Full-automatic information release management method and system

Abstract

The invention discloses a full-automatic information release management method and system, and relates to the technical field of information release management, and the method comprises the steps: firstly obtaining to-be-released information uploaded by an uploading user, forming a sequence according to a time sequence, then carrying out the label binding, and carrying out the distribution processing of the information according to the bound label; then, a rule base associated with each label is selected according to the distribution result, and illegal content auditing is carried out on the corresponding to-be-published information, so that normal and abnormal to-be-published information is locked; abnormal information is fed back to an uploading user, and modification is prompted; for the normal to-be-published information, the preference index of the target user of each publishing platform for the published information is evaluated by combining the binding label of the normal to-be-published information, so that the optimal publishing platform is evaluated, the efficient and accurate publishing management of the normal to-be-published information is realized, and the publishing effect is improved while the compliance is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of information release management, and specifically relates to a fully automatic information release management method and system. Background Art

[0002] With the rapid development of science and technology, the demand for automated information release is growing. Driven by big data and artificial intelligence technologies, information release methods are gradually moving from traditional manual review and release to a new stage of intelligent and automated release. While existing technologies have a foundation for information management and dissemination, they suffer from numerous shortcomings. First, information processing efficiency is low, making it difficult to cope with the rapid influx of massive amounts of information. This is especially true for multimodal information processing, which is both time-consuming and error-prone. Second, review standards are inconsistent, and auditors are easily influenced by subjective factors during manual review, resulting in unclear definitions of illegal content and unstable review results. Third, publishing platforms are poorly adaptable and fail to accurately assess user preferences across different platforms, often leading to information mismatches and reduced dissemination effectiveness. In order to solve the above problems, this application proposes a fully automatic information release management method and system. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the present invention provides a fully automatic information release management method and system, which solves the problems of poor information classification effect and low platform release adaptability in the existing technology.

[0004] The purpose of the present invention can be achieved through the following technical solutions: A method for fully automatic information release management, the method comprising the following steps: Step 1: Obtain the information to be published uploaded by the uploading user, organize the information to be published into a sequence in the order of the acquisition time, traverse the determined sequence of information to be published to perform tag binding, and distribute the information to be published according to the tags bound to different information to be published; Step 2: Based on the results of the allocation of the information to be released, the rule bases associated with different tags are selected, and the information to be released that is bound to the corresponding tags is audited for illegal content. Based on the audit results of the illegal content audit, the normal information to be released and the abnormal information to be released are locked; Step 3: For the abnormal information to be released that has been determined, feedback is given to the uploading user, and the uploading user is reminded to modify the illegal content; for the normal information to be released that has been determined, the label bound to the normal information to be released is obtained, the preference index of the target users on different publishing platforms for the corresponding normal information to be released is evaluated, and the optimal publishing platform is evaluated based on the preference index, and the normal information to be released is released.

[0005] As a further solution of the present invention, in step 1, the specific method of traversing the determined sequence of information to be published and performing tag binding is: S21. Obtain the cache window period preset by the operator , using the current time as the cache window period The start time is recorded as ; At On top of that, duration, get a The end time is recorded as ; S22, take to The total number of information to be published uploaded by users is recorded as , according to the upload time of the information to be released Sort the information to be released and get the sequence of information to be released ; S23, from Start traversal ; from Separate the video information, the sound information and the text information, and extract the video information text from the video information; Extracting sound information text from sound information; Combine video information text, sound information text and text information to build The text features of ; S24. Obtain the label table pre-built by the operator , where any label Related to no less than Keywords, Preset values ​​for operators, is the count index, 1 , n represents the total number of tags in the tag table; Will Associated with any tag in the tag table Keywords are compared to determine the tags with the largest proportion of the same keywords, and Binding; S25, yes Medium Repeat steps S23 to S24 for the information to be published other than the information to be published, and perform tag binding.

