Content processing method and device, electronic equipment and medium

Abstract

The embodiment of the invention relates to a content processing method and device, electronic equipment and a medium. The method comprises the following steps: acquiring to-be-published content, wherein the to-be-published content comprises at least one of a text, an image or a video; the method further comprises the step of obtaining a plurality of related contents associated with the to-be-published contents based on the content characteristics of the to-be-published contents. In addition, the method further comprises the step of determining the publishing state of the to-be-published content based on the publishing states of the multiple related contents. According to the method disclosed by the embodiment of the invention, the publishing state of the to-be-published content is determined by acquiring the publishing state of the plurality of related contents associated with the to-be-published content, so that the effect of identifying the low-quality content can be improved, the missing of the low-quality content and the misjudgment of the forward beneficial content are reduced, the homogenized low-quality content is screened out, and the quality of the to-be-published content is improved. Therefore, the user experience is improved.

Description

Technical Field

[0001] The present disclosure relates generally to the field of computers, and more particularly, to methods, apparatuses, electronic devices, and media for content processing. Background Art

[0002] In the digital age, information distribution has become increasingly important. However, since information is very complex and contains a lot of homogeneous, low-quality content, it is necessary to filter out the homogeneous, low-quality content before distributing it to users. Content processing is an important way to filter out homogeneous, low-quality content. Before information is distributed, content processing filters the information based on its content to ensure that most of the homogeneous, low-quality content can be filtered out.

[0003] At the same time, according to relevant statistical data and user habits, information distribution is prone to concentrated distribution changes. This phenomenon is more obvious during holidays or when relevant hot spots appear, resulting in a large amount of homogeneous low-quality content, which greatly affects the user experience. How to quickly process homogeneous low-quality content in batches and maintain a high-quality and diverse space for users has become an urgent need. Summary of the invention

[0004] The embodiments of the present disclosure provide a method, apparatus, electronic device and medium for content processing, which determine the publishing status of content to be published by acquiring the publishing status of multiple related content associated with the content to be published, thereby improving the effect of identifying low-quality content, reducing the omission of low-quality content and the misjudgment of positive and beneficial content, and screening out homogeneous low-quality content, thereby improving user experience.

[0005] In a first aspect of the present disclosure, a method for content processing is provided. The method includes obtaining content to be published, wherein the content to be published includes at least one of text, image, or video. The method also includes obtaining multiple related content associated with the content to be published based on content features of the content to be published. In addition, the method also includes determining the publishing status of the content to be published based on the publishing status of the multiple related content.

[0006] In a second aspect of the present disclosure, a device for content processing is provided. The device includes a publishing content acquisition module configured to acquire content to be published, and the content to be published includes at least one of text, image, or video. The device also includes a related content acquisition module configured to acquire multiple related content associated with the content to be published based on content features of the content to be published. In addition, the device also includes a publishing status determination module configured to determine the publishing status of the content to be published based on the publishing status of the multiple related content.

[0007] In a third aspect of the present disclosure, an electronic device is provided. The electronic device includes one or more processors; and a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement a method for content processing. The method includes obtaining content to be published, and the content to be published includes at least one of text, image, or video. The method also includes obtaining multiple related content associated with the content to be published based on content features of the content to be published. In addition, the method also includes determining the publishing status of the content to be published based on the publishing status of the multiple related content.

[0008] In a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, a method for content processing is implemented. The method includes obtaining a sequence of objects of historical user interactions, and the sequence of objects is sorted by interaction time. The method includes obtaining content to be published, and the content to be published includes at least one of text, image, or video. The method also includes obtaining multiple related content associated with the content to be published based on content features of the content to be published. In addition, the method also includes determining the publishing status of the content to be published based on the publishing status of the multiple related content.

[0009] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to limit the scope of the claimed subject matter. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0011] Figure 1 A schematic diagram illustrating an example environment in which various embodiments of the present disclosure may be implemented;

[0012] Figure 2 A flowchart of a method for content processing according to some embodiments of the present disclosure is shown;

[0013] Figure 3 A schematic diagram showing a process of determining the publishing status of to-be-published content based on statistical information of related content according to some embodiments of the present disclosure;

[0014] Figure 4 is a schematic diagram of a process of training a prediction model according to an embodiment of the present disclosure;

[0015] Figure 5A schematic diagram showing a process of determining the publishing status of to-be-published content based on clustering information of related content according to some embodiments of the present disclosure;

[0016] Figure 6 is a flowchart of a process of updating a clustering query table according to an embodiment of the present disclosure;

[0017] Figure 7 is a schematic diagram of a process of determining the publishing status of to-be-published content based on statistical information and clustering information of related content according to an embodiment of the present disclosure;

[0018] Figure 8 A block diagram showing an apparatus for content processing according to some embodiments of the present disclosure; and

[0019] Fig. 9 A block diagram of an electronic device according to some embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0020] It is understood that all user-related data involved in this technical solution should be obtained and used after the user's authorization. This means that in this technical solution, if the user's personal information needs to be used, the user's explicit consent and authorization are required before obtaining this data, otherwise the relevant data will not be collected and used. It should also be understood that when implementing this technical solution, relevant laws and regulations should be strictly observed during the collection, use and storage of data, and necessary technologies and measures should be taken to protect the user's data security and ensure the safe use of data.

[0021] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0022] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects, unless explicitly stated. Other explicit and implicit definitions may also be included below.

[0023] Nowadays, various content platforms play an important role in daily life, and a huge amount of information is distributed to users on content platforms. However, due to the huge amount of information, it is difficult to filter out even if a large amount of homogeneous low-quality content is reset. However, in order to protect users and improve their experience, as well as to strengthen the information security of the platform, it is essential to filter out homogeneous low-quality content.

[0024] The common approach now is to use machine models to automatically predict the content to be published based on the content features of the content to be published. However, due to the poor generalization of the model for location data and the periodicity and unpredictability of low-quality content, the model needs to be iterated quickly and rely on a large amount of data annotation. Otherwise, it is very easy to cause the omission of homogeneous low-quality content or the misjudgment of positive and beneficial content, which greatly affects the user experience and is not conducive to the distribution and interaction of information.

