A multimedia content management and push system

Abstract

The application discloses a multimedia content management and pushing system, relates to the technical field of content management and pushing, and comprises a control center, wherein the control center is in communication connection with a database, a user portrait generation module, a derivative portrait set acquisition module, a pushing content generation module and a feedback result processing module; the database is used for storing multimedia content information, user information and browsing records; the user portrait generation module is used for generating a user portrait according to user information and browsing records; the derivative portrait set acquisition module is used for acquiring a corresponding derivative portrait set of the user portrait; the pushing content generation module is used for generating pushing content according to the user portrait and the corresponding derivative portrait set; and the feedback result processing module is used for acquiring a browsing result of the pushing content by the user and generating a corresponding adjustment strategy according to the browsing result.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of content management and pushing technology, and specifically relates to a multimedia content management and pushing system. BACKGROUND

[0002] With the development of big data technology, user portrait has become one of the core technologies in personalized pushing system. By analyzing user behavior data (such as browsing records, click habits, search preferences, etc.), a portrait that conforms to the user's characteristics can be constructed, and then the user's interest can be predicted to provide relevant content.

[0003] Therefore, the pushing content generated based on the user feature portrait tends to have a high degree of repetition, which is easy to cause aesthetic fatigue of the user. How to generate corresponding extended pushing content based on the user feature portrait so that the user can further expand new interest content on the basis of enjoying the original favorite content is a problem that we need to consider at present. SUMMARY

[0004] In order to solve the above problems, the purpose of the present application is to provide a multimedia content management and pushing system.

[0005] The purpose of the present application can be realized by the following technical scheme: a multimedia content management and pushing system, comprising a control center, wherein the control center is communicatively connected with a database, a user portrait generation module, a derivative portrait set acquisition module, a pushing content generation module and a feedback result processing module;

[0006] The database is used for storing multimedia content information, user information and browsing records;

[0007] The user portrait generation module is used for generating a user portrait according to user information and browsing records;

[0008] The derivative portrait set acquisition module is used for acquiring a corresponding derivative portrait set of the user portrait;

[0009] The pushing content generation module is used for generating pushing content according to the user portrait and the corresponding derivative portrait set;

[0010] The feedback result processing module is used for acquiring a browsing result of the pushing content by the user, and generating a corresponding adjustment strategy according to the browsing result.

[0011] Further, the database comprises a content information unit, a user information unit and a browsing record unit;

[0012] The content information unit is used for storing multimedia content information, and the content information comprises a content name, a content link, a content category, a name keyword and a content duration;

[0013] The user information unit is used for storing identity information of the user, and the identity information includes a user name, a user ID, a mobile phone number and a login password.

[0014] The browsing record unit includes a plurality of browsing record sub-units, each of which is used for storing browsing records of the user and has a corresponding user ID; the browsing records include browsing links and browsing times, and the browsing links are in a corresponding relationship with the content links in the content information unit.

[0015] Further, the process of generating a user portrait according to the user information and the browsing records includes:

[0016] Obtaining a user ID corresponding to the user, and denoting the user ID as a target ID;

[0017] Obtaining the browsing records corresponding to the target ID, and then obtaining the content links corresponding to the browsing links in the browsing records, and marking the content categories and name keywords corresponding to the content links as explicit categories and explicit keywords respectively;

[0018] Generating an explicit category set according to the explicit categories;

[0019] Generating an explicit keyword set according to the explicit keywords;

[0020] Setting an explicit threshold, obtaining the explicit categories in the explicit category set and the corresponding occurrence times, and denoting the explicit categories with the occurrence times greater than or equal to the explicit threshold as high-frequency categories;

[0021] Generating a user portrait corresponding to the target ID according to the high-frequency categories and the explicit keyword set.

[0022] Further, the process of obtaining a corresponding derivative portrait set according to the user portrait includes:

[0023] Obtaining the user portraits corresponding to other user IDs except the target ID, obtaining the user IDs corresponding to the user portraits in which the high-frequency categories are consistent with the high-frequency categories in the user portrait corresponding to the target ID, and marking the user IDs as derivative IDs;

[0024] Obtaining the user portrait corresponding to the derivative ID, and marking the user portrait as a derivative portrait;

[0025] Generating a derivative portrait set according to the derivative ID and the derivative portrait.