[0006] As a further solution of the present invention, in step 1, the specific manner of performing the distribution processing on each piece of information to be published according to the tags bound to different pieces of information to be published is as follows: Based on the determined sequence of information to be released The tags associated with all the information to be published; from Extract , and Add to the first processing queue and continue to obtain the information to be published , if the information is to be released The tag of the information to be published is the same as the tag of the information to be published in the first processing queue, then the information to be published included in the first processing cohort; Otherwise, the information will be released Add to the second processing queue and continue to obtain and the labels of the information to be published in the first processing queue and the second processing queue The signatures are compared separately and put into the corresponding processing queue; And so on, until All the information to be released are put into the processing queue; The total number of statistical processing queues is recorded as ,Will The processing queues are arranged from the first processing queue to the The order of processing queues is recorded as the processing queue sequence , among which 1 ; extract The tags associated with each processing queue in Tags, according to The order will be The tags are serialized to get ; .

[0007] As a further solution of the present invention, in step 2, based on the result of the distribution of the information to be released, the specific method of selecting the rule bases associated with different tags is as follows: Get the operator pre-built Table of tags The rule base associated with each tag in , where any rule base includes o verification rules pre-built by the operator; extract rule base and in The rule base corresponding to the tag: ; Sure Any tag in ,in is the count index, 1 ; For labels Processing queue ,Sure To process the queue The associated rule base.

[0008] As a further solution of the present invention, in step 2, the specific method of locking the normal information to be released and the abnormal information to be released is: for Any processing queue in Adopting a rule base The o verification rules in Review the information to be published for illegal content; like Any information to be published is passed If there are o verification rules in the , it means that the information to be released has passed the illegal content review and is determined to be normal information to be released; Otherwise, the information to be released is determined to be abnormal information to be released.

[0009] As a further solution of the present invention, in step three, if it is determined that the information to be published is abnormal information to be published, the abnormal information to be published together with the verification rules that the abnormal information to be published fails to pass will be fed back to the corresponding uploading user to remind the uploading user to make modifications.

[0010] As a further solution of the present invention, in step 3, the specific method of publishing the normal information to be published is: S71, confirm Any normal information to be released ,extract The label is marked as ,in, is the counting index, ; S72. Obtain all publishing platforms and record them in the order of acquisition as a publishing platform sequence ,in, is the total number of publishing platforms; S73, from Select any publishing platform ,in, is the counting index, ; S74, from Get the current time point As the end time point of an evaluation cycle, all the browsing tags are The target users of the published information constitute the target user set, where the duration of the evaluation cycle is determined by the operator; S75, determine that all target users in the target user set browse tags The average completion rate of published information is recorded as ; Then determine that all target users in the target user set have browsing tags Published The number of times the information is viewed, recorded as ; S76, will 、 Remove the dimension and take only the numerical part, and get it by the equal weighted average algorithm The target user pair label in preference index of published information; S77, repeat steps S73 to S76 to obtain The target user pairs in all publishing platforms are labeled The preference index of the published information is calculated, and the publishing platform with the largest preference index is selected as the optimal publishing platform; Determine the best time to publish Publish on the best publishing platform.

[0011] As a further solution of the present invention, in step 3, the best publishing time is determined to be the time for normal publishing of information. The specific methods for publishing on this optimal publishing platform are: The normal information to be released will be The associated optimal publishing platform is marked as ; from Get the label as The associated target user set; And extract the historical online time periods of all target users from the target user set; Divide the time of the day into x time periods and determine the average total number of online target users in any time period, where x is a value preset by the operator; Determine the time period with the largest average number of online target users, and use the start time of this time period as the optimal publishing platform for publishing. The best time to publish.

[0012] An information fully automatic publishing management system, the system includes: The information tagging module obtains the information to be published uploaded by the uploading user, forms a sequence of information to be published in the order of acquisition time, traverses the determined sequence of information to be published to perform tag binding, and distributes each information to be published according to the tags bound to different information to be published; The review module selects the platform rule base associated with each tag based on the results of the allocation of each piece of information to be released, and conducts a review of the illegal content of the information to be released that is bound to the corresponding tag. Based on the review results of the illegal content review, the normal information to be released and the abnormal information to be released are locked; The intelligent execution module feeds back the abnormal information to be released to the uploading user and reminds the uploading user to modify the illegal content; for the normal information to be released, the intelligent execution module obtains the label bound to the normal information to be released, evaluates the preference index of target users in different publishing platforms for the corresponding normal information to be released, and evaluates the optimal publishing platform based on the preference index, and releases the normal information to be released.

