Comment processing method, device, computing device and computer storage medium

Abstract

The embodiment of the present application discloses a comment processing method, device, computing device and computer storage medium. The method includes: storing the instant comments sent by the user in a comment pool, and recording the comment sending time of the instant comments; sorting the instant comments in the comment pool in order of the comment sending time from late to early, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool; displaying the target instant comment in the review interface for review of the target instant comment. This solution can review the latest instant comments, reduce the display delay of the instant comments, improve the matching degree between the displayed instant comments and the played electronic resources, and improve the user experience.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of data processing technology, and specifically to a comment processing method, apparatus, computing device, and computer storage medium. Background Art

[0002] With the continuous development of science and technology and society, the emergence of live broadcast services has greatly enriched people's work and life. Among them, many live broadcast services provide users with instant comment functions such as barrage functions, through which users can send instant comments such as barrage while watching live broadcasts.

[0003] In order to ensure the healthy and safe operation of live broadcast services, it is necessary to review the instant comments sent by users. The existing instant comment review method is to review each instant comment in order from the earliest to the latest time they were sent, and display the instant comments after they are reviewed. However, when this method is used, when there are a large number of instant comments, the displayed instant comments will lag significantly behind the currently viewed resource content, reducing the user service experience.

[0004] Application Contents

[0005] In view of the technical problem in the prior art that the displayed content is significantly delayed compared with the currently viewed resource content, the embodiments of the present application are proposed to provide a comment processing method, apparatus, computing device and computer storage medium that overcome the above problem or at least partially solve the above problem.

[0006] According to a first aspect of an embodiment of the present application, a comment processing method is provided, comprising:

[0007] Store the instant comments sent by users in the comment pool and record the time when the instant comments were sent;

[0008] Sorting the instant comments in the comment pool in order of comment sending time from latest to earliest, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool;

[0009] The target instant comments are displayed in the review interface for reviewing the target instant comments.

[0010] In an optional implementation, after displaying the target instant comment in the review interface, the method further includes:

[0011] Determine whether the display time of the target instant comment on the review interface reaches a first preset time;

[0012] If so, the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface are performed to update the review interface.

[0013] In an optional implementation, before determining whether the display duration of the target instant comment on the review interface reaches a first preset duration, the method further includes:

[0014] Obtain a first number of target instant comments that have been reviewed through the review interface within a first historical time window;

[0015] The first preset duration is determined according to a first window length of the first historical time window and the first quantity.

[0016] In an optional implementation, before determining whether the display duration of the target instant comment on the review interface reaches a first preset duration, the method further includes:

[0017] Obtain a second number of instant comments stored in the comment pool within a second historical time window;

[0018] The first preset duration is determined according to a second window length of the second historical time window and the second quantity.

[0019] In an optional implementation, after displaying the target instant comment in the review interface, the method further includes:

[0020] When the review interface update operation is detected, the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface are performed to update the review interface.

[0021] In an optional implementation, after the review interface is updated, the method further includes: re-storing the target instant comments that have not been reviewed in the review interface before the update in the comment pool.

[0022] In an optional implementation, extracting at least one target instant comment from the comment pool according to the ranking result further includes:

[0023] At least one target instant comment is extracted from the comment pool according to the sorting result and the time difference between the sending time of the instant comment and the current time; wherein the time difference between the sending time of the target instant comment and the current time is less than a second preset time length.

[0024] In an optional implementation, the step of storing the instant comments sent by the user in the comment pool and recording the comment sending time of the instant comments further includes: for any instant comment sent, storing the instant comment in the comment pool, and adding an index record of the instant comment in the index table corresponding to the comment pool; wherein the index record includes a comment identifier and a comment sending time;

[0025] The step of sorting the instant comments in the comment pool in the order of the comment sending time from the latest to the earliest further comprises: after adding the index record of the instant comment in the index table corresponding to the comment pool, sorting the index records in the order of the comment sending time from the latest to the earliest in the index record to update the index table;

[0026] The step of extracting at least one target instant comment from the comment pool according to the sorting result further includes: searching the first N index records in the index table, determining the instant comment corresponding to the comment identifier in the first N index records as the target instant comment, extracting the target instant comment from the comment pool, and deleting the index record of the target instant comment in the index table; wherein N is a positive integer greater than or equal to 1.

[0027] In an optional implementation, storing the instant comments sent by the user in the comment pool further includes: storing the instant comments sent by the user for any electronic resource in the comment pool corresponding to the electronic resource;

[0028] The displaying of the target instant comment in the review interface further includes: displaying the target instant comment and a playback screen of an electronic resource corresponding to the target instant comment in the review interface.

[0029] In an optional implementation, the instant comment is an instant comment on a live stream;

[0030] Then the step of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface further comprises: during the broadcast of the live stream, extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface;

[0031] The method also includes: after the live stream is turned off, extracting instant comments from the comment pool, and displaying the extracted instant comments in a playback review interface for reviewing the instant comments, and displaying the instant comments that have passed the review in the playback interface of the live stream.

[0032] In an optional implementation manner, storing the instant comments sent by the user in the comment pool further includes:

[0033] Receive instant comments from users;

[0034] Identify the exempted comments from the instant comments sent by the users, and store the instant comments other than the exempted comments in the comment pool.