[0025] Therefore, the present disclosure aims to utilize the high similarity and high homogeneity of window period data in information distribution to establish a screening system for content to be released, while reducing the reliance on content features themselves, so as to avoid the impact of the periodicity and unpredictability of low-quality content on the system. At the same time, it avoids the need for rapid iteration and reliance on a large amount of annotated data to reduce the cost of using the system. Furthermore, it improves the prediction effect of the system, reduces the omission of low-quality content and the misjudgment of positive and beneficial content.

[0026] To this end, an embodiment of the present disclosure proposes a method for content processing. The method obtains multiple related contents associated with the content to be published based on the content features of the content to be published, and then determines the publishing status of the content to be published based on the publishing status of the multiple related contents. Since the publishing status of the content to be published is mainly based on the publishing status of the related content, it is avoided to use the content features of the content to be published itself to make judgments. According to some embodiments of the present disclosure, it can be based on the features of multiple related contents rather than the content features of the content to be published itself, so as to avoid the influence of the periodicity and unpredictability of low-quality content on the system. In addition, since it is based on multiple related contents, more reference information is provided, avoiding relying solely on the information of the content to be published itself, improving the robustness of the system, avoiding the interference of random factors, so as to improve the prediction effect of the system, reduce the omission of low-quality content and the misjudgment of positive and beneficial content.

[0027] Figure 1 1 is a schematic diagram of an example environment 100 in which various embodiments of the present disclosure may be implemented. Figure 1As shown, the example environment 100 includes content to be published 110, a related content acquisition module 120, a content library 130, and a decision module 140. The content to be published 110 can be different content types, including but not limited to text, images, or videos, and combinations thereof. For example, in some embodiments, the content to be published can be pure text content such as news, blogs, or a combination of text and images, or video content such as long videos or short videos. Since the amount of content to be published 110 is huge in the content publishing platform, including a large amount of high-quality content, but also including a large amount of homogenized low-quality content, these contents once published will greatly affect the user experience. Therefore, in the information distribution of the content platform, it is necessary to process the content to be published 110 to determine the publishing status of the content to be published 110. For high-quality content, it should be allowed to be published, while for homogenized low-quality content, it is not allowed to be published.

[0028] The example environment 100 also includes a related content acquisition module 120. The related content acquisition module 120 acquires multiple contents associated with the content to be published 110 from the content library 130 based on the content features of the content to be published 110. In some embodiments, the related content acquisition module 120 can acquire the content features of the content to be published 110, and acquire the content features of all contents 131-1 to 131-N in the content library 130, where N identifies the number of contents in the content library 130, and then acquire multiple contents 131-1 to 131-N that are most relevant to the content to be published 110 through the relevance of these content features. In some embodiments, the related content acquisition module 120 can acquire vectorized content features of the content to be published 110. The vectorized content features can be acquired based on the content of the content to be published 110. For example, when the content to be published 110 is text, the vectorized content features corresponding to the text of the content to be published 110 can be acquired through a vectorization tool. Similarly, when the content 110 to be published is text, image, or video and any combination thereof, the vectorized content features corresponding to the content 110 to be published can also be obtained through a vectorization tool. The vectorization tool can be any machine model or algorithm, and the present disclosure does not limit this. In addition, the related content acquisition module 120 can obtain the vectorized content features of all contents 131-1 to 131-N in the content library 130. By respectively calculating the vector distance of the vectorized content features of the content 110 to be published and the vectorized content features of all contents 131-1 to 131-N, a plurality of contents 131-1 to 131-N that are most relevant to the content 110 to be published can be obtained. For example, the M contents that are most relevant to the content 110 to be published can be obtained from the contents 131-1 to 131-N, and M can be 5, 10, 15 or other numbers not greater than the total number of contents in the content library 130, and the present disclosure does not limit this.

[0029] As shown in the example environment 100, each content in the content library 130 includes its corresponding publishing status, for example, the content 131-1 includes its corresponding status 132-1, the content 131-2 includes its corresponding status 132-2, and the content 131-N includes its corresponding status 132-N. That is, the status of each content stored in the content library 130 is determined, so it provides a reference for determining the status of the content 110 to be published. The contents 131-1 to 131-N are different, representing different contents to be published, and their content types may also be different, and their corresponding publishing statuses 132-1 to 132-N may be the same or different. In some embodiments, the publishing status may be two types: allowing publishing and not allowing publishing. For example, the publishing status 132-1 may be allowing publishing, and the publishing status 132-2 may be not allowing publishing, and other publishing statuses also belong to one of allowing publishing and not allowing publishing. However, in other embodiments, the publishing status may be multiple, such as allowed publishing, recommended publishing, not recommended publishing, and not allowed publishing, or other situations, which mainly depends on the settings of the content platform, and the present disclosure does not limit it here.

[0030] After the related content module 120 obtains the publishing status of multiple related contents associated with the content to be published 110, the decision module 140 can determine the publishing status of the content to be published 110 through the publishing status of the multiple related contents. For example, in some embodiments, the publishing status of the content to be published 110 can be determined by the publishing status that accounts for the largest proportion among the multiple related contents. In some embodiments, multiple related content historical data can be obtained, such as the number of likes and dislikes of multiple related contents, and the publishing status of the content to be published 110 can be determined in combination with the publishing status of the multiple related contents. In this way, when determining the publishing status of the content to be published 110, the decision module 140 is based only on the publishing status of the multiple related contents and other features, without the need to make a judgment based on the content characteristics of the content to be published 110 itself, so as to avoid the influence of the periodicity and unpredictability of low-quality content on the decision module 140.

[0031] Figure 2A flow chart of a method 200 for content processing according to some embodiments of the present disclosure is shown. In box 202, method 200 obtains content to be published, and the content to be published includes at least one of text, image, or video. For example, the content to be published is content in a content platform, and it is necessary to determine whether the content can be published before distributing it to users. For example, when the content to be published is low-quality content, it cannot be published. In the content platform, the content to be published can be at least one of text, image, or video, for example, it can be a separate text, image, or video, or a combination of text and image, or a combination of text and video, etc., and the present disclosure does not limit this.

