristics of the sub-group is screened from the deployed AI cluster, and the content homogeneity of the sub-group is calculated based on the screened part of the AI , including:

[0172] content homogeneity of the sub-group in a certain region;

[0173] content homogeneity of the sub-group of a certain specific age or age range;

[0174] The calculated content homogeneity is transmitted to the content homogeneity query module.

[0175] In practical application, the present embodiment will describe the process of measuring the content homogeneity of the target group in the weibo platform (weibo.com);

[0176] The attribute characteristics of the target group are determined, including the following steps:

[0177] Step S100, determining the group attribute characteristics of the target group, including:

[0178] Step S110, determining the user portrait of the target group, in the present embodiment, the user portrait of the target group is: young women living in first-tier cities, paying attention to fashion, health or life content categories.

[0179] ​Step S120, based on the user portrait of the target group, determining the specific demographic attribute characteristics of the target group, including: gender attribute, age attribute, geographic location attribute and active time attribute.

[0180] In this embodiment, based on the user portrait of the target group, the gender attribute is determined to be female, the age attribute is determined to be 14 to 35 years old, the geographic location attribute is determined to be a first-tier city, i.e. Beijing, Shanghai, Shenzhen or Guangzhou, and the active time attribute is determined according to the average active time of the microblog platform users, because the user portrait of the target group does not contain clues about the active time.

[0181] According to the research results of this embodiment, the active time peak of the microblog platform users starts at 8am to 10am and ends at 21pm to 22pm, and the active market is relatively uniform in this time period. Accordingly, the active time distribution , wherein , i.e. the probability of the user using the microblog platform at time point t (in hours) is .

[0182] According to the research results of this embodiment, there is a power-law distribution characteristic between the two activities of the user, as shown in the following formula 1:

[0183] (1)

[0184] In the formula, t is the time difference between the two activities of the user, P is the probability density function, is the power-law distribution coefficient. According to the research results, the embodiment selects .

[0185] The above activity time distribution Q(t) and activity interval distribution P( t) are used as the active time attribute of the target group.

[0186] Step S130, based on the user portrait of the target group, determining the content preference attribute characteristics of the target group, including: content category attribute and keyword attribute.

[0187] The content preference attribute includes two dimensions of content category attribute and keyword attribute, wherein the content category attribute includes content categories such as current affairs, sports, entertainment, technology, military, games, health, etc.

[0188] In this embodiment, based on the user portrait of the target group, the content preference attribute is determined to be fashion, health or life.

[0189] The keywords include: sub-interest field keywords, geographic location keywords and gender keywords.

[0190] In this embodiment, based on the user portrait of the target group, a keyword library is formulated in combination with relevant content categories, containing 28 words.

[0191] Step S140, according to the initialization account information table prepared according to the attribute characteristics of the target group, the user nickname of each microblog account is generated according to the language habits of the target group.

[0192] Step S200, deploying an artificial agent cluster based on the group attribute characteristics of the target group.

[0193] One of the core features of this embodiment is that the artificial agent replaces the investigation of real users in the traditional method to obtain the recommended content presented by the Internet media platform in the use process of the corresponding user group in a simulated form.

[0194] The artificial agent aims to simulate the use behavior of real users in the Internet media platform, and approaches the experience situation of real users in the Internet media platform.

[0195] In this embodiment, deploying an artificial agent cluster based on the group attribute characteristics of the target group in step S200 includes the following steps:

[0196] Step S210, setting the registration information of the Internet media platform account, including: nickname, gender, age, IP address, content preference and initial attention user.

[0197] In this embodiment, a mobile phone number is used to register a microblog account in batches, and the gender, age, geographical location and other demographic characteristic information and preferred content categories are set according to the results of step S140 in the registration process of the microblog account, and 3-5 related accounts are randomly followed according to the content category setting.

[0198] Step S220, designing and implementing an artificial agent with the ability to simulate the use behavior of the target group in the Internet media platform.

[0199] In this embodiment, the mobile terminal automation control tool Appium is used to realize the simulation of human Internet media platform use behavior by artificial agents, and the artificial agents realize the automation control of the microblog platform through Appium.

[0200] The designed and implemented artificial agent has two running modes, including: simulation mode and recording mode.

[0201] The behavior logic and running mode of the artificial agent are saved as an executable program using at least but not limited to Java, JavaScript and Python programming languages, and placed in the artificial agent running module for execution.