[0035] In an optional implementation, identifying the review-free comments from the instant comments sent by the user further includes:

[0036] For any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, determine that the instant comment is an exempted comment;

[0037] And / or, for any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine that the instant comment is an exempted comment.

[0038] In an optional embodiment, the method further includes:

[0039] The review results of the target instant comments on the electronic resource that have been reviewed through the review interface within the third historical period are obtained, and the white list and / or black list corresponding to the electronic resource are updated according to the review results.

[0040] In an optional implementation, identifying the review-free comments from the instant comments sent by the user further includes:

[0041] For any instant comment, the instant comment is input into a pre-built sentiment analysis model, and the sentiment prediction result output by the sentiment analysis model is obtained, and the review-free comments are identified according to the sentiment prediction result.

[0042] According to a second aspect of an embodiment of the present application, a comment processing method is provided, including:

[0043] Obtain any instant comment sent by the user, and use the negative comment model to determine whether the instant comment is a negative comment;

[0044] If the negative comment model determines that the instant comment is not a negative comment, then using the positive comment model to determine whether the instant comment is a positive comment;

[0045] If the positive comment model determines that the instant comment is a positive comment, the instant comment is sent to the manual review platform.

[0046] In an optional implementation, the negative comment model includes a sensitive word recognition rule, and the use of the negative comment model to determine whether the instant comment is a negative comment further includes: using the sensitive word recognition rule to determine whether there is a sensitive word in the instant comment; if so, determining that the instant comment is a negative comment; if not, determining that the instant comment is not a negative comment;

[0047] And / or, the negative comment model is a negative sentiment analysis model generated based on a machine learning algorithm, then the use of the negative comment model to determine whether the instant comment is a negative comment further includes: inputting the instant comment into the negative comment model, and obtaining the sentiment analysis result of the instant comment output by the negative comment model; if the sentiment analysis result is a negative sentiment, then the instant comment is determined to be a negative comment; if not, then the instant comment is determined not to be a negative comment.

[0048] In an optional implementation, the positive comment model includes a positive word recognition rule, and the using the positive comment model to determine whether the instant comment is a positive comment further includes: using the positive word recognition rule to determine whether there is a positive word in the instant comment; if so, determining that the instant comment is a positive comment; if not, determining that the instant comment is not a positive comment;

[0049] And / or, the positive comment model is a positive sentiment analysis model generated based on a machine learning algorithm, then using the positive comment model to determine whether the instant comment is a positive comment further includes: inputting the instant comment into the positive comment model, and obtaining the sentiment analysis result of the instant comment output by the positive comment model; if the sentiment analysis result is a positive sentiment, then the instant comment is determined to be a positive comment; if not, then the instant comment is determined not to be a positive comment.

[0050] In an optional implementation, after obtaining any instant comment sent by the user, the method further includes: using a preset review-free rule to determine whether the instant comment is a review-free comment;

[0051] Then the using of the negative comment model to determine whether the instant comment is a negative comment further includes: if the instant comment is not an exempted review comment, using the negative comment model to determine whether the instant comment is a negative comment.

[0052] In an optional implementation, the using of a preset review-free rule to determine whether the instant comment is a review-free comment further includes:

[0053] The electronic resource targeted by the instant comment is determined, and the information of the instant comment is matched with the information in the whitelist corresponding to the electronic resource. If the information of the instant comment matches any information in the whitelist, the instant comment is determined to be an exempted comment.

[0054] In an optional implementation, after obtaining any instant comment sent by the user, the method further includes: using a preset filtering rule to determine whether the instant comment is a filtered comment;

[0055] Then the using of the negative comment model to determine whether the instant comment is a negative comment further includes: if the instant comment is not a filtered comment, using the negative comment model to determine whether the instant comment is a negative comment.

[0056] In an optional implementation, the using of a preset filtering rule to determine whether the instant comment is a filtered comment further includes:

[0057] The electronic resource targeted by the instant comment is determined, and the information of the instant comment is matched with the information in the blacklist corresponding to the electronic resource. If the information of the instant comment matches any information in the blacklist, the instant comment is determined to be a filtered comment.

[0058] In an optional embodiment, the method further includes:

[0059] The review results of the instant comments on the electronic resources that have been reviewed by the manual review platform are obtained, and the preset exemption rules and / or preset filtering rules are updated according to the review results.

[0060] In an optional implementation, sending the instant comment to the manual review platform further includes:

[0061] Storing the instant comment in a comment pool and recording the time when the instant comment was sent;

[0062] Sorting the instant comments in the comment pool in order of comment sending time from latest to earliest, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool;

[0063] The target instant comments are sent to the manual review platform so as to be displayed in the review interface of the manual review platform.

[0064] According to a third aspect of an embodiment of the present application, a comment processing device is provided, including:

[0065] A comment processing module is used to store the instant comments sent by the user in a comment pool and record the comment sending time of the instant comments; and sort the instant comments in the comment pool in order of comment sending time from the latest to the earliest;

[0066] A comment extraction module, used to extract at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool;

[0067] The display module is used to display the target instant comments in the review interface for reviewing the target instant comments.