[0032] In block 204, based on the content characteristics of the content to be published, a plurality of related contents associated with the content to be published are obtained. In some embodiments, the content characteristics of the content to be published are obtained, and all contents in the content library (e.g., Figure 1 The content features of the content 131-1 to 131-N shown in the figure are then used to obtain the multiple contents in the content library that are most relevant to the content to be published. In some embodiments, the vectorized content features of the content to be published can be obtained. The vectorized content features can be obtained based on the content of the content to be published. For example, when the content to be published is text, the vectorized content features corresponding to the text of the content to be published can be obtained through a vectorization tool. Similarly, when the content to be published is text, image, or video and any combination thereof, the vectorized content features corresponding to the content of the content to be published can also be obtained through a vectorization tool. The vectorization tool can be any machine model or algorithm, and the present disclosure does not limit this. In addition, the vectorized content features of all the contents in the content library can be obtained. By respectively calculating the vector distance between the vectorized content features of the content to be published and the vectorized content features of all the contents, the multiple contents in the content library that are most relevant to the content to be published can be obtained.

[0033] In block 206, the publishing status of the to-be-published content is determined based on the publishing status of the plurality of related content. For example, each content in the content library includes its corresponding publishing status (e.g., Figure 1132-1 to 132-N shown). That is to say, the state of each content stored in the content library is determined, so as to provide a reference for determining the state of the content to be published. Each content in the content library is different, representing different content to be published, and its content type may also be different, and its corresponding publishing state may be the same or different. In some embodiments, the publishing state may be two kinds of allowing publishing and not allowing publishing, but in other embodiments, the publishing state may be multiple, such as allowing publishing, suggesting publishing, not suggesting publishing and not allowing publishing, or other situations, mainly depending on the setting of the content platform, which is not limited by the present disclosure. After obtaining the publishing state of multiple related contents associated with the content to be published, the method 200 may determine the publishing state of the content to be published by the publishing state of multiple related contents. For example, in some embodiments, the publishing state of the content to be published may be determined by the publishing state with the largest proportion among multiple related contents. In some embodiments, multiple related content historical data may be obtained, such as the number of likes and the number of dislikes of multiple related contents, and the publishing state of the content to be published may be determined in combination with the publishing state of multiple related contents.

[0034] according to Figure 2 The method 200 shown, when determining the publishing status of the content to be published, is based only on the publishing status and other features of multiple related contents, without the need for content features of the content to be published itself, so as to avoid the influence of the periodicity and unpredictability of low-quality content on the process of determining the publishing status. At the same time, a comprehensive judgment is made through the publishing status of multiple related contents, avoiding judging the publishing status only based on the content to be published itself, improving the robustness of the whole process, and improving the ability to identify low-quality content.

[0035] Figure 3 A schematic diagram of a process 300 for determining the publishing status of content to be published based on statistical information of related content according to some embodiments of the present disclosure is shown. In process 300, the content to be published 302 is first obtained. For example, the content to be published is content in a content platform. Before distributing it to users, it is necessary to determine whether the content can be published. For example, when the content to be published is low-quality content, it cannot be published. In the content platform, the content to be published can be at least one of text, image, or video, for example, it can be a separate text, image, or video, or a combination of text and image, or a combination of text and video, etc., and the present disclosure does not limit this.

[0036] The content features of the content to be published 302 are extracted by the feature extraction module 304, and the content features are related to the content of the content to be published itself. For example, in some embodiments, the content to be published is news information, so the classification category, multiple keywords, publishing account, article topic, etc. of the news information can be extracted as content features. In some embodiments, the content to be published can be converted from the original text, picture or video to a vector as the content feature of the content to be published by a vectorization tool, and the vector is associated with the content to be published to represent the content to be published in the vector dimension. For example, when two contents are related, their corresponding vectors are also close in distance. The vectorization tool can be any suitable machine learning model, which can receive the content to be published as input, and can output a corresponding vector to represent the input content to be published.

[0037] The related content acquisition module 306 receives the content features of the content 302 to be published from the content feature extraction module 304, and acquires multiple related contents that are most related to the content 302 to be published from the content library 308 based on the content features. For example, in some embodiments, the content to be published is news information, wherein the content features of the content to be published are classification categories, multiple keywords, publishing accounts, article topics, etc., then the related content acquisition module 306 searches for multiple related features that are most related to the content 302 to be published from the content library 308 based on these content features. For example, the multiple related contents can be content that matches the content to be published in one or more of the classification categories, multiple keywords, publishing accounts, and article topics. In some embodiments, the content features of the content 302 to be published can be vectors corresponding to its content, and the related content acquisition module 306 searches for multiple vectors similar to the content 302 to be published in the content library 308. The similarity between vectors can be represented by the distance, angle, and correlation coefficient of the vectors, which is not limited in the present disclosure.

[0038] The content library 308 stores information such as content features, historical data, and publishing status of multiple contents. For example, the content features of the content to be published 302 are stored in the content library 308, and after the publishing status of the content to be published 302 is determined, its publishing status is also stored in the content library 308 accordingly. In some embodiments, the content library 308 has timeliness, for example, some expired content is regularly cleared from the content library 308 to reduce storage costs, and also prevent the relevant content acquisition module 306 from acquiring expired content from the content library 308.

[0039] The statistical feature construction module 310 constructs statistical features for the content to be published 302 based on the information of multiple related contents. In some embodiments, based on the statistical features of all related contents and the corresponding publishing status, statistical features are constructed for the content to be published 302. Various similarity statistical information can be constructed as statistical features using the content with the same publishing status in multiple related contents, such as the maximum similarity, minimum similarity, average similarity, etc. of the related contents whose status is allowed to be published, and the maximum similarity, minimum similarity, average similarity, etc. of the related contents whose status is not allowed to be published, so as to construct statistical features as statistical features of the content to be published. In addition, historical data of multiple related contents can also be counted, such as the average number of viewers, average number of likes, and other positive content information statistics of multiple related contents, or the number of dislikes, number of reports, and other negative content information statistics of multiple related contents, so as to construct statistical features of historical data for the content to be published 302. The above method is only an example of constructing statistical features, and other information of multiple related contents can also be used to construct statistical features for the content to be published 302, which is not limited by the present disclosure.

[0040] After obtaining the statistical features of the content to be published 302 through the statistical feature construction module 310, the prediction model 314 can be called by the prediction model calling module 312 to input the statistical features into the prediction model 314. In some embodiments, the prediction model 314 can be a tree model, such as an XGBoost model, which is a boosted tree model, so it integrates many tree models together to form a classifier with good results. In some embodiments, the prediction model 314 can be other machine learning or deep learning models, which are not limited by the present disclosure.