[0202] Step S230, binding the internet media platform account with the artificial intelligence agent.

[0203] Randomly matching the artificial intelligence agent with the microblog account registered in step S210.

[0204] Step S240, setting the activity time parameters of the artificial intelligence agent, including: activity time distribution Q(t) and activity interval distribution P(t).

[0205] Step S250, setting the interaction rule parameters of the artificial intelligence agent, including: selection source page probability P(source), whether to click content probability P(click), whether to like probability P(like), and whether to follow probability P(follow).

[0206] As shown in Figure 2 , the behaviors of the artificial intelligence agent in the simulation mode include: browsing, liking, and following.

[0207] First, based on the preset probability P(source), one source page is selected from the content source pages, including: recommendation page, content classification page, and search page.

[0208] Second, based on the preset probability P(click), according to whether the content title or label contains the specified keyword, the artificial intelligence agent randomly clicks a piece of content from the current page to enter the detail information page to perform the browsing behavior; if there is no information meeting the condition, the operation is skipped.

[0209] Third, in the detail information page, the artificial intelligence agent judges whether to perform the like behavior based on the preset probability P(like), and judges whether to perform the follow content publisher behavior based on the preset probability P(follow).

[0210] In this embodiment, the activity parameters of the artificial intelligence agent described above are saved as a configuration file in text format such as JSON or XML, and the file is placed in the artificial intelligence agent running module.

[0211] Step S300, running the artificial intelligence agent to simulate the target group's internet media platform usage behavior.

[0212] Step S310, starting the simulation mode of the artificial intelligence agent to continuously simulate the target group's microblog platform usage behavior, including:

[0213] Logging in the registered microblog account to the artificial intelligence agent matched therewith under the assistance of the artificial intelligence agent.

[0214] Browsing the microblog recommendation page, clicking the content appearing the keyword, and entering the detail information page;

[0215] ​In the micro-blog content classification page, select the corresponding content category, browse the content under the content category, randomly click the content, and enter the detail information page.

[0216] In the micro-blog search page, search the keyword, click the content in the search result to enter the detail information page.

[0217] In the micro-blog detail information page, perform the like behavior and the behavior of following the content publisher.

[0218] Activity parameters of the artificial intelligence agent.

[0219] Step S320, issue a control command to the artificial intelligence agent.

[0220] For the artificial intelligence agent in the artificial intelligence agent running module, the control personnel can issue a control command to the artificial intelligence agent through the artificial intelligence agent management module, including: suspending the task of the artificial intelligence agent; aborting the task of the artificial intelligence agent; restarting the artificial intelligence agent; modifying the activity parameters of the artificial intelligence agent.

[0221] Step S400, record the content information browsed by the artificial intelligence agent in the recommendation page.

[0222] Step S410, start the recording mode of the artificial intelligence agent, and the behavior of the artificial intelligence agent in the recording mode is as shown in the table. Figure 3 In this mode, the artificial intelligence agent browses the micro-blog recommendation page, and records the text content of the micro-blog appearing in the page, and the publisher, the publishing time, the like amount, the comment amount and the forwarding amount of the content.

[0223] Step S420, for the micro-blog content entering the detail page, record the popular comment content of the micro-blog, and the publisher, the publishing time and the like amount of the comment content.

[0224] In this embodiment, all content information required for calculating the content homogeneity degree is not pursued in a single recording process, but the time length and frequency of recording content information are determined according to the user active time attribute, and are obtained in multiple times.

[0225] Step S430, convert the content information into a result file in a text format such as JSON, return the result file to the artificial intelligence agent management module through a content information uploading channel possessed between the artificial intelligence agent running module and the artificial intelligence agent management module, and save the data after the artificial intelligence agent management module receives the result file.

[0226] Step S440, repeat the above steps until the content information obtained by the artificial intelligence agent reaches 500, in order to calculate the content homogeneity.