[0068] In an optional embodiment, the device also includes: a judgment module, which is used to judge whether the display time of the target instant comment in the review interface reaches a first preset time after the target instant comment is displayed in the review interface; if so, executing the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface to update the review interface.

[0069] In an optional implementation, the device further includes: a duration determination module, configured to obtain a first number of target instant comments reviewed through the review interface within a first historical time window;

[0070] The first preset duration is determined according to a first window length of the first historical time window and the first quantity.

[0071] In an optional implementation, the device further includes: a duration determination module, configured to obtain a second number of instant comments stored in the comment pool within a second historical time window;

[0072] The first preset duration is determined according to a second window length of the second historical time window and the second quantity.

[0073] In an optional embodiment, the device also includes: a detection module, which is used to detect the review interface update operation, execute the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface to update the review interface.

[0074] In an optional implementation, the comment processing module is used to re-store the target instant comments that have not been reviewed in the review interface before the update in the comment pool.

[0075] In an optional embodiment, the comment extraction module is used to: extract at least one target instant comment from the comment pool based on the sorting result and the time difference between the sending time of the instant comment and the current time; wherein the time difference between the sending time of the target instant comment and the current time is less than a second preset duration.

[0076] In an optional implementation, the comment processing module is used to: for any instant comment sent, store the instant comment in a comment pool, and add an index record of the instant comment in an index table corresponding to the comment pool; wherein the index record includes a comment identifier and a comment sending time; after adding the index record of the instant comment to the index table corresponding to the comment pool, sort the index records according to the order of comment sending time in the index record from latest to earliest, so as to update the index table;

[0077] A comment extraction module is used to search the first N index records in the index table, determine the instant comment corresponding to the comment identifier in the first N index records as the target instant comment, extract the target instant comment from the comment pool, and delete the index record of the target instant comment in the index table; wherein N is a positive integer greater than or equal to 1.

[0078] In an optional implementation, the comment processing module is used to: store the instant comments sent by the user for any electronic resource in a comment pool corresponding to the electronic resource;

[0079] The display module is used to display the target instant comment and the playback screen of the electronic resource corresponding to the target instant comment in the review interface.

[0080] In an optional implementation, the instant comment is an instant comment on a live stream;

[0081] The comment extraction module is used to: extract at least one target instant comment from the comment pool according to the sorting result during the live streaming process;

[0082] The device also includes: a playback review module, which is used to extract instant comments from the comment pool after the live stream is turned off, and display the extracted instant comments in the playback review interface for reviewing the instant comments, and display the instant comments that have passed the review in the playback interface of the live stream.

[0083] In an optional implementation, the comment processing module is used to: receive instant comments sent by users;

[0084] Identify the exempted comments from the instant comments sent by the users, and store the instant comments other than the exempted comments in the comment pool.

[0085] In an optional implementation, the comment processing module is used to: for any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, determine that the instant comment is an exempted comment;

[0086] And / or, for any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine that the instant comment is an exempted comment.

[0087] In an optional embodiment, the device also includes: a list update module, which is used to obtain the review results of the target instant comments on the electronic resources that have been reviewed through the review interface within the third historical period, and update the whitelist and / or blacklist corresponding to the electronic resource according to the review results.

[0088] In an optional embodiment, the comment processing module is used to: for any instant comment, input the instant comment into a pre-built sentiment analysis model, obtain the sentiment prediction result output by the sentiment analysis model, and identify the exempted comments based on the sentiment prediction result.

[0089] According to a fourth aspect of an embodiment of the present application, a comment processing system is provided, including:

[0090] The acquisition module is used to obtain any instant comment sent by the user;

[0091] Negative comment model, used to determine whether the instant comment is a negative comment;

[0092] A positive comment model, used to determine whether the instant comment is a positive comment if the negative comment model determines that the instant comment is not a negative comment;

[0093] A sending module, configured to send the instant comment to a manual review platform if the positive comment model determines that the instant comment is a positive comment;

[0094] Manual review platform, used for manual review of instant comments.

[0095] In an optional implementation, the negative comment model includes a sensitive word recognition rule, and the negative comment model is used to: use the sensitive word recognition rule to determine whether there is a sensitive word in the instant comment; if so, determine that the instant comment is a negative comment; if not, determine that the instant comment is not a negative comment;

[0096] And / or, the negative comment model is a negative sentiment analysis model generated based on a machine learning algorithm, and the negative comment model is used to: input the instant comment into the negative comment model, and obtain the sentiment analysis result of the instant comment output by the negative comment model; if the sentiment analysis result is a negative sentiment, then the instant comment is determined to be a negative comment; if not, then the instant comment is determined not to be a negative comment.

[0097] In an optional implementation, the positive comment model includes a positive word recognition rule, and the positive comment model is used to: use the positive word recognition rule to determine whether there is a positive word in the instant comment; if so, determine that the instant comment is a positive comment; if not, determine that the instant comment is not a positive comment;

[0098] And / or, the positive comment model is a positive sentiment analysis model generated based on a machine learning algorithm, and the positive comment model is used to: input the instant comment into the positive comment model, and obtain the sentiment analysis result of the instant comment output by the positive comment model; if the sentiment analysis result is a positive sentiment, then the instant comment is determined to be a positive comment; if not, then the instant comment is determined not to be a positive comment.