[0026] Further, the process of generating a push content according to the user portrait and the corresponding derivative portrait set includes:

[0027] The push content includes a large category push content and an anchor push content;

[0028] obtaining high-frequency categories and explicit keywords corresponding to the target ID and the user portrait, and marking the high-frequency categories and the explicit keywords as target categories and target keywords respectively;

[0029] obtaining a derived ID in the derived portrait set;

[0030] generating a browsing data chain according to the derived ID, and generating anchor push content according to the target categories, the target keywords and the browsing data chain;

[0031] generating general category push content according to the target categories and the derived portrait set.

[0032] Further, the process of generating a browsing data chain according to the derived ID, and generating anchor push content according to the target categories, the target keywords and the browsing data chain includes:

[0033] generating query information according to the derived ID, submitting the query information to a browsing record unit, and obtaining a browsing record subunit corresponding to the derived ID;

[0034] obtaining browsing records in the browsing record subunit, sorting browsing links according to browsing times in the browsing records, marking the sorted browsing links as browsing nodes, and generating a browsing data chain according to the browsing nodes;

[0035] if a browsing node in the browsing data chain has a content category corresponding to a browsing link of the browsing node consistent with the target categories, marking the browsing node as an anchor node;

[0036] if a browsing node in the browsing data chain has a name keyword corresponding to a browsing link of the browsing node consistent with the target keywords, marking the browsing node as an anchor node;

[0037] obtaining a browsing node adjacent to the anchor node and located in front of the anchor node, and marking the browsing node as a push node;

[0038] obtaining a browsing link corresponding to the push node, and marking a content link corresponding to the browsing link as anchor push content

[0039] Further, the process of generating general category push content according to the target categories and the derived portrait set includes:

[0040] setting a frequency threshold;

[0041] obtaining high-frequency categories other than the target categories and the number of occurrences of the high-frequency categories corresponding to each derived portrait in the derived portrait set, and marking a high-frequency category corresponding to a number of occurrences greater than or equal to the frequency threshold as a push category;

[0042] According to the push category, a query signal is generated, the query signal is submitted to the content information unit, a content link corresponding to the content category consistent with the push category in the content information unit is obtained, and the content link is recorded as a large category push content.

[0043] Further, the process of generating a corresponding adjustment strategy according to the browsing result of the user on the push content includes:

[0044] The offline duration, the repetition threshold, and the statistical threshold are set.

[0045] The gray list and the black list are set.

[0046] Step a1: obtaining a user ID corresponding to a user, generating a browsing data chain according to the principle process of generating a browsing data chain through a derived ID, and generating a browsing data chain according to the user ID.

[0047] Step a2: obtaining a content link corresponding to the push content, obtaining a browsing node corresponding to the push content in the browsing data chain according to the content link, and marking the browsing node as a verification node.

[0048] Step a3: obtaining a browsing link corresponding to the verification node, and recording a content duration corresponding to a content link corresponding to the browsing link as an effective duration of the verification node.

[0049] Step a4: obtaining a browsing node adjacent to the verification node position and located in front of the verification node position, obtaining a time interval between the verification node and the browsing node, and if the time interval is greater than or equal to the effective duration of the verification node and less than the offline duration, the browsing result of the push content corresponding to the verification node is successful, and the verification node is an effective node; otherwise, the browsing result of the push content corresponding to the verification node is failed, and the verification node is an invalid node.

[0050] Step a5: obtaining a browsing link corresponding to the invalid node, and sending a content category and a name keyword corresponding to a content link corresponding to the browsing link to a gray list.

[0051] Step a6: steps a1 to a5 are repeatedly executed, and the gray list is counted in the process of repeated execution, and the content category, name keyword, content category occurrence frequency and name keyword occurrence frequency existing in the gray list are obtained according to the counting result; if the content category occurrence frequency corresponding to a certain content category in the gray list is greater than the repetition threshold, the content category is sent to the blacklist; if the name keyword occurrence frequency corresponding to a certain name keyword in the gray list is greater than the repetition threshold, the name keyword is sent to the blacklist; when a certain content category appears for the first time in the gray list counting result, the subsequent (Num-1) times of counting results are obtained, if the content category occurrence frequency corresponding to the content category in the first time to the subsequent (Num-1) times of counting results is less than or equal to the repetition threshold, the content category is deleted in the gray list; when a certain name keyword appears for the first time in the gray list counting result, the subsequent (Num-1) times of counting results are obtained, if the name keyword occurrence frequency corresponding to the name keyword in the first time to the subsequent (Num-1) times of counting results is less than or equal to the repetition threshold, the name keyword is deleted in the gray list; wherein Num=statistical threshold;

[0052] If the high-frequency category corresponding to the derivative image exists in the derivative image set and is consistent with the content category in the blacklist, the derivative image is deleted.