[0013] Beneficial effects of the present invention: (1) The core advantage of this application is that it realizes the full process automation management, from information uploading, label binding and allocation, to illegal content review based on the label rule library, and then to the automatic feedback of abnormal information to be released and the intelligent release decision of normal information to be released, which greatly reduces the cost and time consumption of manual intervention. While optimizing the upload user experience, the entire process also improves the allocation efficiency of release resources, and realizes efficient, compliant and accurate information release management; (2) This application achieves efficient organization and classification of information through the special design of the tag binding and allocation process. During the tag binding process, the cache window mechanism is used to obtain and serialize the information to be published, and the information text features are extracted by combining the video understanding model and audio spectrum analysis. Then, the text features are compared with the pre-built tag table to accurately match the most relevant tags. Then, queues are constructed based on the tags to achieve automatic classification of information with the same tags, effectively avoiding the tediousness and inefficiency of traditional manual classification, and reducing the risk of information misclassification due to human factors. (3) This application determines the rule base corresponding to each tag and conducts targeted review of information under specific tags, thus avoiding the problem of inaccurate review that may be caused by the general rule base; at the same time, it adopts preset verification rules to automatically determine whether the information to be released is qualified, which greatly reduces the workload of manual review and saves time and labor costs; this application automatically locks normal and abnormal information to be released, realizes the quality control of information release, ensures the compliance of release, improves the release efficiency, and enhances the reliability and security of information review; (4) When this application detects abnormal information to be published, it not only notifies the uploading user in a timely manner, but also clearly points out the violation verification rules to help users quickly correct the problem, reduce invalid publishing attempts, and improve publishing efficiency; for normal information, it adopts intelligent publishing decisions based on user behavior data, and accurately locates the optimal publishing platform and the best publishing time by analyzing the target user preference index and historical online time period of different publishing platforms, greatly improving the exposure rate and dissemination effect of information; this method optimizes the entire process of information release, reduces the cost of manual review and decision-making, and at the same time enhances the compliance of the published content and the user reception effect, which has significant practical value for both the information publisher and the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention will be further described below with reference to the accompanying drawings.

[0015] Figure 1 It is a schematic flow chart of the method of the present invention; Figure 2 Schematic diagram of the structure of the system of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0017] Example 1 A fully automatic information release management method, such as Figure 1 As shown, specifically including the following: Step 1: Acquire the information to be published uploaded by the uploading user, organize it into a sequence of information to be published in the order of acquisition time, traverse the determined sequence of information to be published to perform tag binding, and distribute and process each piece of information to be published according to the tags bound to different pieces of information to be published. Specifically, in the process of implementing this method, it relies on a fully automatic information publishing management system. This system interacts with several information publishing platforms (hereinafter referred to as publishing platforms) in real time and serves as an interactive medium between the uploading user (the user who uploaded the information) and the publishing platform. This system can acquire the information to be published uploaded by the uploading user in real time (at this time, the information to be published uploaded by the uploading user has not yet entered the publishing platform, but has entered the interactive medium between the uploading user and the publishing platform, that is, a fully automatic information publishing management system); This step involves obtaining the information to be published uploaded by the uploading user. This step requires defining a time period for obtaining the information to be published. This step provides a cache window period. , each cache window period The corresponding information to be published is obtained in each of them, among which the cache window period The operator shall determine this based on the actual situation (determined by the upload time requirements of the uploading user and the performance of the publishing platform); The following section is used as an example to give a solution for the cache window period in a specific situation, taking the current moment as a cache window period. The start time of Then, the cache window period determined above is The duration is added to the determined start time , you can get a cache window period The time point at which the duration ends and this time point is used as the cache window period The associated end time, marked as ; So far, a time length is determined as a cache window period The time period is a cache window period starting at the current time. , during this cache window Time range (that is, start time To end time (Inside) Get the information to be published uploaded by the uploading user and the total number of information to be published, and record the total number of information to be published as , and extract them separately The time when each uploading user associated with each piece of information to be published uploaded the information to be published, and The information to be published is serialized according to the upload time (sorted in the order of upload time), and the sorted result is recorded as the sequence of information to be published, which is expressed as: ; Based on the determined sequence of information to be released , from the first information to be released Execute traversal until the last information to be published ; The following steps are also included in the traversal process. Here we take the first information to be released as the example. As an example process: Because the information to be published may be in various forms, such as video, text or a combination of video and text, the information to be published is first separated by audio separation technology. The sound information is extracted from the video, the text information is extracted through text extraction technology, and finally the video part is extracted through video separation technology; Next, a pre-built video understanding model is used to extract video information text from the extracted video information; audio spectrum analysis is used to extract sound information text from the extracted sound information. These two steps are to convert both video information and sound information into text form. Finally, the information to be released is constructed by combining the video information text, the sound information text and the text information The associated text features are recorded as (It needs to be explained that text features It is necessary to split the keywords into multiple keywords to facilitate subsequent processing).