[0099] In an optional implementation, the system further includes: a session rule module, which is used to use a preset review-free rule to determine whether any instant comment sent by the user is a review-free comment after obtaining the instant comment;

[0100] The negative comment model is used to: if the instant comment is not an exempted comment, the negative comment model is used to determine whether the instant comment is a negative comment.

[0101] In an optional implementation, the session rule module is used to: determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, determine that the instant comment is an exempted comment.

[0102] In an optional implementation, the system further includes: a session rule module, configured to use a preset filtering rule to determine whether the instant comment is a filtered comment;

[0103] The negative comment model is used to: if the instant comment is not a filtered comment, the negative comment model is used to determine whether the instant comment is a negative comment.

[0104] In an optional implementation, the session rule module is used to: determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine that the instant comment is a filtered comment.

[0105] In an optional embodiment, the system further includes: a rule updating module for obtaining the review results of the instant comments on the electronic resources that have been reviewed by the manual review platform, and updating the preset exemption rules and / or preset filtering rules according to the review results.

[0106] In an optional implementation, the sending module is used to: store the instant comment in a comment pool, and record the comment sending time of the instant comment;

[0107] Sorting the instant comments in the comment pool in order of comment sending time from latest to earliest, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool;

[0108] The target instant comments are sent to the manual review platform so as to be displayed in the review interface of the manual review platform.

[0109] According to a fifth aspect of an embodiment of the present application, there is provided a computing device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other through the communication bus;

[0110] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned comment processing method.

[0111] According to a sixth aspect of an embodiment of the present application, a computer storage medium is provided, wherein at least one executable instruction is stored in the storage medium, and the executable instruction enables a processor to perform operations corresponding to the above-mentioned comment processing method.

[0112] The embodiment of the present application stores the instant comments in a comment pool, and sorts the instant comments in the comment pool in order of the comment sending time from late to early, and extracts the target instant comments with late comment sending time for display, so that the latest instant comments can be reviewed, the display delay of the instant comments can be reduced, the matching degree of the displayed instant comments and the played electronic resources can be improved, and the user experience can be improved.

[0113] The embodiment of the present application reacquires a new target instant comment display when the display time of the target instant comment on the review interface reaches a first preset time, thereby facilitating the review of the latest instant comment and further reducing the display delay of the instant comment.

[0114] The embodiment of the present application determines the first preset duration according to the review speed, so that the acquisition speed of the target instant comments can match the review speed, thereby improving the overall review effect.

[0115] The embodiment of the present application determines the first preset duration according to the storage speed of the instant comments, so that the acquisition speed of the target instant comments can match the storage speed of the instant comments, thereby improving the overall review effect.

[0116] The embodiment of the present application detects the update operation of the audit interface and updates the audit interface, thereby meeting the audit needs of the auditor.

[0117] This application re-stores the target instant comments that have not been reviewed in the review interface before the update in the comment pool, so that the comment pool stores the instant comments that have not been reviewed, which is convenient for centralized processing of the instant comments that have not been reviewed.

[0118] The embodiment of the present application extracts target instant comments from the comment pool based on the sorting results and the time difference between the sending time of the instant comments and the current time, so that the time difference between the sending time of the target instant comments and the current time is less than a second preset time length, thereby avoiding displaying instant comments with too early sending time in the review interface.

[0119] The embodiment of the present application has a corresponding index table in the comment pool. Whenever a new instant comment is added and / or a target instant comment is extracted, the index record in the index table is updated, and the index records in the index table are arranged in order from late to early according to the sending time, so that the target instant comment can be quickly extracted.

[0120] The embodiment of the present application stores the instant comments sent by users for any electronic resource in a comment pool corresponding to the electronic resource, thereby avoiding direct interference between instant comments from different electronic resources; and displays the target instant comments and the playback screen of the electronic resource corresponding to the target instant comments in the review interface, thereby facilitating reviewers to review the target instant comments and improving the review accuracy.

[0121] During the live streaming process, the present application can extract at least one target instant comment from the comment pool according to the sorting result and display the target instant comment in the review interface, thereby reducing the display delay of the instant comments during the live streaming process; and after the live streaming is turned off, the instant comments in the comment pool are reviewed, and the instant comments that have passed the review are displayed in the playback interface of the live streaming, so that users can watch the complete instant comments that have passed the review in the live playback.

[0122] After receiving the instant comments sent by the user, the embodiment of the present application identifies the exempted comments from the instant comments sent by the user, and stores the instant comments other than the exempted comments in a comment pool, thereby reducing the number of instant comments displayed in the review interface, saving review resources and improving the overall review efficiency.

[0123] This application can identify exempted comments based on the whitelist and / or blacklist of the corresponding electronic resources, thereby improving the recognition efficiency and accuracy of exempted comments.

[0124] This application obtains the review results of the target instant comments on the electronic resources that have been reviewed through the review interface within the third historical period, and updates the whitelist and / or blacklist corresponding to the electronic resource according to the review results, which is conducive to improving the recognition accuracy of subsequent exempted review comments.

[0125] The embodiment of the present application can also identify review-exempt comments through a sentiment analysis model, thereby further improving the recognition accuracy of review-exempt comments.