[0053] If the explicit keyword corresponding to the derivative image exists in the derivative image set and is consistent with the name keyword in the blacklist, the derivative image is deleted.

[0054] Compared with the prior art, the beneficial effects of the present application are:

[0055] 1、The present application obtains the derivative image set corresponding to the user image of the user, generates the push content of the user through the derivative image set, obtains the corresponding derivative push content on the basis of the multi-media content with the same interest as the user, reasonably expands the exploration range of the user to the multi-media content, and improves the user stickiness.

[0056] 2、The present application sets the gray list and the blacklist, realizes the screening of the derivative image through the blacklist, realizes the buffer area of the screening of the derivative image through the gray list, prevents the deletion operation of the derivative image generation due to operation errors, and makes the derivative image set further fit the user image of the user through the gray list and the blacklist, so that the push multi-media content generated according to the derivative image set in the future is more in line with the interests and hobbies of the user. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 The schematic diagram of the present application. DETAILED DESCRIPTION

[0058] As Figure 1As shown, a multimedia content management and pushing system comprises a control center connected with a database, a user portrait generation module, a derivative portrait set acquisition module, a pushing content generation module and a feedback result processing module.

[0059] The database is used for storing multimedia content information, user information and browsing records.

[0060] The user portrait generation module is used for generating a user portrait according to user information and browsing records.

[0061] The derivative portrait set acquisition module is used for acquiring a corresponding derivative portrait set of a user according to a user portrait.

[0062] The pushing content generation module is used for generating pushing content according to a user portrait and its corresponding derivative portrait set.

[0063] The feedback result processing module is used for acquiring a browsing result of a user on pushing content and generating a corresponding adjustment strategy according to the browsing result.

[0064] It needs to be further explained that, in the specific implementation process, the database comprises a content information unit, a user information unit and a browsing record unit.

[0065] The content information unit is used for storing multimedia content information, and the content information comprises content name, content link, content category, name keyword and content duration; the name keyword is a keyword in the content name.

[0066] The user information unit is used for storing identity information of a user, and the identity information comprises user name, user ID, mobile phone number and login password; the user ID has uniqueness.

[0067] The browsing record unit comprises a plurality of browsing record sub-units, and each browsing record sub-unit is used for storing browsing records of a user and has its corresponding user ID, i.e. each browsing record sub-unit is used for storing browsing records of the user corresponding to the user ID of the browsing record sub-unit; the browsing records comprise browsing link and browsing time, wherein the browsing link and the content link in the content information unit are in a corresponding relationship.

[0068] It needs to be further explained that, in the specific implementation process, the process of generating a user portrait according to user information and browsing records comprises:

[0069] Acquiring a user ID corresponding to a user, and recording the user ID as a target ID;

[0070] Obtaining the browsing record corresponding to the target ID, and obtaining the content link corresponding to the browsing link in the browsing record, marking the content category and name keyword corresponding to the content link as explicit category and explicit keyword respectively;

[0071] Generating an explicit category set according to the explicit category;

[0072] The explicit category set includes one or more explicit categories; , wherein n is a positive integer;

[0073] Generating an explicit keyword set according to the explicit keyword; , wherein n is a positive integer;

[0074] Setting an explicit threshold;

[0075] Obtaining the explicit category and its corresponding occurrence frequency in the explicit category set, and recording the explicit category with an occurrence frequency greater than or equal to the explicit threshold as a high-frequency category;

[0076] Generating a user portrait corresponding to the target ID according to the high-frequency category and the explicit keyword set;

[0077] It needs to be further explained that in the specific implementation process, the process of obtaining the corresponding derivative portrait set of the user portrait includes:

[0078] Obtaining the user portrait corresponding to other user IDs except the target ID, obtaining the user ID corresponding to the user portrait in which the high-frequency category is consistent with the high-frequency category in the user portrait corresponding to the target ID, and marking the user ID as a derivative ID;