[0018] The audio separation technology, text extraction technology, video separation technology, video understanding model, and audio spectrum analysis involved in the above steps are all existing technologies. Among them, audio separation technology includes: spectral subtraction, non-negative matrix decomposition, and deep learning methods; text extraction technology includes: optical character recognition, natural language processing, and regular expressions; video separation technology includes: 3D-CNN, convolutional neural network model; video understanding models include: convolutional neural network (CNN), recurrent neural network (RNN) and its variants (such as LSTM, GRU); audio spectrum analysis includes: short-time Fourier transform (STFT), fast Fourier transform (FFT), and wavelet transform (WT).

[0019] Then, the operator obtains a tag table pre-built based on big data. The tag table contains all tags associated with any information to be published that has appeared before. When information to be published that has never appeared before appears, the tags associated with the information to be published will be automatically added to the tag table. The label table is represented as: , where any tag in the tag table Corresponding to no less than Keywords, Preset values ​​for operators, is a counting index, ranging from 1 to , n represents the total number of tags in the tag table; For example, there is a tag in the tag table called sports. The number of keywords associated with this tag is is 3, and the keywords are: football, basketball, Olympics; and the tags can be compatible with each other. For example, there is a tag called sports: the keywords are: football, basketball, swimming, and the keywords need to be associated with the corresponding tags.

[0020] In the above steps, the information to be released is obtained Associated text features , then the text features Compare with any tag and the associated keywords in the tag table (using string verification method) and determine the text features The keyword percentage of any tag in the tag table is the same as the keyword percentage, and the tag with the largest percentage of the same keyword is extracted, and this tag is associated with the information to be released. Binding; Exemplary text features Including: football, basketball, Olympics, then we can know the text features The proportion of keywords identical to sports tags is 100%, while text features The proportion of the same keywords as the sports tag is 66.67%. In comparison, the sports tag is the same as the text feature. If there is a tag with the largest proportion of the same keyword, then the sports tag will be added to the information to be released. Bind.

[0021] At this point, the information to be released has been determined Repeat the above steps for the associated tags to publish the information sequence All the information to be published in is processed and obtained The tags associated with each piece of information to be published.

[0022] This step also involves the allocation of each piece of information to be published. The specific method is as follows: In the above steps, the sequence of information to be released can be determined The tags associated with each piece of information to be published; Extract the first information to be published , and the information to be released Including the first processing queue, wherein the first processing queue is distinguished from other queues by a difference in labels; Then continue to obtain the information to be released Adjacent information to be released If the information to be released Tags and pending release information Same, then the information to be released Include information to be released The first processing queue; If the information is to be released Tags and pending release information If they are not the same, then the information to be released will be Included in the second processing queue, and the distinction method associated with the second processing queue depends on the information to be published Then, get the information to be released. , and the information to be released The tags are respectively related to the information to be released And information to be released Compare the labels and determine which processing queue to include; And so on, until All the information to be published is put into the processing queue, and then the total number of all processing queues is counted and recorded as , and then follow the order of building the processing queue (that is, the first processing queue, the second processing queue, ..., the Processing queue) will determine the The processing queues are recorded as processing queue sequences, which are expressed as: ,in, Greater than 0, less than or equal to ; Extract the processing queue sequence again The tags associated with each processing queue in The labels are matched with the number of processing queues, and then the The tags are serialized to obtain the processed Tags, represented as: , what needs to be explained here is: , The tags in Find the same tags.