[0227] Step S500, based on the content information, calculating and outputting the content homogeneity of the target group when using the Internet media platform, including the following steps:

[0228] Step S510, based on the content information returned by the single artificial intelligence agent i in the artificial intelligence agent running module, calculating the content homogeneity of the artificial intelligence agent i, including the following steps:

[0229] Step S511, using the pre-trained language model "Chinese BERT" to extract the text semantic vector for each piece of content information returned by the artificial intelligence agent i;

[0230] Step S512, using the "DBSCAN" method to cluster all semantic vectors, dividing all content information into m semantic categories;

[0231] Step S513, counting the frequency of content occurrence in each semantic category and calculating the proportion of each category ;

[0232] Step S514, based on the probability distribution of each semantic category, calculating the semantic entropy as the content homogeneity of the artificial intelligence agent, the calculation formula is:

[0233] (2)

[0234] In the formula, is the content homogeneity of the content information returned by the artificial intelligence agent i, m is the total number of semantic categories after semantic clustering, is the proportion of the number of content in the jth semantic category to the total number of content, that is, the number of content in the category divided by the total number of content returned by the artificial intelligence agent.

[0235] Step S520, based on all artificial intelligence agents i, calculating the content homogeneity of the target group, including the following steps:

[0236] Step S521, for the content information returned by each artificial intelligence agent i, constructing its content distribution vector according to the semantic category set in step S512: , wherein m is the number of content categories;

[0237] Step S522, calculating the average vector of the semantic distribution of all artificial intelligence agents , the calculation formula is:

[0238] (3)

[0239] Step S523, based on and each , the average of the Jensen-Shannon divergences between the content distribution of each artificial intelligence agent and the overall average distribution is taken as the content homogeneity index of the target group, and the calculation formula is:

[0240] (4)

[0241] In the formula, represents the content homogeneity of the target group, is the content distribution of the i-th artificial intelligence agent, i is the average of the content distributions of all artificial intelligence agents, represents the content distribution of the i-th artificial intelligence agent, represents the Jensen-Shannon divergence between the content distribution of the i-th artificial intelligence agent and the overall average content distribution, and the calculation formula is: i

[0242] (5)

[0243] wherein, , .

[0244] Optionally, based on the attribute characteristics, a part of the artificial intelligence agents are screened, and the content homogeneity of the subgroups in the target group can be calculated, including the following steps:

[0245] Step S530, determining the attribute characteristic set of the subgroups The attribute characteristics that can be subdivided include:

[0246] The geographical location attribute (g) of the artificial intelligence agent, for example: one or more of Beijing, Shanghai, Shenzhen, Guangzhou, Hangzhou, etc.

[0247] The age attribute (a) of the artificial intelligence agent, for example: one or more of specific ages or age intervals such as 20 years old, 30 years old, 30 to 35 years old, etc.

[0248] Step S540, screening the artificial intelligence agents that meet the subdivided attribute characteristics, and calculating the content homogeneity of the subgroups based on a part of the artificial intelligence agents , including:

[0249] The content homogeneity of the subgroup whose geographical location is a certain region, for example represents the content homogeneity of the platform recommended content of the artificial intelligence agent whose account geographical location is Shanghai in the target group.

[0250] The content homogeneity of the subgroup whose age is a certain specific age or age range, for example represents the content homogeneity of the platform recommended content of the artificial intelligence agent whose account age is 20 years old in the target group. ​

[0251] The above conditions can be set simultaneously, for example... This indicates the homogeneity of content recommended by AI-powered content platforms for users whose accounts are located in Shenzhen and are 30 years old, within the target audience.

[0252] Step S550: The calculation results in the content homogeneity calculation module are transmitted to the content homogeneity query module.

[0253] Based on the content of this embodiment, the homogeneity of content across multiple internet media platforms for a target group can be measured by extending the implementation process.

[0254] Based on this embodiment (which describes the process of measuring the content homogeneity of a target group on the Weibo platform (weibo.com), the process of applying this invention on multiple Internet media platforms is demonstrated by adding the Douyin short video platform.

[0255] Based on the process of content homogenization on the Weibo platform (weibo.com), the following modifications are made:

[0256] In step S200 above, an AI agent is added to simulate the operation logic of a real user using the Douyin short video platform, and the account registration process and account binding process of the Douyin short video platform are added.

[0257] In step S300 above, the AI ​​agent is run on the Douyin short video platform;

[0258] In step S400 above, a calculation of the homogeneity of content on the Douyin short video platform is added;

[0259] Content information from the Douyin short video platform can be merged into content information from the Weibo platform to calculate the content homogeneity of the target group across multiple internet media platforms.

[0260] Alternatively, the content homogeneity of the two platforms can be calculated separately using a formula, for example... and .