[0079] Obtaining the user portrait corresponding to the derivative ID, and marking the user portrait as a derivative portrait;

[0080] Generating a derivative portrait set according to the derivative ID and the derivative portrait;

[0081] The derivative portrait set includes one or more derivative portraits; , wherein n is a positive integer;

[0082] It needs to be further explained that in the specific implementation process, the process of generating the push content according to the user portrait and the corresponding derivative portrait set includes:

[0083] The push content includes a large category push content and an anchor push content;

[0084] Obtaining the high-frequency category and the explicit keyword corresponding to the target ID corresponding user portrait, and marking them as target category and target keyword respectively;

[0085] Obtaining the derivative ID in the derivative portrait set;

[0086] According to the derived ID, a browsing data chain is generated, according to the target category, the target keyword and the browsing data chain, anchor push content is generated;

[0087] According to the target category and the derived image set, large category push content is generated;

[0088] According to the derived ID, a browsing data chain is generated, according to the target category, the target keyword and the browsing data chain, anchor push content is generated, and the process includes:

[0089] According to the derived ID, query information is generated, the query information is submitted to the browsing record unit, and the browsing record subunit corresponding to the derived ID is obtained;

[0090] The browsing record in the browsing record subunit is obtained, the browsing links are sorted according to the browsing time in the browsing record, the sorted browsing links are recorded as browsing nodes, and the browsing data chain is generated according to the browsing nodes;

[0091] The browsing data chain includes a plurality of browsing nodes, one browsing node corresponds to one browsing link, and the browsing nodes are sequentially connected in position. The time interval between the browsing time corresponding to the browsing node in the rear position and the current time is less than the time interval between the browsing time corresponding to the browsing node in the front position and the current time;

[0092] Suppose there are browsing nodes and browsing nodes , the browsing nodes and the browsing nodes are adjacent two browsing nodes, and the browsing node is the browsing node in the front position, and the browsing node is the browsing node in the rear position;

[0093] If the browsing time corresponding to the browsing node is , the browsing time corresponding to the browsing node is , and the current time is set as Nt, then the time interval between the browsing time corresponding to the browsing node in the rear position and the current time is less than the time interval between the browsing time corresponding to the browsing node in the front position and the current time, that is ;

[0094] If the corresponding content category of the browsing link corresponding to the browsing node in the browsing data chain is consistent with the target category, the browsing node is marked as an anchor node;

[0095] If the name keyword corresponding to the browsing link of the browsing node in the browsing data chain is consistent with the target keyword, the browsing node is marked as an anchor node;

[0096] If the browsing node adjacent to the anchor node position is obtained, and the anchor node position is in front of the browsing node position, the browsing node is marked as a push node;

[0097] The browsing link corresponding to the push node is obtained, and the content link corresponding to the browsing link is recorded as anchor push content;

[0098] The process of generating the large category push content according to the target category and the derived portrait set includes:

[0099] A frequency threshold is set;

[0100] The high-frequency categories other than the target category and the number of occurrences of each derived portrait in the derived portrait set are obtained, and the high-frequency category corresponding to the number of occurrences greater than or equal to the frequency threshold is recorded as a push category;

[0101] The query signal is generated according to the push category, and the query signal is submitted to the content information unit. The content link corresponding to the content category consistent with the push category in the content information unit is obtained, and the content link is recorded as large category push content;

[0102] It needs to be further explained that in the specific implementation process, the process of obtaining the browsing result of the user on the push content and generating the corresponding adjustment strategy according to the browsing result includes:

[0103] The offline duration, the repetition threshold and the statistical threshold are set;

[0104] The gray list and the black list are set;

[0105] Step a1: obtaining the user ID corresponding to the user, generating the browsing data chain according to the user ID according to the principle process of generating the browsing data chain through the derived ID;

[0106] Step a2: obtaining the content link corresponding to the push content, obtaining the browsing node corresponding to the push content in the browsing data chain according to the content link, and marking the browsing node as a verification node;

[0107] Step a3: obtaining the browsing link corresponding to the verification node, recording the content duration corresponding to the content link corresponding to the browsing link as the effective duration of the verification node;