[0023] Step 2: Based on the result after the information to be released is assigned, the rule bases associated with different tags are selected, and the information to be released that is bound to the corresponding tags is audited for illegal content. Based on the audit result of the illegal content audit, the normal information to be released and the abnormal information to be released are locked. Specifically, a pre-built tag table can be obtained in step 1, wherein each tag corresponds to a unique rule base, and the rule base has different verification rules for each tag. It should be noted that the rule base is constructed by the operator in combination with the actual situation and the pre-built tag table, and each rule base contains o pre-built verification rules. As above, the verification rules are also pre-built by the operator, which need to include: text content verification rules, picture content verification rules, video content verification rules and format verification rules. The detailed determination method of the verification rules is part of the existing technology, and will not be elaborated in this solution. Get and process queue sequences Associated Tags: , and extract from the pre-built rule base Tags The rule bases associated with each are expressed as: , and In order one by one, that is correspond , correspond ; Sure Any tag is marked as: ,in is a counting index, ranging from 1 to ; For labels The associated processing queue ,Sure To process the queue The associated rule base, which is then used to process the queue and processing queues Associated rule base Example processing performed: Utilizing the rule base The o verification rules in the processing queue sequence Any processing queue in Conduct review of illegal content; Among them, the processing queue It contains several messages to be published. If the processing queue Any information to be published passes the rule library during the review of illegal content If there are o verification rules in , it means that the information to be released has passed the review of illegal content and is considered normal information to be released; If the processing queue Any information to be published does not pass the rule library during the review of illegal content If any of the o verification rules in the , it means that the information to be released has not passed the illegal content review and the information to be released is regarded as abnormal information to be released.

[0024] Step 3: For the abnormal information to be released that has been determined, it is fed back to the uploading user, and the uploading user is reminded to modify the illegal content; for the normal information to be released that has been determined, the label bound to the normal information to be released is obtained, the preference index of the target users in different publishing platforms for the corresponding normal information to be released is evaluated, and the optimal publishing platform is evaluated based on the preference index, and the normal information to be released is released. Specifically, in step 2, the abnormal information to be released and the normal information to be released that have been determined after the illegal content review can be obtained. The normal information to be released enters the execution step of release, and for the abnormal information to be released, it is necessary to feed back the abnormal information to be released together with the verification rules that the abnormal information to be released fails to pass to the uploading user who uploaded the abnormal information to be released (if one verification rule fails, one feedback is given, and similarly, if multiple verification rules fail, multiple feedback is given), and the uploading user is informed that the illegal content in the uploaded abnormal information to be released needs to be modified.

[0025] Next, it is necessary to publish the normal information to be published. Based on step 1, the queue sequence can be obtained. , which includes the sequence of information to be released Each piece of information to be released in From the queue sequence The sequence of information to be released contained in Extract any normal information to be published and mark it as: , and then obtain the normal information to be released The bound tags are marked as: (same as described in the above steps Not for the same one, express Any label in ), where is a counting index, i ranges from 1 to ; It needs to be explained that in this step, the normal information to be released will be This is an example process. Other normal information to be released is also processed in the same way as described in this step.