[0261] By setting weight coefficients for the two platforms and Weighted calculation of content homogeneity across multiple platforms As shown in Formula 6:

[0262] (6)

[0263] like Figure 4 As shown, this embodiment also provides a device for measuring content homogeneity in an internet media platform, including:

[0264] An artificial intelligence agent running module 1 is configured to deploy and execute a plurality of distributively independent running artificial intelligence agents, and provide a network access environment required for running the artificial intelligence agents;

[0265] In the embodiment, the artificial intelligence agent running module 1 is implemented by a distributed cloud host, and the features include:

[0266] Basic hardware: 4-core central processor, 8 GB running memory, 100 GB solid state storage; operating system: Ubuntu 22.04 LTS 64-bit version; running environment: OpenJDK 17, Firefox ESR 115 and Java execution framework of artificial intelligence agent program; network configuration: 2 Mbps transmission bandwidth; independent public IPv4 address.

[0267] In the embodiment, 50 artificial intelligence agent running module instances are configured, which are deployed in a distributed architecture, and the specific distribution is: 12 units in Shenzhen, 12 units in Hangzhou, 8 units in Chongqing, 10 units in Xi'an, and 8 units in Shenyang. Each artificial intelligence agent running module 1 is interconnected through a BGP multi-line network, but the running environments of each unit are isolated from each other.

[0268] An artificial intelligence agent management module 2 is configured to receive running data and behavior logs of artificial intelligence agents in the artificial intelligence agent running module 1, and issue scheduling commands to the artificial intelligence agent running module 1; wherein the artificial intelligence agent management module 2 includes a data storage unit 6 for storing data and a policy scheduling unit 5 for receiving and issuing scheduling commands.

[0269] In the embodiment, the data storage unit 6 is implemented by a cloud database, and the features include: 4-core central processor, 8 GB running memory, 800 GB solid state storage; operating system: Ubuntu 22.04 LTS 64-bit version; database system: MySQL 64-bit version; network configuration: internal transmission bandwidth 2500 Mbps.

[0270] In the embodiment, the policy scheduling unit 5 is implemented by a distributed cloud host, and the features include: 4-core central processor, 8 GB running memory, 100 GB solid state storage; system environment: Ubuntu 22.04 LTS 64-bit version; control software: policy scheduling unit software implementation; network configuration: 2 Mbps transmission bandwidth.

[0271] A content homogeneity calculation module 3 is configured to read data in a targeted manner and calculate the homogeneity index of the content in the Internet media platform.

[0272] In the present embodiment, the content homogeneity calculation module 3 is implemented through a distributed cloud host, and the features include: basic hardware: 4-core computing unit, 8GB running memory, 100GB solid state storage; operating system: Ubuntu 22.04 LTS 64-bit version; control software: content homogeneity calculation module 3 software implementation; network configuration: 2Mbps transmission bandwidth.

[0273] The content homogeneity query module 4 is used to interactively query the calculation results of content homogeneity.

[0274] In the present embodiment, the content homogeneity query module 4 is implemented through a distributed cloud host, and the features include: basic hardware: 4-core computing unit, 8GB running memory, 100GB solid state storage; operating system: Ubuntu 22.04 LTS 64-bit version; control software: content homogeneity query module 4 software implementation; network configuration: 20Mbps transmission bandwidth.

[0275] In the present embodiment, in addition to the internal artificial intelligence agent management module 2, the signal paths between the modules are via the public Internet, including: between the multiple artificial intelligence agent running modules 1 and the multiple artificial intelligence agent management modules 2, there are content information uploading channels and artificial intelligence agent control instruction transmission channels; between the content homogeneity calculation module 3 and the artificial intelligence agent management module 2, there are query instruction transmission channels and data result back transmission channels; between the content homogeneity query module 4 and the content homogeneity calculation module 3, there are query instruction transmission channels and content homogeneity result back transmission channels; the internal bidirectional channel of the artificial intelligence agent management module 2 is via a local area network line.

[0276] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present disclosure.

[0277] The above-described embodiments are merely preferred embodiments of the present application, and cannot be used to limit the scope of protection of the present application. Any non-essential changes and replacements made by those skilled in the art based on the present application shall fall within the scope of protection of the present application.