[0108] Step a4: if the position of the checking node is adjacent to the position of the browsing node and the position of the browsing node is in front of the position of the checking node, obtaining the time interval between the checking node and the browsing node, if the time interval is greater than or equal to the valid duration of the checking node and less than the offline duration, the browsing result of the pushing content corresponding to the checking node is success, and the checking node is a valid node; otherwise, the browsing result of the pushing content corresponding to the checking node is failure, and the checking node is an invalid node;

[0109] Step a5: obtaining the browsing link corresponding to the invalid node, and sending the content category and name keyword corresponding to the content link of the browsing link to the gray list;

[0110] Step a6: repeatedly executing steps a1 to a5, and counting the gray list in the process of repeating execution, obtaining the content category, name keyword, content category occurrence frequency and name keyword occurrence frequency existing in the gray list according to the counting result; if the content category occurrence frequency corresponding to a content category in the gray list is greater than a repetition threshold, the content category is sent to the black list; if the name keyword occurrence frequency corresponding to a name keyword in the gray list is greater than the repetition threshold, the name keyword is sent to the black list; when a content category appears for the first time in the counting result of the gray list, subsequent (Num-1) counting results are obtained, if the content category occurrence frequency corresponding to the content category in the first to the subsequent (Num-1) counting results is less than or equal to the repetition threshold, the content category is deleted from the gray list; when a name keyword appears for the first time in the counting result of the gray list, subsequent (Num-1) counting results are obtained, if the name keyword occurrence frequency corresponding to the name keyword in the first to the subsequent (Num-1) counting results is less than or equal to the repetition threshold, the name keyword is deleted from the gray list; wherein, Num=statistical threshold;

[0111] If the high-frequency category corresponding to the derivative portrait in the derivative portrait set is consistent with the content category in the black list, the derivative portrait is deleted;

[0112] If the explicit keyword corresponding to the derivative portrait in the derivative portrait set is consistent with the name keyword in the black list, the derivative portrait is deleted;

[0113] The above examples are only used to illustrate the technical method of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A multimedia content management and push system, comprising a control center, characterized in that, The control center is connected to a database, a user profile generation module, a derived profile set acquisition module, a push content generation module, and a feedback result processing module. The database is used to store multimedia content information, user information, and browsing history; The user profile generation module is used to generate user profiles based on user information and browsing history; This includes: obtaining the user ID corresponding to the user, and recording the user ID as the target ID; Obtain the browsing history corresponding to the target ID, then obtain the content links corresponding to the browsing links in the browsing history, and mark the content type and name keywords corresponding to the content links as explicit types and explicit keywords, respectively; Generate a set of dominant species based on the dominant species; Generate a set of explicit keywords based on explicit keywords; Set a dominance threshold; obtain the dominant species that appear in the dominant species set and their corresponding occurrence counts, and record the dominant species whose occurrence count is greater than or equal to the dominance threshold as high-frequency species; Generate user profiles corresponding to target IDs based on high-frequency categories and explicit keyword sets; The derivative profile set acquisition module is used to acquire the corresponding derivative profile set based on the user profile; including: acquiring user profiles corresponding to user IDs other than the target ID, acquiring user IDs corresponding to user profiles whose high-frequency categories in the user profiles are consistent with the high-frequency categories in the user profiles corresponding to the target ID, and marking the user IDs as derivative IDs. Obtain the user profile corresponding to the derived ID, and mark the user profile as a derived profile; Generate a set of derivative images based on the derived ID and the derived image; The push content generation module is used to generate push content based on user profiles and their corresponding derived profile sets. The push content includes broad category push content and anchor push content; Obtain the high-frequency categories and explicit keywords corresponding to the user profile of the target ID, and mark them as target categories and target keywords respectively; Retrieve the derived ID from the derived image set; Generate browsing data chains based on derived IDs, and generate targeted push content based on target type, target keywords, and browsing data chains; Generate category-specific push content based on target type and derived profile sets; The feedback result processing module is used to obtain the user's browsing results of the pushed content and generate corresponding adjustment strategies based on the browsing results.

2. The multimedia content management and push system according to claim 1, characterized in that, The database includes content information units, user information units, and browsing history units; The content information unit is used to store multimedia content information, which includes content name, content link, content type, name keywords, and content duration. The user information unit is used to store the user's identity information, which includes the user name, user ID, mobile phone number, and login password. The browsing history unit includes several browsing history sub-units, which are used to store the user's browsing history and each browsing history sub-unit has its corresponding user ID; the browsing history includes browsing links and browsing time, wherein the browsing links correspond to the content links in the content information unit.