[0126] The embodiment of the present application adopts a negative comment model, a positive comment model and a manual review platform in turn to review instant comments, thereby reducing the number of instant comments entering the manual review platform, simplifying the manual review process and improving review efficiency.

[0127] The above description is only an overview of the technical solution of the embodiment of the present application. In order to more clearly understand the technical means of the embodiment of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present application more obvious and easy to understand, the specific implementation method of the embodiment of the present application is listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0128] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only used for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the embodiments of the present application. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0129] Figure 1 A schematic diagram showing a flow chart of a comment processing method provided in an embodiment of the present application;

[0130] Figure 2 A flowchart of another comment processing method provided in an embodiment of the present application is shown;

[0131] Figure 3 A schematic diagram of an audit interface provided in an embodiment of the present application is shown;

[0132] Figure 4 A flowchart of another comment processing method provided in an embodiment of the present application is shown;

[0133] Figure 5 A flowchart of another comment processing method provided in an embodiment of the present application is shown;

[0134] Figure 6 A schematic diagram of a process flow of a bullet screen processing method provided in an embodiment of the present application is shown;

[0135] Figure 7 A schematic diagram of the structure of a comment processing device provided in an embodiment of the present application is shown;

[0136] Figure 8 A structural diagram of a comment processing system provided by an embodiment of the present application is shown;

[0137] Fig. 9 A schematic diagram of the architecture of comment processing provided by an embodiment of the present application is shown;

[0138] Fig.10 A schematic diagram of the structure of a computing device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0139] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application embodiments and to enable the scope of the present application embodiments to be fully communicated to those skilled in the art.

[0140] Figure 1 A flow chart of a comment processing method provided by an embodiment of the present application is shown. The comment processing method provided by this embodiment is specifically used for processing instant comments. The execution subject of this embodiment can be an instant comment review platform.

[0141] Specifically, Figure 1 As shown, the method comprises the following steps:

[0142] Step S110, storing the instant comments sent by the user in a comment pool, and recording the time when the instant comments were sent.

[0143] Instant comments are commentary data sent by users in real time to electronic resources such as videos and audios while they are watching them. The electronic resources may be live streams, and the instant comments may be bullet comments sent by users during the live stream.

[0144] This embodiment stores the instant comments sent by the user in the comment pool, and records the comment sending time of any instant comment in the comment pool, wherein the comment sending time of the instant comment is the time when the user sends the instant comment.

[0145] In an optional implementation, in order to facilitate the management of instant comments and improve the processing efficiency of instant comments, in this implementation, the instant comments sent by users for any electronic resource are stored in the comment pool corresponding to the electronic resource, so that each comment pool only stores the instant comments for the corresponding electronic resource, avoiding interference between instant comments of different electronic resources. For example, there are currently live broadcast rooms A and B in the live broadcast platform, then the bullet comments for the live stream in live broadcast room A are stored in the comment pool M corresponding to live broadcast room A, and the bullet comments for the live stream in live broadcast room B are stored in the comment pool N corresponding to live broadcast room B.

[0146] Step S120, sorting the instant comments in the comment pool in order of comment sending time from latest to earliest, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool.

[0147] Different from the prior art that reviews each instant comment in order from the earliest to the latest comment sending time, this embodiment sorts the instant comments in the comment pool in order from the latest to the earliest comment sending time. The later the comment sending time, the higher the sorting order. Then, at least one instant comment with a high order is extracted from the instant comments in the current comment pool. The extracted instant comment is the target instant comment, and the instant comments other than the target instant comment in the current comment pool are non-target instant comments. The comment sending time of each extracted target instant comment is later than the comment sending time of the non-target instant comments in the current comment pool. Therefore, this embodiment obtains instant comments with a later sending time from the comment pool each time, thereby ensuring the subsequent review of the latest instant comments. Among them, after the target instant comment is extracted, the target instant comment in the comment pool is deleted synchronously.

[0148] In an optional embodiment, in order to improve the extraction efficiency of the target instant comment, in this embodiment, the comment pool has a corresponding index table, then step S110 stores the instant comment sent by any user in the comment pool, and adds the index record of the instant comment in the index table corresponding to the comment pool, wherein the index record includes the comment ID and the comment sending time. After adding the index record of the instant comment in the index table corresponding to the comment pool, the index records are sorted according to the order of the comment sending time in the index record from late to early to update the index table. Then, when determining the target instant comment, this step searches the first N index records in the index table, and determines the instant comment corresponding to the comment ID in the first N index records as the target instant comment, wherein N is a positive integer greater than or equal to 1. Extract the target instant comment from the comment pool and delete the index record of the target instant comment in the index table. In this embodiment, the index table is updated in time after the instant comment is stored in the comment pool, and the index records in the index table are arranged in order from the latest to the earliest time the comment is sent, so that when it is necessary to extract the target instant comment from the comment pool, the target instant comment to be extracted can be quickly determined by searching the index record, thereby improving the extraction efficiency of the target instant comment. Moreover, whenever a new instant comment is stored in the comment pool or whenever a target instant comment is extracted, the index table is updated in time to ensure the extraction accuracy of subsequent target instant comments.