Claims

1. A method for measuring content homogeneity in an internet media platform, characterized in that, Comprise: Determine the group attribute characteristics of the target group, and take the activity time distribution Q(t) and the activity interval distribution P(t) as the active time attributes of the target group; wherein, t) as the target group's active time attributes; wherein, , t is the time difference between two activities of the user, P is the probability density function, is the power law distribution coefficient; Based on the determined target group of population attribute characteristics, deploy a cluster of artificial agents in the Internet media platform environment; Run artificial agent simulation target group Internet media platform use behavior; During the process of artificial agent simulation use behavior, record the content information browsed by each artificial agent on the recommended page of the Internet media platform; Based on the content information, calculate and output the content homogeneity of the target group when using the Internet media platform; wherein, based on the content information returned by a single artificial agent, calculate the content homogeneity of the artificial agent; based on all artificial agents, calculate the content homogeneity of the target group; Running a single artificial intelligence agent based on an artificial intelligence agent operating module i The returned content information, computing the content homogeneity of the artificial intelligence agent i includes the following steps: Artificial intelligence agent i Each piece of content information returned uses a pre-trained language model to extract a text semantic vector; Clustering all semantic vectors, all content information is divided into m semantic categories; Statistics of the frequency of occurrence of content in each semantic category, calculate the proportion of each category ; Based on the probability distribution of each semantic category, calculate the semantic entropy as the content homogeneity of the artificial agent, the calculation formula is: wherein, is an artificial intelligence agent i content homogeneity of the returned content information, m is the total number of semantic clusters, is the proportion of the number of content items in the j semantics category to the total number of content items, i.e., the number of content items in the category divided by the total number of content items returned by the artificial intelligence agent. Based on all artificial agents i computing content homogeneity of the target group, including the following steps: For each artificial intelligence agent i The returned content information is constructed into its content distribution vector according to the calculated semantic category set: Wherein m is the number of content categories; computing the average vector of all artificial agent content distributions with the formula: Based on With each , the Jensen-Shannon divergence between the content distribution of all artificial agents and the overall average distribution is calculated and averaged as the content homogeneity index of the target group, and the calculation formula is: where, content homogeneity of the target group, is the i content distribution of the individual AI, is the average of all AI content distributions, content distribution of the individual AI, i Jensen-Shannon divergence between the content distribution of the individual AI and the overall average content distribution, calculated as: 。 2. The method of claim 1, wherein, Also include: Based on the attribute characteristics, filter part of the artificial agents, and calculate the content homogeneity of the subgroups in the target group: Determining a set of attribute features of a subpopulation The set of attribute features includes an artificial agent geographic location attribute g and an artificial agent age attribute a; Attribute feature set Filtering artificial intelligences meeting the attribute features of the subdivision from the deployed artificial intelligence cluster, calculating the content homogeneity of the subdivision group based on the filtered part of artificial intelligences , comprising: The content homogeneity of the subgroup whose geographical location is a certain area; The content homogeneity of the subgroup whose age is a specific age or age range; The calculation results of the calculated content homogeneity are transmitted to the content homogeneity query module.

3. The method of claim 1, wherein, Deploy a cluster of artificial agents in the Internet media platform environment, including: Based on the group attribute characteristics of the target group, generate Internet media platform account registration information for artificial agents, wherein the nickname is generated according to the language habits of the target group; the gender and age are set according to the proportion of the target group; the IP address is selected according to the address corresponding to the region of the target group; the content preference is determined by data mining the content commonly interacted by the target group; the initial attention user is selected from the account commonly followed by the target group; Design and implement artificial agents with the ability to simulate the use behavior of the target group on the Internet media platform; Develop an account binding interface to bind the Internet media platform account with the artificial agent; Set the activity parameters of the artificial agent.

4. The method of claim 3, wherein, When designing and implementing artificial agents with the ability to simulate the use behavior of the target group on the Internet media platform, including: For web page simulation, use browser automation control tools or browser plugins to simulate human Internet media platform use behavior of artificial agents, browser automation control tools and browser plugins at least include Webdriver, puppeteer; For mobile client simulation, use mobile terminal automation control tools to simulate human Internet media platform use behavior of artificial agents, mobile terminal automation control tools at least include Appium, Auto.js; The artificial agent has a simulation mode and a recording mode: in the simulation mode, it simulates the use behavior of the target group on the Internet media platform according to the preset parameters; in the recording mode, it records its own behavior and platform content information; The behavior logic and running mode of the artificial agent are written as an executable program and executed in the running module, and the running state is monitored and managed.