[0041] After calling the prediction model 314 , the prediction score acquisition module 316 obtains the prediction model 314 's prediction of the score of the content to be published, and the score is input into the decision module 318 , which compares the score with a predetermined threshold to determine the publishing status of the content to be published 312 .

[0042] It can be seen that when the prediction model 314 is used to determine the score of the content to be published 302, the statistical features used are constructed through multiple related contents associated with the content to be published 302, avoiding the problem of insufficient information caused by only using the features of the content to be published 302, and improving the ability to identify low-quality content. At the same time, using statistical features instead of content features for prediction avoids the impact of the periodicity and unpredictability of low-quality content, so that the prediction model 314 has better generalization and prediction effect.

[0043] Figure 44 is a schematic diagram of a process 400 for training a prediction model according to an embodiment of the present disclosure. The content publishing library 402 stores all the content that has been recently published, which may be text, images, videos, or any combination thereof. The content feature extraction model 404 extracts the content features of all the content in the content publishing library 402, such as Figure 3 As described, content features can be in various forms, such as category features, keyword features, theme features, etc., or can be vectorized representations of content. The related content 406 acquisition module obtains multiple related contents for each content in the content publishing library 402, and obtains historical data, publishing status, and other information of the multiple related contents. The statistical feature construction module 410 uses the information of multiple related contents to construct statistical features for each content in the content publishing library 402. After the statistical features are constructed, combined with the publishing status corresponding to each content (i.e., label), the training set data and the validation set data can be obtained. Then the prediction model training module 412 uses the training set data to train the prediction model. For example, in some embodiments, the prediction model can be a tree model. The tree model can be trained according to the statistical features of multiple contents in the training set data (i.e., the input features of the training data) and their corresponding publishing status (i.e., the label of the training data) to obtain a model with better effect. The threshold selection module 414 is used to perform reasoning on the validation set data. For example, some different thresholds can be used to obtain the precision and recall rates under the corresponding thresholds for use in the decision-making process (for example, such as Figure 3 The decision module 318 shown in the figure may select an appropriate threshold value according to the scenario.

[0044] From the above, we can see that the prediction model is trained mainly by using the statistical features constructed from multiple related contents of each content, rather than using content features to train the prediction model. This avoids the impact of the periodicity and unpredictability of low-quality content. Then, there is no need to train the prediction model frequently, and the prediction model can also have a good prediction effect, saving the training cost.

[0045] Figure 5 A schematic diagram of a process 500 for determining the publishing status of content to be published based on clustering information of related content according to some embodiments of the present disclosure is shown. In process 500, the content to be published 502 is first obtained. For example, the content to be published is content in a content platform. Before distributing it to users, it is necessary to determine whether the content can be published. For example, when the content to be published is low-quality content, it cannot be published. In the content platform, the content to be published can be at least one of text, image, or video, for example, it can be a separate text, image, or video, or a combination of text and image, or a combination of text and video, etc., and the present disclosure does not limit this.

[0046] The content features of the content to be published 502 are extracted by the feature extraction module 504, and the content features are related to the content of the content to be published itself. For example, in some embodiments, the content to be published is news information, so the classification category, multiple keywords, publishing account, article theme, etc. of the news information can be extracted as content features. In some embodiments, the content to be published can be converted from the original text, picture or video to a vector as the content feature of the content to be published by a vectorization tool, and the vector is associated with the content to be published to represent the content to be published in the vector dimension. For example, when two contents are related, their corresponding vectors are also close in distance. The vectorization tool can be any suitable machine learning model, which can receive the content to be published as input, and can output a corresponding vector to represent the input content to be published.

[0047] The related content acquisition module 506 receives the content features of the content 502 to be published from the content feature extraction module 504, and acquires multiple related contents that are most related to the content 502 to be published from the content library 508 based on the content features. For example, in some embodiments, the content to be published is news information, wherein the content features of the content to be published are classification categories, multiple keywords, publishing accounts, article topics, etc., then the related content acquisition module 306 searches for multiple related features that are most related to the content 502 to be published from the content library 508 based on these content features. For example, the multiple related contents can be content that matches the content to be published in one or more of the classification categories, multiple keywords, publishing accounts, and article topics. In some embodiments, the content features of the content 502 to be published can be vectors corresponding to its content, and the related content acquisition module 506 searches for multiple vectors similar to the content 502 to be published in the content library 508. The similarity between vectors can be represented by the distance, angle, and correlation coefficient of the vectors, which is not limited in the present disclosure.

[0048] The content library 508 stores information such as content features, historical data, clustering information, and publishing status of multiple contents. For example, the content features of the content to be published 502 are stored in the content library 508, and after the publishing status of the content to be published 502 is determined, its publishing status is also stored in the content library 508 accordingly. In some embodiments, the content library 308 has timeliness, for example, some expired content is regularly cleared from the content library 508 to reduce storage costs, and also prevent the related content acquisition module 506 from acquiring expired content from the content library 508.

[0049] The clustering feature construction module 510 constructs clustering features for the content to be published 502 based on the clustering information of the multiple related contents. In some embodiments, the clustering features are constructed for the content to be published 502 based on the clustering features and the corresponding publishing status of all the related contents. For example, the clustering features can be constructed using the clustering categories of each related content in the multiple related contents and their corresponding publishing status. The clustering category determination module 512 uses the clustering features to determine the clustering category of the content to be published 502 through the clustering query table 514.

[0050] After the clustering category of the content 502 to be released is determined, the content 502 to be released can be scored by the clustering score acquisition module 516. For example, if the clustering category of the content 502 to be released is C, and in the content library 508, there are M contents corresponding to the clustering category C, and the publishing status of M1 contents is allowed to be released, while the publishing status of M2 contents is not allowed to be released, then the clustering score of the content 502 to be released can be obtained as M1 / M. In other words, the clustering score of the content 502 to be released is determined by all the contents with the same clustering category as the content 502 to be released and their publishing status. The clustering score is input into the decision module 518, and the score is compared with a predetermined threshold to determine the publishing status of the content 512 to be released.

[0051] It can be seen that when determining the cluster score of the content to be released 502, the cluster features used are constructed by using multiple related contents associated with the content to be released 502, avoiding the problem of insufficient information caused by using only the features of the content to be released 502, and improving the ability to identify low-quality content. At the same time, cluster features are used instead of content features for prediction, and the release status of related content is used to "guide" the determination of the release status of the content to be released 502, avoiding the impact of the periodicity and unpredictability of low-quality content.