3. The multimedia content management and push system according to claim 2, characterized in that, The process of generating browsing data chains based on derived IDs, and generating targeted push content based on target categories, target keywords, and browsing data chains includes: Generate query information based on the derived ID, submit the query information to the browsing history unit, and obtain the browsing history sub-unit corresponding to the derived ID; Obtain the browsing records in the browsing record subunit, sort the browsing links according to the browsing time in the browsing records, record the sorted browsing links as browsing nodes, and generate a browsing data chain based on the browsing nodes; If a browsing node exists in the browsing data chain and the content type of its corresponding browsing link is consistent with the target type, then the browsing node is marked as an anchor node. If a browsing node exists in the browsing data chain and the name keyword of its corresponding browsing link matches the target keyword, then the browsing node is marked as an anchor node. Obtain the browsing node that is adjacent to the anchor node and is located behind the anchor node when the anchor node is in front of it, and mark the browsing node as a push node; Obtain the browsing link corresponding to the push node, and record the content link corresponding to the browsing link as the anchor push content.

4. The multimedia content management and push system according to claim 3, characterized in that, The process of generating broad-category push content based on target categories and derived user profiles includes: Set the frequency threshold; Obtain the high-frequency categories (excluding the target category) and their frequency of occurrence for each derivative image in the derivative image set, and record the high-frequency categories whose frequency is greater than or equal to the frequency threshold as push categories; A query signal is generated based on the push type, and the query signal is submitted to the content information unit. The content link corresponding to the content type that matches the push type is obtained from the content information unit, and the content link is recorded as the major category push content.

5. A multimedia content management and push system according to claim 4, characterized in that, The process of obtaining the user's browsing results of the pushed content and generating a corresponding adjustment strategy based on the browsing results includes: Set offline duration, repetition threshold, and statistical threshold; Set up graylists and blacklists; Step a1: Obtain the user ID corresponding to the user, and generate a browsing data chain based on the user ID according to the principle of generating a browsing data chain through derived ID; Step a2: Obtain the content link corresponding to the pushed content, obtain the browsing node corresponding to the pushed content in the browsing data chain according to the content link, and mark the browsing node as a verification node; Step a3: Obtain the browsing link corresponding to the verification node, and record the content duration corresponding to the content link of the browsing link as the valid duration of the verification node; Step a4: Obtain the browsing node that is adjacent to the verification node and is located behind the verification node. Obtain the time interval between the verification node and the browsing node. If the time interval is greater than or equal to the effective duration of the verification node and less than the offline duration, the browsing result of the corresponding push content of the verification node is successful, and the verification node is a valid node; otherwise, the browsing result of the corresponding push content of the verification node is unsuccessful, and the verification node is an invalid node. Step a5: Obtain the browsing link corresponding to the invalid node, and send the content type and name keywords corresponding to the content link to the gray list; Step a6: Repeat steps a1 to a5, and statistically analyze the gray list during the repetition process. Based on the statistical results, obtain the content categories, name keywords, frequency of occurrence of content categories, and frequency of occurrence of name keywords in the gray list. If the frequency of occurrence of a content category in the gray list is greater than the duplication threshold, then send the content category to the blacklist. If the frequency of occurrence of a name keyword in the gray list is greater than the duplication threshold, then send the name keyword to the blacklist. When a content category appears for the first time in the gray list statistical results, obtain the subsequent (Num-1) statistical results. If the frequency of occurrence of the content category in all (from the first to the subsequent (Num-1) statistical results is less than or equal to the duplication threshold, then delete the content category from the gray list. When a name keyword appears for the first time in the gray list statistical results, obtain the subsequent (Num-1) statistical results. If the frequency of occurrence of the name keyword in all (from the first to the subsequent (Num-1) statistical results is less than or equal to the duplication threshold, then delete the name keyword from the gray list. Wherein, Num = statistical threshold. If a derivative image in the derivative image set has a high-frequency category that matches the content category in the blacklist, then the derivative image will be deleted. If a derivative profile in the derivative profile set contains explicit keywords that match the name keywords in the blacklist, then that derivative profile will be deleted.

Images