5. The method of claim 3, wherein, When setting the activity parameters of the artificial agent, including: The activity time parameter includes: setting the activity time distribution Q(t) and the activity interval distribution P(t) according to the activity law of the target group. t) Interaction rule parameters, including: According to the access frequency of the target group to different content source pages, select content source pages based on a preset probability P(source) Setting keywords according to the content theme of interest of the target group, based on the preset probability P (click), the artificial intelligence entity randomly clicks a piece of content from the current page to enter the detail information page to perform the browsing behavior; if there is no information meeting the condition, skip this operation; Wherein, in the detail information page, the artificial intelligence entity judges whether to perform the like behavior based on the preset probability P (like), and judges whether to perform the follow content publisher behavior based on the preset probability P (follow) ; Wherein, the content source page includes: a recommendation page, a content classification page, a search page; Save the activity parameters of the artificial intelligence entity as a configuration file in JSON or XML text format, and place the configuration file in the artificial intelligence entity running module.

6. The method according to any one of claims 1 to 5, characterized in that, When running the artificial intelligence entity to simulate the target group's internet media platform usage behavior, including: Start the simulation mode of the artificial intelligence entity, so that the artificial intelligence entity continuously simulates the target group's internet media platform usage behavior; The behavior of the artificial intelligence entity in simulation mode includes: Log in to the internet media platform account bound to the artificial intelligence entity; The artificial intelligence entity enters the recommendation page of the internet media platform, scans the page content according to the preset keyword list, and when it finds that the content title or abstract contains the keyword, the artificial intelligence entity randomly selects a piece of content that meets the condition to click, and enters the detail information page of the content; In the content classification page, the artificial intelligence entity selects a corresponding content category according to the preset content category list; after entering the category page, the artificial intelligence entity browses the content under the page and randomly clicks one of the contents to enter the detail information page; The artificial intelligence entity enters the search page and inputs the preset keyword to search; in the search result page, the artificial intelligence entity randomly selects a result to click and enters the detail information page of the content; wherein, the selection of the search keyword is also based on the analysis of the target group's search behavior to simulate the target group's active information search process; When the artificial intelligence entity enters the detail information page, it performs the like behavior and the follow content publisher behavior according to the preset probability value; The artificial intelligence entity will accept control commands from the artificial intelligence entity management module during operation, including: Suspend the task of the artificial intelligence entity, abort the task of the artificial intelligence entity, restart the artificial intelligence entity, modify the activity parameters of the artificial intelligence entity.

7. The method according to any one of claims 1 to 5, characterized in that, When recording the content information browsed by the artificial intelligence entity on the recommendation page, including: After meeting the preset recording trigger condition, start the recording mode of the artificial intelligence entity; The behavior of the artificial intelligence entity in recording mode includes: Browsing the recommendation page: after entering the recording mode, the artificial intelligence entity browses the recommendation page according to its preset browsing rules; Text information saving: during the process of browsing the recommendation page, the artificial intelligence entity extracts and saves the text information contained in the content appearing on the page; Result file generation and return: the saved text information is converted into a result file in JSON text format, and the result file is returned to the artificial intelligence agent management module through the content information upload channel between the artificial intelligence agent running module and the artificial intelligence agent management module.

8. A device for measuring content homogeneity in an internet media platform, characterized in that, The measuring device is applied to the content homogeneity measuring method of the Internet media platform as claimed in any one of claims 1 to 7, and the measuring device comprises: An artificial intelligence agent running module is configured to deploy and execute a plurality of distributively independently running artificial intelligence agents, and provide a network access environment required for running of the artificial intelligence agents; An artificial intelligence agent management module is configured to receive running data and behavior logs of the artificial intelligence agents in the artificial intelligence agent running module, and issue a scheduling command to the artificial intelligence agent running module; A content homogeneity calculating module is configured to directionally read data, and calculate a homogeneity index of content in the Internet media platform; A content homogeneity querying module is configured to interactively query a calculation result of the content homogeneity; A plurality of the artificial intelligence agent running modules and a plurality of the artificial intelligence agent management modules are provided with a content information upload channel and an artificial intelligence agent control instruction transmission channel; The content homogeneity calculating module and the artificial intelligence agent management module are provided with a query instruction transmission channel and a data result return channel; The content homogeneity querying module and the content homogeneity calculating module are provided with a query instruction transmission channel and a content homogeneity result return channel.

Images