[0052] Figure 66 is a flowchart of a process 600 of updating a clustering query table according to an embodiment of the present disclosure. The clustering information of each content can be obtained through the clustering query table. In box 602, the clustering result of the previous clustering is obtained. For example, all nodes (i.e., published content) and edges (i.e., the association relationship between all content) in the previous clustering are obtained. In box 604, the latest published content is obtained. For example, after the clustering query table was last updated, a plurality of new contents were published, but they were not updated to the clustering query table, so they need to be processed. In box 606, expired nodes and edges in the clustering graph are deleted. For example, in some content platforms, many contents are time-sensitive, and they need to be deleted and cannot be published after expiration, so they need to be deleted from the clustering graph. In box 608, clustering operations are performed. For example, in some embodiments, a graph clustering method can be used to perform clustering operations, such as using vectors in a vector library to construct an undirected graph, connecting each content as a node and vector distance as the weight of the edge, and then clustering using a community discovery algorithm. In block 610, the latest clustering result is merged with the previous clustering result to ensure that the clustering category of the same content does not change over time. In block 612, the merged clustering result is written into a clustering query table. In some embodiments, the clustering query table can be in the form of a key-value pair to facilitate querying by other tasks.

[0053] Figure 7 It is a schematic diagram of process 700 for determining the publishing status of content to be published based on statistical information and clustering information of related content according to an embodiment of the present disclosure. As shown in process 700, and with reference to process 300, first obtain the content to be published 702. For example, the content to be published is content in a content platform. Before distributing it to users, it is necessary to determine whether the content can be published. For example, when the content to be published is low-quality content, it cannot be published. In the content platform, the content to be published can be at least one of text, image, or video. For example, it can be a separate text, image, or video, or a combination of text and image, or a combination of text and video, etc. The present disclosure does not limit this.

[0054] Then, the content feature extraction module 704 is used to extract the content features of the content to be published 702, and the content features are related to the content of the content to be published itself. For example, in some embodiments, the content to be published is news information, so the classification category, multiple keywords, publishing account, article topic, etc. of the news information can be extracted as content features. In some embodiments, the content to be published can be converted from the original text, picture or video to a vector as the content feature of the content to be published through a vectorization tool, and the vector is associated with the content to be published to represent the content to be published in the vector dimension. For example, when two contents are related, their corresponding vectors are also close in distance. The vectorization tool can be any suitable machine learning model, which can receive the content to be published as input, and can output a corresponding vector to represent the input content to be published.

[0055] The related content acquisition module 708 acquires multiple related contents from the content library 708 based on the extracted content features, and simultaneously acquires the historical data, clustering information, and release status of the multiple related contents. For example, in some embodiments, the content to be released is news information, wherein the content features of the content to be released are classification categories, multiple keywords, release accounts, article topics, etc., then the related content acquisition module 706 searches for multiple related features that are most relevant to the content to be released 702 from the content library 708 based on these content features. For example, the multiple related contents may be content that matches the content to be released in one or more of the classification categories, multiple keywords, release accounts, and article topics. In some embodiments, the content features of the content to be released 702 may be vectors corresponding to its content, and the related content acquisition module 706 searches for multiple vectors similar to the content to be released 702 in the content library 708. The similarity between vectors may be represented by the distance, angle, and correlation coefficient of the vectors, which are not limited in the present disclosure.

[0056] The statistical feature construction module 710 uses the historical data of multiple related contents and the similarity with the content to be released 702 to construct statistical features. Various similarity statistical information can be constructed as statistical features using the content with the same release status in multiple related contents, such as the maximum similarity, minimum similarity, average similarity, etc. of the related contents whose status is allowed to be released, and the maximum similarity, minimum similarity, average similarity, etc. of the related contents whose status is not allowed to be released, so as to construct statistical features as statistical features of the content to be released. In addition, the historical data of multiple related contents can also be counted, such as the average number of viewers, average number of likes, and other positive content information statistics of multiple related contents, or the number of dislikes, number of reports, and other negative content information statistics of multiple related contents, so as to construct statistical features of historical data for the content to be released 702. The above method is only an example of constructing statistical features, and other information of multiple related contents can also be used to construct statistical features for the content to be released 702, which is not limited by the present disclosure.

[0057] The prediction model calling module 712 processes the constructed statistical features by calling the prediction model 714 , and the prediction score obtaining model 716 obtains the first score for the to-be-published content 702 from the prediction model 714 and sends the first score to the decision module 726 .

[0058] Meanwhile, the cluster feature construction module 718 constructs cluster features using cluster information and similarity with the to-be-published content 702. In some embodiments, cluster features are constructed for the to-be-published content 702 based on the cluster features and corresponding publishing statuses of all related content, for example, cluster features can be constructed using the cluster categories of each related content in a plurality of related content and their corresponding publishing statuses.

[0059] The cluster category confirmation module 720 obtains the cluster category of the to-be-published content 702 from the cluster query table 722 based on the constructed cluster features. The cluster score acquisition module 724 calculates the proportion of different publishing states of all the content corresponding to the cluster category according to the cluster category to generate a second score for the to-be-published content 702, and sends the second score to the decision module 726.

[0060] The decision model 726 compares the first score with the first threshold and the second score with the second threshold to determine the publishing status of the published content 702. For example, the first score is required to be greater than the first threshold and the second score is greater than the second threshold to determine that the publishing status of the published content 702 is allowed to be published; or the first score is required to be greater than the first threshold or the second score is greater than the second threshold to determine that the publishing status of the published content 702 is allowed to be published. In some embodiments, the first score can be a vector, for example, each dimension of the vector is a score for each state, and the corresponding threshold is also a vector in this case for comparison. The threshold can be manually specified or determined according to other methods. For example, the first threshold can be determined according to, for example, Figure 4 The threshold is determined by the selection module shown.

[0061] Through the embodiment shown in process 700, the publishing status of the content to be published can be determined by using the statistical features and clustering features of multiple related contents associated with the content to be published. The use of statistical features and clustering features can improve the accuracy of identification, effectively avoid omissions and misjudgments, and improve the user experience.