[0026] Then, all publishing platforms that interact in real time with a fully automatic information publishing management system that this method relies on are extracted, and all the extracted publishing platforms are recorded as a publishing platform sequence in the order in which the publishing platforms are extracted, which is expressed as: ,in, Indicates the total number of publishing platforms; Then from the determined release platform sequence Determine any publishing platform Perform example processing, where is a counting index, ranging from 1 to ; Then from any of the determined publishing platforms Get the current time point (Different from the current moment ) as the end time point of an evaluation cycle, all browsed tags are Published information (here, it is to collect the relevant data of the target user on the published information. Because the normal information to be published has not been published yet, the target user cannot see it. Therefore, the historical browsing data of all users in the corresponding platform is used to determine all the users who have browsed the information tagged as The target users are users who have posted information as target users) and these users who have browsed the information are labeled as The target users of the published information constitute the target user set, and the duration of the evaluation cycle is determined by the operator based on the actual situation; Through this publishing platform Extract all user browsing tags in the target user set as The average completion rate of the published information is recorded as The average completion rate refers to the proportion of users who have watched (or read) information in its entirety. The average value is taken. If the average completion rate of a tag on a publishing platform is high, it can be interpreted as indicating that the information published with this tag is highly attractive to the target users on this publishing platform. Conversely, if the average completion rate is low, it indicates that the target users are not interested in the information published with this tag. From the publishing platform Extract all target users’ browsing tags in the target user set as The number of views of the published information is recorded as Here, browsing history refers to the target user clicking on the label The total number of times the information has been published and viewed, including the total number of times it is recommended to the target user for browsing (this data is related to the recommendation algorithm in the corresponding publishing platform and will not be further studied in this solution. We only focus on and record the number of times the target user browses it) and the total number of times the target user searches and browses it on his own; Next, get the average completion rate and number of views The value of is the pure value after removing the dimension, without considering the unit and dimension; Through the equal weighted average algorithm (average completion rate and number of views The calculation weight of the values ​​is 0.5) The average completion rate after removing the dimension and number of views The value of the publishing platform is calculated The target user pair label in The preference index of the published information (the calculation weights in the equal weighted average algorithm described here are not fixed and can be calculated by the operator for the average completion rate after removing the dimension) and number of views The values ​​of are assigned corresponding weights for weighted calculation, which requires that the sum of the two calculated weights is 1 and both calculated weights are greater than 0); So far, we have obtained the publishing platform The target user pair label in The preference index of the published information, and so on, to obtain the publishing platform sequence The target user pairs in all publishing platforms are labeled The preference index of the published information is preference index; and from the determined The value of the preference index is extracted from the preference indexes as the largest one, and the publishing platform associated with this preference index is used as the publishing label. The best publishing platform for normal information to be released; After determining the optimal publishing platform, we need to further determine the optimal publishing time to ensure that the information uploaded by the uploading user can receive the best views and attention after publishing (that is, be seen by more people); From the above steps, we can know that the normal information to be released is The associated optimal publishing platform is ; Then from the best publishing platform Get the label as The historical online time periods of all target users in the associated target user set (the reason is that the recommendation algorithms of each platform all serve the same principle: they will give priority to recommending information that the target user is interested in to the target user). Therefore, the normal information to be released is determined in the above steps. The label is , you can publish on the best platform Get the label as The associated target user set collects the historical online time periods of all target users. Based on the historical online time periods of all target users, it can be determined that the average total number of online target users in a certain hour of the 24 hours of the day is the largest (one hour is selected as an example here. In actual processing, the operator can adjust the unit time of the hour according to the actual situation, for example, adjust it to 30 minutes, or 10 minutes, or divide the time of the day into x time periods, and the specific length of each time period is determined according to the specific value of x, and x is determined by the operator). Then the start time of this hour is the time when this optimal publishing platform publishes the normal information to be published. The best time to publish information will be released normally Optimal publishing time and optimal publishing platform Publish in.

[0027] This embodiment describes a fully automatic information release management method, which aims to achieve automated and precise processing of information from acquisition to release. Its core purpose is to obtain and serialize information to be released through a cache window period, use multiple technologies to extract information features to construct text features and bind tags to form a processing queue; retrieve corresponding rules from the rule library based on the tags to conduct illegal content review and distinguish normal and abnormal information; feedback reminders are provided for abnormal information to be modified, and for normal information, the preference index of target users of different release platforms is evaluated to select the optimal release platform and time; this method aims to improve the efficiency, accuracy and effectiveness of information release, ensure that information can be effectively received by the target user group, and at the same time reduce manual intervention, improve the automation level of release management, and ultimately achieve efficient and precise delivery of information release, and enhance the pertinence and influence of information dissemination.