[0149] Step S130, displaying the target instant comment in the review interface for reviewing the target instant comment.

[0150] The extracted target instant comments are displayed in the review interface of the instant comment review platform. Reviewers can review the latest instant comments. The target instant comments that pass the review will be displayed in a timely manner to the user terminal that plays the electronic resources.

[0151] It can be seen that the embodiment of the present application stores the instant comments in a comment pool, and sorts the instant comments in the comment pool in order from late to early comment sending time, and extracts the target instant comments with late comment sending time for display, so that the latest instant comments can be reviewed, the display delay of the instant comments can be reduced, the matching degree of the displayed instant comments and the played electronic resources can be improved, and the user experience can be improved.

[0152] Figure 2 A flowchart of another comment processing method provided by an embodiment of the present application is shown. The comment processing method provided by this embodiment is specifically used for processing instant comments. The execution subject of this embodiment can be an instant comment review platform.

[0153] Specifically, Figure 2 As shown, the method comprises the following steps:

[0154] Step S210, storing the instant comments sent by the user in a comment pool, recording the comment sending time of the instant comments, and sorting the instant comments in the comment pool in order of comment sending time from latest to earliest.

[0155] Step S220, extracting a preset number of target instant comments from the comment pool according to the sorting result.

[0156] The preset number can be determined based on the number of instant comments displayed on each review interface at a time and / or the number of review interfaces. For example, in this embodiment, P review terminals can concurrently review the target instant comments, each review terminal includes an review interface, and each review interface can display Q instant comments at a time, so the preset number can be P*Q.

[0157] In an optional implementation, when determining the target instant comment, in addition to the ranking result of the instant comment, the target instant comment can be further determined by combining the time difference between the sending time of the instant comment in the comment pool and the current time, that is, according to the ranking result and the time difference between the sending time of the instant comment and the current time, at least one target instant comment is extracted from the comment pool; wherein, the time difference between the sending time of the target instant comment and the current time is less than the second preset time length, thereby avoiding the instant comment sent at a time far from the current time from being displayed in the user interface. In the specific implementation process, after the candidate target instant comment is screened out according to the ranking result, the candidate target instant comment whose time difference between the comment sending time and the current time is less than the second preset time length is used as the final target instant comment. Specifically, it is determined whether the time difference between the candidate target instant comment with the latest comment sending time and the current time is less than the second preset time length. If so, all the current candidate target instant comments are used as the target instant comment; if not, the candidate target instant comment is eliminated, and it is determined whether the time difference between the candidate target instant comment with the latest comment sending time and the current time is less than the second preset time length. By analogy, the final target instant comment is determined.

[0158] Step S230, displaying the target instant comments in the review interface.

[0159] If there is a single review end, the obtained target instant comments will be displayed in the review interface of the review end; if there are multiple review ends, the obtained target instant comments will be split into multiple groups, each group contains at least one target instant comment, and the number of target instant comments contained in each group can be the same or different. The number of split groups matches the number of review ends, and then each group of target instant comments is sent to the corresponding review end, and the received target instant comments are displayed in the review interface of the review end.

[0160] In an optional implementation, if the target real-time comment is a comment on a certain electronic resource, the target real-time comment and the playback screen of the electronic resource corresponding to the target real-time comment are displayed in the review interface, so as to facilitate the reviewer to accurately review the real-time comment. For example, if the target real-time comment is a bullet screen for the live stream in live broadcast room A, the review interface will display the current playback screen of the live stream in live broadcast room A in addition to the bullet screen extracted from the comment pool.

[0161] In another optional implementation, the review interface also includes a review operation item for the displayed target instant comment, and the reviewer can realize the review of the corresponding target instant comment by triggering the review operation item. Wherein, the review operation item includes: push and / or mark. After triggering the push review operation item, it is determined that the corresponding target instant comment is reviewed and passed, and the target instant comment is pushed to the user terminal that plays the electronic resource in time for displaying the instant comment in the user terminal interface. After triggering the mark review operation item, it is determined that the corresponding target instant comment is not reviewed, and the target instant comment is placed in the mark collection for further review of the target instant comment when idle, or directly abandoning the target instant comment, etc. For example, if the reviewer triggers the push control next to the bullet screen 1, the bullet screen 1 is displayed in the playback interface of the user terminal of the live stream; if the reviewer triggers the mark control next to the bullet screen 1, the target instant comment can be placed in the mark collection, and the target instant comment is reviewed again when the review is idle later, and so on.

[0162] Step S240, determining whether the display duration of the target instant comment reaches a first preset duration; if so, executing step S220.

[0163] In order to further reduce the display delay of the instant comment, after displaying the target instant comment in the review interface, it is detected whether the display duration of the currently displayed target instant comment reaches the first preset duration. If so, steps S220-S230 are executed in time to obtain a new target instant comment and display it in the review interface, so as to update the review interface, thereby facilitating the reviewer to review the latest instant comment in time.