[0062] Figure 8 FIG. 8 is a block diagram of an apparatus 800 for content processing according to some embodiments of the present disclosure. Figure 8 As shown, the device 800 includes a published content acquisition module 802, which is configured to acquire the content to be published, wherein the content to be published includes at least one of text, image, or video. The device 800 also includes a related content acquisition module 804, which is configured to acquire multiple related contents associated with the content to be published based on the content features of the content to be published. In addition, the device 800 also includes a publishing status determination module 806, which is configured to determine the publishing status of the content to be published based on the publishing status of the multiple related contents.

[0063] It can be understood that, through the device 800 of the present disclosure, at least one of the advantages that can be achieved by the method or process described above can be achieved. For example, by determining the publishing status of the content to be published based on the publishing status of multiple related contents, it is avoided to determine its publishing status based on the content characteristics of the content to be published, improve the accuracy of determining the publishing status, reduce the omission of low-quality content and the misjudgment of positive and beneficial content, and improve the user experience.

[0064] Fig. 9 1 shows a block diagram of an electronic device 900 according to some embodiments of the present disclosure. The device 900 may be a device or apparatus described in an embodiment of the present disclosure. Fig. 9As shown, the device 900 includes a central processing unit (CPU) and / or a graphics processing unit (GPU) 901, which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) 902 or computer program instructions loaded from a storage unit 908 to a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the device 900 can also be stored. The CPU / GPU 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904. Although not shown in FIG. Fig. 9 As shown in FIG. 9 , device 900 may further include a co-processor.

[0065] A number of components in the device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0066] The various methods or processes described above may be performed by the CPU / GPU 901. For example, in some embodiments, the methods may be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the CPU / GPU 901, one or more steps or actions in the methods or processes described above may be performed.

[0067] In some embodiments, the methods and processes described above may be implemented as a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present disclosure.

[0068] Computer readable storage medium can be a tangible device that can hold and store instructions used by an instruction execution device. Computer readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (non-exhaustive list) of computer readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, for example, a punch card or a convex structure in a groove on which instructions are stored, and any suitable combination thereof. The computer readable storage medium used here is not interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated by a waveguide or other transmission medium (for example, a light pulse by an optical fiber cable), or an electrical signal transmitted by a wire.

[0069] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0070] The computer program instructions for performing the disclosed operation may be assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, programming languages ​​including object-oriented programming languages, and conventional procedural programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, executed as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., utilizing an Internet service provider to connect via the Internet). In certain embodiments, by utilizing the state information of a computer-readable program instruction to customize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit may execute a computer-readable program instruction, thereby realizing various aspects of the present disclosure.

[0071] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processing unit of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0072] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0073] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the equipment, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each frame in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the frame can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous frames can actually be executed substantially in parallel, and they can also be executed in the opposite order sometimes, depending on the functions involved. It should also be noted that each frame in the block diagram and / or flow chart, and the combination of frames in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0074] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

[0075] Some example implementations of the present disclosure are listed below.

[0076] Example 1. A method for content processing, comprising:

[0077] Acquire content to be published, where the content to be published includes at least one of text, image, or video;

[0078] Based on the content features of the content to be published, acquiring a plurality of related contents associated with the content to be published; and

[0079] Based on the publishing status of the plurality of related contents, the publishing status of the to-be-published content is determined.

[0080] Example 2. The method according to Example 1, wherein obtaining the plurality of related contents associated with the to-be-published content comprises:

[0081] The plurality of related contents associated with the content to be published are obtained by determining the correlation between the content feature of each content in the content library and the content feature of the content to be published.

[0082] Example 3. The method according to Example 1, wherein determining the publishing status of the to-be-published content comprises:

[0083] Determining statistical features of the content to be published based on the statistical information of the multiple related contents; and

[0084] Based on the statistical features of the content to be published, the publishing status of the content to be published is determined.

[0085] Example 4. The method according to Example 3, wherein determining the publishing status of the to-be-published content further comprises:

[0086] Determining clustering features of the to-be-published content based on clustering information of the plurality of related content; and

[0087] The publishing status of the content to be published is determined based on the clustering features of the content to be published.

[0088] Example 5. The method according to Example 4, wherein determining the publishing status of the to-be-published content further comprises:

[0089] Based on the statistical features and the clustering features, the publishing status of the to-be-published content is determined.

[0090] Example 6. The method according to Example 4, wherein determining the statistical features of the content to be published comprises:

[0091] Obtaining the relevance between the to-be-published content and the plurality of related contents and statistical information of the plurality of related contents; and

[0092] Based on the relevance and the statistical information, the statistical features of the content to be published are determined.

[0093] Example 7. The method according to Example 6, wherein determining the publishing status of the to-be-published content comprises:

[0094] Obtaining a first score of the content to be published by inputting the statistical features of the content to be published into a machine learning model; and

[0095] The publishing status is determined by comparing the first score with a first threshold, wherein the first threshold is predetermined.

[0096] Example 8. The method according to Example 7, wherein determining the clustering feature of the content to be published comprises:

[0097] Acquire clustering information of each relevant content in the plurality of relevant contents by querying the clustering query table; and

[0098] The clustering feature of the to-be-published content is determined based on the clustering information of each of the plurality of related contents.

[0099] Example 9. The method according to Example 8, wherein determining the publishing status of the to-be-published content based on the clustering feature of the to-be-published content comprises:

[0100] Determining the cluster category of the content to be published based on the cluster feature of the content to be published;

[0101] Determining a second score of the to-be-published content based on the publishing status of the content in the content library belonging to the cluster category; and

[0102] The publishing status is determined by comparing the second score with a second threshold, wherein the second threshold is predetermined.

[0103] Example 10. The method according to Example 9, wherein determining the publishing status of the to-be-published content comprises:

[0104] The publishing status of the to-be-published content is determined by comparing the first score with the first threshold and comparing the second score with the second threshold.

[0105] Example 11. The method according to Example 7, further comprising:

[0106] The machine learning model is trained using training data, wherein the training data includes statistical features and corresponding release statuses.

[0107] Example 12. The method of Example 11, wherein training the machine learning model comprises:

[0108] Acquire a plurality of related contents associated with each content in the content library and a corresponding publishing status for each content;

[0109] Determining the statistical features of each content based on each content and the plurality of related contents; and

[0110] The machine learning model is trained based on the statistical features of each content and the corresponding publishing status.

[0111] Example 13. The method according to Example 8, further comprising:

[0112] The clustering query table is periodically updated using a clustering algorithm.