[0028] Example 2 An information automatic publishing management system, such as Figure 2 As shown, specifically including the following: The information tagging module interacts with several publishing platforms in real time and obtains the information to be published uploaded by the uploading users. It forms a sequence of information to be published in the order of acquisition time, traverses the determined sequence of information to be published, and performs tag binding. According to the tags bound to different information to be published, each information to be published is assigned and processed; The review module selects the platform rule base associated with each tag based on the results of the allocation of each piece of information to be released, and conducts a review of the illegal content of the information to be released that is bound to the corresponding tag. Based on the review results of the illegal content review, it locks the normal information to be released and the abnormal information to be released; The intelligent execution module, for the identified abnormal pending information, will feedback it to the uploading user and remind the uploading user to modify the illegal content; for the identified normal pending information, it will obtain the tags bound to the normal pending information, evaluate the preference index of target users on different publishing platforms for the corresponding normal pending information, and evaluate the optimal publishing platform based on the preference index, and publish the normal pending information; Central processing module, used to support and implement the analysis steps and calculation steps described in any module of this system; The data storage module is used to store the fully automatic information publishing management method described in Example 1, and is also used to store the analysis steps, calculation steps, analysis results of the analysis steps, and calculation results of the calculation steps involved in the fully automatic information publishing management method described in Example 1.

[0029] Example 3 As the third embodiment of the present invention, the focus is on implementing the first embodiment and the second embodiment in combination.

[0030] Some of the data in the formulas described above are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0031] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in similar ways. As long as they do not deviate from the invention or exceed the scope defined by the claims, they should all fall within the scope of protection of the present invention.

[0032] It is important to note that all user data collected in this application is collected with the user's consent and authorization. Furthermore, the use of user data is legal and compliant, and the use and processing of user data complies with the relevant laws, regulations, and standards of the relevant regions.

Claims

1. A fully automatic information publishing management method, characterized in that: This method comprises the following steps: Step 1: Obtain the information to be published uploaded by the uploading user, organize the information to be published into a sequence in the order of the acquisition time, traverse the determined sequence of information to be published to perform tag binding, and distribute the information to be published according to the tags bound to different information to be published; Step 2: Based on the results of the allocation of the information to be released, the rule bases associated with different tags are selected, and the information to be released that is bound to the corresponding tags is audited for illegal content. Based on the audit results of the illegal content audit, the normal information to be released and the abnormal information to be released are locked; Step 3: For the abnormal information to be released that has been determined, feedback is given to the uploading user, and the uploading user is reminded to modify the illegal content; for the normal information to be released that has been determined, the label bound to the normal information to be released is obtained, the preference index of the target users on different publishing platforms for the corresponding normal information to be released is evaluated, and the optimal publishing platform is evaluated based on the preference index, and the normal information to be released is released.

2. The method for fully automatic information release management according to claim 1, characterized in that: In step 1, the specific method of traversing the determined sequence of information to be published and performing tag binding is: S21. Obtain the cache window period preset by the operator , using the current time as the cache window period The start time is recorded as ; At On top of that, duration, get a The end time is recorded as ; S22, take to The total number of information to be published uploaded by users is recorded as , According to the upload time of the information to be published Sort the information to be released and get the sequence of information to be released ; S23, from Start traversal ; from Separate the video information, the sound information and the text information, and extract the video information text from the video information; Extracting sound information text from sound information; Combine video information text, sound information text and text information to build The text features of ; S24. Obtain the label table pre-built by the operator , where any label Related to no less than Keywords, Preset values ​​for operators, is the count index, 1 , n represents the total number of tags in the tag table; Will Associated with any tag in the tag table Keywords are compared to determine the tags with the largest proportion of the same keywords, and Binding; S25, yes Medium Repeat steps S23 to S24 for the information to be published other than the information to be published, and perform tag binding.