[0164] The first preset time length may be a fixed value, such as 5 seconds, etc. Alternatively, the first preset time length may be dynamically determined so that the review page update speed matches the actual review scenario. When the first preset time length is a dynamic value, it may be determined in one or more of the following ways:

[0165] Determination method 1: Determine the first preset duration according to the review speed, specifically to make the acquisition speed of the target instant comments compatible with the review speed, for example, the acquisition speed of the target instant comments can be made consistent with the review speed or the acquisition speed of the target instant comments can be slightly greater than the review speed. Specifically, obtain the first number of target instant comments that have been reviewed through the review interface within the first historical time window, and determine the first preset duration according to the first window length and the first number of the first historical time window. Among them, the first historical time window is a historical time period before the execution of this step, and the window length of the first historical time window is the first window length, that is, the time length of the historical time period is the first window length. This determination method counts the number of target instant comments reviewed by the reviewer within the first historical time window, and the number is the first number. Then the review speed within the first historical time window is the ratio of the first number to the first window length. The acquisition speed of the target instant comments is the ratio of the preset number to the first preset duration, and the preset number is the number of target instant comments extracted from the comment pool each time. Therefore, the first preset duration can be determined according to the first quantity, the first preset duration, and the preset quantity, and the first preset duration ≈ (preset quantity*first window length) / first quantity.

[0166] Determination method 2: Determine the first preset duration according to the speed at which the instant comments are stored in the comment pool, specifically to adapt the acquisition speed of the target instant comments to the speed at which the instant comments are stored in the comment pool. For example, the acquisition speed of the target instant comments can be equal to or slightly less than the speed at which the instant comments are stored in the comment pool. Specifically, obtain the second number of instant comments stored in the comment pool in the second historical time window, and determine the first preset duration according to the second window length and the second number of the second historical time window. In this determination method, the second historical time window is a historical time period before the execution of this step, and the second historical time window can be consistent with or inconsistent with the first historical time window. The length of the second historical time window is the second window length, and the number of instant comments stored in the comment pool in the second historical time window is the second number. Then, the speed at which the instant comments are stored in the comment pool in the second historical time window is the ratio of the second number to the second window length. Since the acquisition speed of the target instant comments is the ratio of the preset number to the first historical time window, the first preset duration can be determined according to the second number, the second window length, and the preset number, and the first preset duration ≈ (preset number*second window length) / second number.

[0167] Determination method three: Determine the first preset duration according to the review speed and the speed at which instant comments are stored in the comment pool, that is, the acquisition speed of the target instant comments is adapted to the review speed and the speed at which instant comments are stored in the comment pool. Specifically, the first preset duration determined by the above determination method one can be used as the first candidate first preset duration, and the first preset duration determined by the above determination method two can be used as the second candidate first preset duration. Finally, the final first preset duration is determined based on the first candidate first preset duration and the second candidate first preset duration. For example, the average of the first candidate first preset duration and the second candidate first preset duration can be used as the final first preset duration.

[0168] In an optional implementation, if the display duration of the target instant comment does not reach the first preset duration at a certain time, and the review interface update operation is detected, the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface are performed to update the review interface. That is, in this implementation, the reviewer can actively trigger the update of the review interface at a certain time that does not exceed the first preset duration, thereby meeting the reviewer's review needs.

[0169] After updating the review interface, if there are unreviewed target instant comments in the review interface before the update, the unreviewed target instant comments can be discarded; or the unreviewed target instant comments in the review interface before the update can be re-stored in the comment pool, so that the comment pool contains unreviewed instant comments.

[0170] by Figure 3 Take the audit interface shown as an example. Figure 3 The review interface shown shows the playback screen of the live stream with room number "111", and shows bullet comments 1, 2, 3, 4, and 5 for the live stream with room number "111". The current reviewer has completed the review of bullet comments 1 and 2. Bullet comment 1 has passed the review and has been pushed to the user terminal watching the live stream with room number "111", so that the user can view bullet comment 1 in the user terminal; Bullet comment 2 has not passed the review, and the bullet comment 2 has been marked and sent to the corresponding tag library. Bullet comments 3, 4, and 5 have not been reviewed. The reviewer can click the "Push" control on the side of bullet comment 3 to push bullet comment 3 to the user terminal watching the live stream with room number "111"; or click the "Mark" control on the side of bullet comment 3 to mark the bullet comment 3 and send it to the corresponding tag library. Furthermore, the review interface also includes a progress bar "Automatically change a batch of bullet comments after X seconds" to remind the reviewer that the review interface will be automatically updated after X seconds. In addition, the reviewer can click the "Change Batch" control within X seconds to trigger the update of the review interface.

[0171] In addition, as an optional implementation of the embodiment of the present application, if the instant comment is an instant comment for the live stream, the embodiment of the present application can adopt the above implementation to reduce the display delay of the instant comment during the live stream broadcast process to ensure the matching degree between the displayed instant comment and the live screen. That is, during the live stream broadcast process, at least one target instant comment is extracted from the comment pool according to the sorting result, and the target instant comment is displayed in the review interface.

[0172] After the live stream is turned off, the comment pool still contains the real-time comments corresponding to the live stream that have not been reviewed. In this embodiment, the real-time comments can be extracted from the comment pool after the live stream is turned off, and the real-time comments can be displayed in the playback review interface for review, and the real-time comments that have passed the review can be displayed in the playback interface of the live stream. Therefore, when replaying the live stream, the complete user comments that have passed the review can be viewed.