[0113] Example 14. The method of Example 13, wherein updating the clustering query table comprises:

[0114] Generate updated clustering results based on previous clustering results and the latest content that needs to be clustered;

[0115] merging the updated clustering result with the previous clustering result; and

[0116] The updated clustering result is written into the clustering query table.

[0117] Example 15. An apparatus for content processing, comprising:

[0118] A publishing content acquisition module is configured to acquire content to be published, wherein the content to be published includes at least one of text, image, or video;

[0119] A related content acquisition module, configured to acquire a plurality of related contents associated with the content to be published based on content features of the content to be published; and

[0120] The publishing status determining module is configured to determine the publishing status of the to-be-published content based on the publishing status of the multiple related contents.

[0121] Example 16. The apparatus according to Example 15, wherein obtaining the plurality of related contents associated with the to-be-published content comprises:

[0122] The plurality of related contents associated with the content to be published are obtained by determining the correlation between the content feature of each content in the content library and the content feature of the content to be published.

[0123] Example 17. The apparatus of Example 15, wherein determining the publishing status of the to-be-published content comprises:

[0124] Determining statistical features of the content to be published based on the statistical information of the multiple related contents; and

[0125] Based on the statistical features of the content to be published, the publishing status of the content to be published is determined.

[0126] Example 18. The apparatus according to Example 17, wherein determining the publishing status of the to-be-published content further comprises:

[0127] Determining clustering features of the to-be-published content based on clustering information of the plurality of related content; and

[0128] The publishing status of the content to be published is determined based on the clustering features of the content to be published.

[0129] Example 19. The apparatus according to Example 18, wherein determining the publishing status of the to-be-published content further comprises:

[0130] Based on the statistical features and the clustering features, the publishing status of the to-be-published content is determined.

[0131] Example 20. The apparatus of Example 18, wherein determining the statistical features of the content to be published comprises:

[0132] Obtaining the relevance between the to-be-published content and the plurality of related contents and statistical information of the plurality of related contents; and

[0133] Based on the relevance and the statistical information, the statistical features of the content to be published are determined.

[0134] Example 21. The apparatus of Example 20, wherein determining the publishing status of the to-be-published content comprises:

[0135] Obtaining a first score of the content to be published by inputting the statistical features of the content to be published into a machine learning model; and

[0136] The publishing status is determined by comparing the first score with a first threshold, wherein the first threshold is predetermined.

[0137] Example 22. The apparatus of Example 21, wherein determining the clustering feature of the to-be-published content comprises:

[0138] Acquire clustering information of each relevant content in the plurality of relevant contents by querying the clustering query table; and

[0139] The clustering feature of the to-be-published content is determined based on the clustering information of each of the plurality of related contents.

[0140] Example 23. The apparatus according to Example 22, wherein determining the publishing status of the to-be-published content based on the clustering feature of the to-be-published content comprises:

[0141] Determining the cluster category of the content to be published based on the cluster feature of the content to be published;

[0142] Determining a second score of the to-be-published content based on the publishing status of the content in the content library belonging to the cluster category; and

[0143] The publishing status is determined by comparing the second score with a second threshold, wherein the second threshold is predetermined.

[0144] Example 24. The apparatus of Example 23, wherein determining the publishing status of the to-be-published content comprises:

[0145] The publishing status of the to-be-published content is determined by comparing the first score with the first threshold and comparing the second score with the second threshold.

[0146] Example 25. The apparatus of Example 20, further comprising:

[0147] The machine learning model is trained using training data, wherein the training data includes statistical features and corresponding release statuses.

[0148] Example 26. The apparatus of Example 25, wherein training the machine learning model comprises:

[0149] Acquire a plurality of related contents associated with each content in the content library and a corresponding publishing status for each content;

[0150] Determining the statistical features of each content based on each content and the plurality of related contents; and

[0151] The machine learning model is trained based on the statistical features of each content and the corresponding publishing status.

[0152] Example 27. The apparatus of Example 22, further comprising:

[0153] The clustering query table is periodically updated using a clustering algorithm.

[0154] Example 28. The method of Example 27, wherein updating the clustering query table comprises:

[0155] Generate updated clustering results based on previous clustering results and the latest content that needs to be clustered;

[0156] merging the updated clustering result with the previous clustering result; and

[0157] The updated clustering result is written into the clustering query table.

[0158] Example 29. An electronic device comprising:

[0159] Processor; and

[0160] A memory coupled to the processor, the memory having instructions stored therein, the instructions causing the electronic device to perform actions when executed by the processor, the actions comprising:

[0161] Acquire content to be published, where the content to be published includes at least one of text, image, or video;

[0162] Based on the content features of the content to be published, acquiring a plurality of related contents associated with the content to be published; and

[0163] Based on the publishing status of the plurality of related contents, the publishing status of the to-be-published content is determined.

[0164] Example 30. The electronic device according to Example 29, wherein acquiring the plurality of related contents associated with the to-be-published contents comprises:

[0165] The plurality of related contents associated with the content to be published are obtained by determining the correlation between the content feature of each content in the content library and the content feature of the content to be published.

[0166] Example 31. The electronic device of Example 29, wherein determining the publishing status of the to-be-published content comprises:

[0167] Determining statistical features of the content to be published based on the statistical information of the multiple related contents; and

[0168] Based on the statistical features of the content to be published, the publishing status of the content to be published is determined.

[0169] Example 32. The electronic device according to Example 31, wherein determining the publishing status of the to-be-published content further comprises:

[0170] Determining clustering features of the to-be-published content based on clustering information of the plurality of related content; and

[0171] The publishing status of the content to be published is determined based on the clustering features of the content to be published.

[0172] Example 33. The electronic device according to Example 30, wherein determining the publishing status of the to-be-published content further comprises:

[0173] Based on the statistical features and the clustering features, the publishing status of the to-be-published content is determined.

[0174] Example 34. The electronic device of Example 32, wherein determining the statistical features of the content to be published comprises:

[0175] Obtaining the relevance between the to-be-published content and the plurality of related contents and statistical information of the plurality of related contents; and

[0176] Based on the relevance and the statistical information, the statistical features of the content to be published are determined.

[0177] Example 35. The electronic device of Example 34, wherein determining the publishing status of the to-be-published content comprises:

[0178] Obtaining a first score of the content to be published by inputting the statistical features of the content to be published into a machine learning model; and

[0179] The publishing status is determined by comparing the first score with a first threshold, wherein the first threshold is predetermined.