3. The method for fully automatic information release management according to claim 2, characterized in that: In step 1, the specific method of performing distribution processing on each piece of information to be published according to the tags bound to the different pieces of information to be published is as follows: Based on the determined sequence of information to be released The tags associated with all the information to be published; from Extract , and Add to the first processing queue and continue to obtain the information to be published , if the information is to be released The tag of the information to be published is the same as the tag of the information to be published in the first processing queue, then the information to be published included in the first processing cohort; Otherwise, the information will be released Add to the second processing queue and continue to obtain , and compare them with the tags of the information to be published in the first processing queue and the second processing queue respectively, and put them into the corresponding processing queue; And so on, until All the information to be released are put into the processing queue; The total number of statistical processing queues is recorded as ,Will The processing queues are arranged from the first processing queue to the The order of processing queues is recorded as the processing queue sequence , among which 1 ; extract The tags associated with each processing queue in Tags, according to The order will be The tags are serialized to get ; 。 4. The method for fully automatic information release management according to claim 3, characterized in that: In step 2, based on the result of the distribution of the information to be released, the specific method of selecting the rule bases associated with different tags is as follows: Get the operator pre-built Table of tags The rule base associated with each tag in , where any rule base includes o verification rules pre-built by the operator; extract rule base and in The rule base corresponding to the tag: ; Sure Any tag in ,in is the count index, 1 ; For labels Processing queue ,Sure To process the queue The associated rule base.

5. A method for fully automatic information release management according to claim 4, wherein The characteristic is that, in the step 2, the specific method of locking the normal information to be released and the abnormal information to be released is: for Any processing queue in Adopting a rule base The o verification rules in Review the information to be published for illegal content; like Any information to be published is passed If there are o verification rules in the , it means that the information to be released has passed the illegal content review and is determined to be normal information to be released; Otherwise, the information to be released is determined to be abnormal information to be released.

6. The method for fully automatic information release management according to claim 5, characterized in that: In step three, if the information to be published is determined to be abnormal information to be published, the abnormal information to be published together with the verification rules that the abnormal information to be published fails to pass will be fed back to the corresponding uploading user to remind the uploading user to make modifications.

7. The method for fully automatic information release management according to claim 5, characterized in that: In step 3, the specific method for publishing the normal information to be published is: S71, confirm Any normal information to be released ,extract The label is marked as ,in, is the counting index, ; S72. Obtain all publishing platforms and record them in the order of acquisition as a publishing platform sequence ,in, is the total number of publishing platforms; S73, from Select any publishing platform ,in, is the counting index, ; S74, from Get the current time point As the end time point of an evaluation cycle, all the browsing tags are The target users of the published information constitute the target user set, where the duration of the evaluation cycle is determined by the operator; S75, determine that all target users in the target user set browse tags The average completion rate of published information is recorded as ; Then determine that all target users in the target user set have browsing tags The number of views of the published information is recorded as ; S76, will 、 Remove the dimension and take only the numerical part, and get it by the equal weighted average algorithm The target user pair label in preference index of published information; S77, repeat steps S73 to S76 to obtain The target user pairs in all publishing platforms are labeled The preference index of the published information is calculated, and the publishing platform with the largest preference index is selected as the optimal publishing platform; Determine the best time to publish Publish on the best publishing platform.

8. The method for fully automatic information release management according to claim 7, characterized in that: In step 3, determine the best time to publish information. The specific methods for publishing on this optimal publishing platform are: The normal information to be released will be The associated optimal publishing platform is marked as ; from Get the label as The associated target user set; And extract the historical online time periods of all target users from the target user set; Divide the time of the day into x time periods and determine the average total number of online target users in any time period, where x is a value preset by the operator; Determine the time period with the largest average number of online target users, and use the start time of this time period as the optimal publishing platform for publishing. The best time to publish.

9. An information automatic publishing management system, characterized in that: This system includes: Information labeling module, obtains the information to be published uploaded by the uploading user, press The order of the acquisition time is used to form a sequence of information to be released, the determined sequence of information to be released is traversed to perform tag binding, and each piece of information to be released is assigned according to the tags bound to different pieces of information to be released; The review module selects the platform rule base associated with each tag based on the results of the allocation of each piece of information to be released, and conducts a review of the illegal content of the information to be released that is bound to the corresponding tag. Based on the review results of the illegal content review, the normal information to be released and the abnormal information to be released are locked; The intelligent execution module feeds back the abnormal information to be released to the uploading user and reminds the uploading user to modify the illegal content; for the normal information to be released, the intelligent execution module obtains the label bound to the normal information to be released, evaluates the preference index of target users in different publishing platforms for the corresponding normal information to be released, and evaluates the optimal publishing platform based on the preference index, and releases the normal information to be released.

Images