[0180] Example 36. The electronic device of Example 35, wherein determining the clustering feature of the to-be-published content comprises:

[0181] Acquire clustering information of each relevant content in the plurality of relevant contents by querying the clustering query table; and

[0182] The clustering feature of the to-be-published content is determined based on the clustering information of each of the plurality of related contents.

[0183] Example 37. The electronic device according to Example 36, wherein determining the publishing status of the to-be-published content based on the clustering feature of the to-be-published content comprises:

[0184] Determining the cluster category of the content to be published based on the cluster feature of the content to be published;

[0185] Determining a second score of the to-be-published content based on the publishing status of the content in the content library belonging to the cluster category; and

[0186] The publishing status is determined by comparing the second score with a second threshold, wherein the second threshold is predetermined.

[0187] Example 38. The electronic device of Example 37, wherein determining the publishing status of the to-be-published content comprises:

[0188] The publishing status of the to-be-published content is determined by comparing the first score with the first threshold and comparing the second score with the second threshold.

[0189] Example 39 The electronic device according to Example 35 further includes:

[0190] The machine learning model is trained using the statistical features using training data, wherein the training data includes statistical features and corresponding release statuses.

[0191] Example 38. The electronic device of Example 37, further comprising:

[0192] Based on the predicted probability distribution, the probability distribution of the positive samples, and the probability distribution of the negative samples, a second loss function is generated using a hinge loss function.

[0193] Example 40. The electronic device of Example 39, wherein training the machine learning model comprises:

[0194] Acquire a plurality of related contents associated with each content in the content library and a corresponding publishing status for each content;

[0195] Determining the statistical features of each content based on each content and the plurality of related contents; and

[0196] The machine learning model is trained based on the statistical features of each content and the corresponding publishing status.

[0197] Example 41. The electronic device of Example 37, further comprising:

[0198] The clustering query table is periodically updated using a clustering algorithm.

[0199] Example 42. The electronic device of Example 41, wherein updating the clustering lookup table comprises:

[0200] Generate updated clustering results based on previous clustering results and the latest content that needs to be clustered;

[0201] merging the updated clustering result with the previous clustering result; and

[0202] The updated clustering result is written into the clustering query table.

[0203] Although the disclosure has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Instead, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. A method for content processing, include: Acquire content to be published, where the content to be published includes at least one of text, image, or video; Based on the content characteristics of the content to be published, obtaining a plurality of related contents associated with the content to be published; as well as Based on the publishing status of the plurality of related contents, the publishing status of the to-be-published content is determined.

2. The method according to claim 1, wherein the plurality of related contents associated with the content to be published are obtained include: The plurality of related contents associated with the content to be published are obtained by determining the correlation between the content feature of each content in the content library and the content feature of the content to be published.

3. The method according to claim 1, wherein determining the publishing status of the to-be-published content include: Determining statistical features of the content to be published based on the statistical information of the multiple related contents; as well as Based on the statistical features of the content to be published, the publishing status of the content to be published is determined.

4. The method according to claim 3, wherein determining the publishing status of the to-be-published content further comprises: include: Determining clustering features of the to-be-published content based on clustering information of the plurality of related content; as well as The publishing status of the content to be published is determined based on the clustering features of the content to be published.

5. The method according to claim 4, wherein determining the publishing status of the to-be-published content further comprises: include: Based on the statistical features and the clustering features, the publishing status of the to-be-published content is determined.

6. The method according to claim 4, wherein the statistical characteristics of the content to be published are determined include: Obtaining the relevance between the to-be-published content and the plurality of related contents and statistical information of the plurality of related contents; as well as Based on the relevance and the statistical information, the statistical features of the content to be published are determined.

7. The method according to claim 6, wherein determining the publishing status of the to-be-published content include: Obtaining a first score of the content to be published by inputting the statistical features of the content to be published into a machine learning model; as well as The publishing status is determined by comparing the first score with a first threshold, wherein the first threshold is predetermined.

8. The method according to claim 7, wherein the clustering features of the content to be published are determined include: Acquire clustering information of each relevant content in the plurality of relevant contents by querying the clustering query table; as well as The clustering feature of the to-be-published content is determined based on the clustering information of each of the plurality of related contents.

9. The method according to claim 8, wherein the publishing status of the to-be-published content is determined based on the clustering features of the to-be-published content. include: Determining the cluster category of the content to be published based on the cluster feature of the content to be published; Determining a second score of the to-be-published content based on the publishing status of the content in the content library belonging to the cluster category; as well as The publishing status is determined by comparing the second score with a second threshold, wherein the second threshold is predetermined.

10. The method according to claim 9, wherein determining the publishing status of the to-be-published content include: The publishing status of the to-be-published content is determined by comparing the first score with the first threshold and comparing the second score with the second threshold.

11. The method according to claim 7, further comprising: include: The machine learning model is trained using training data, wherein the training data includes statistical features and corresponding release statuses.

12. The method of claim 11, wherein the machine learning model is trained include: Acquire a plurality of related contents associated with each content in the content library and a corresponding publishing status for each content; Determine the statistical feature of each content based on each content and the plurality of related contents; as well as The machine learning model is trained based on the statistical features of each content and the corresponding publishing status.

13. The method according to claim 8, further comprising: include: The clustering query table is periodically updated using a clustering algorithm.

14. The method of claim 13, wherein updating the clustering query table include: Generate updated clustering results based on previous clustering results and the latest content that needs to be clustered; Merging the updated clustering result with the previous clustering result; as well as The updated clustering result is written into the clustering query table.

15. A device for content processing, include: A publishing content acquisition module is configured to acquire content to be published, wherein the content to be published includes at least one of text, image, or video; A related content acquisition module, configured to acquire a plurality of related contents associated with the content to be published based on the content features of the content to be published; as well as The publishing status determining module is configured to determine the publishing status of the to-be-published content based on the publishing status of the multiple related contents.

16. An electronic device, include: processor; as well as A memory coupled to the processor, the memory having instructions stored therein, wherein when the instructions are executed by the processor, the electronic device executes the method according to any one of claims 1 to 14.

17. A computer-readable storage medium having computer-executable instructions stored thereon, wherein the computer-executable instructions are executed by a processor to implement the method according to any one of claims 1 to 14.

Images