[0173] It can be seen that after the embodiment of the present application displays the target instant comments in the review interface, it detects whether the display time of the target instant comments reaches the first preset time. If so, new target instant comments are re-acquired for display, thereby ensuring that the reviewer can review the latest instant comments in a timely manner, and further improving the matching degree between the instant comments displayed on the user end and the electronic resource screen.

[0174] Figure 4 A flow chart of another comment processing method provided in an embodiment of the present application is shown. The comment processing method provided in this embodiment is specifically used for processing instant comments. The execution subject of this embodiment can be an instant comment review platform.

[0175] Specifically, Figure 4 As shown, the method comprises the following steps:

[0176] Step S410: receiving instant comments sent by users.

[0177] Step S420, identifying the review-exempt comments from the instant comments sent by the user, and storing the instant comments other than the review-exempt comments in a comment pool.

[0178] In order to further reduce the number of instant comments that are manually reviewed, this embodiment first filters the instant comments after receiving the instant comments sent by the user. Specifically, for any received instant comment, it is determined whether the instant comment is an exempted comment or a pending review comment. If the instant comment is determined to be an exempted comment, the exempted comment does not need to be stored in the comment pool. If the instant comment is determined to be a pending review comment, the instant comment is stored in the comment pool for subsequent manual review. Among them, the instant comments other than the exempted comments are pending reviews.

[0179] In an optional implementation, one or more of the following methods may be used to identify review-exempt comments and / or review-pending comments.

[0180] Identification method 1: For any instant comment, determine the electronic resource targeted by the instant comment, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, then determine the instant comment as an exempted comment. In this identification method, each electronic resource has a whitelist corresponding to it, for example, each live broadcast room has a corresponding whitelist. The whitelist contains at least one piece of information, which can be: at least one comment field and / or at least one user identifier. When matching the information of the instant comment with the information in the whitelist corresponding to the electronic resource, specifically, match the comment content of the instant comment with each comment field in the whitelist, if the comment content of the instant comment matches a certain comment field, then determine the instant comment as an exempted comment; and / or, match the user identifier of the user corresponding to the instant comment with each user identifier in the whitelist, if the user identifier of the user corresponding to the instant comment matches a certain user identifier in the whitelist, then determine the instant comment as an exempted comment. The exempted comments determined by this identification method do not need to be manually reviewed, and such exempted comments can be sent to the user terminal that plays the electronic resource so that the user terminal can display such exempted comments.

[0181] Identification method 2: For any instant comment, determine the electronic resource targeted by the instant comment, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine the instant comment as an exempted comment. In this identification method, each electronic resource has a blacklist corresponding to it, for example, each live broadcast room has a corresponding blacklist. The blacklist contains at least one piece of information, which can be: at least one comment field and / or at least one user identifier. When matching the information of the instant comment with the information in the blacklist corresponding to the electronic resource, specifically, match the comment content of the instant comment with each comment field in the blacklist, if the comment content of the instant comment matches a certain comment field, determine the instant comment as an exempted comment; and / or, match the user identifier of the user corresponding to the instant comment with each user identifier in the blacklist, if the user identifier of the user corresponding to the instant comment matches a certain user identifier in the blacklist, determine the instant comment as an exempted comment. This identification method determines that the exempted comment does not need to be manually reviewed, and such exempted comments can be directly discarded.

[0182] Further optionally, in order to improve recognition accuracy, the above-mentioned whitelist and blacklist can be updated in the following way: obtain the review results of the target instant comments on the electronic resources that have been reviewed through the review interface within the third historical period, and update the whitelist and / or blacklist corresponding to the electronic resource according to the review results.

[0183] Specifically, first-category target instant comments on electronic resources that have been reviewed and approved within the third historical period are obtained, and the target instant comments on electronic resources that have been reviewed and approved within the third historical period are first-category target instant comments. The comment content of the first-category target instant comments is obtained, and the frequency of the comment content is counted. If the occurrence frequency of a certain comment content is greater than a first frequency threshold, the comment content is added to the whitelist of the electronic resource; and / or, the user identifier corresponding to the first-category target instant comments is obtained, and the frequency of the user identifier is counted. If the occurrence frequency of a certain user identifier is greater than a second frequency threshold, the user identifier is added to the whitelist of the electronic resource.

[0184] Correspondingly, the second-category target instant comments on the electronic resources that have not passed the review within the third historical period are obtained. The target instant comments on the electronic resources that have not passed the review within the third historical period are the second-category target instant comments. The comment content of the second-category target instant comments is obtained, and the frequency of the comment content is counted. If the occurrence frequency of a certain comment content is greater than the third frequency threshold, the comment content is added to the blacklist of the electronic resource; and / or, the user identifier corresponding to the second-category target instant comments is obtained, and the frequency of the user identifier is counted. If the occurrence frequency of a certain user identifier is greater than the fourth frequency threshold, the user identifier is added to the blacklist of the electronic resource.

[0185] For example, if the reviewer has pushed 11 bullet comments X targeting live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the reviewed and approved comment content X in the last half hour exceeds the first frequency threshold, and X is added to the whitelist corresponding to live broadcast room A, so that when the X bullet comment is received later, the bullet comment is determined as a bullet comment exempted from review and pushed to the live broadcast client for display. Alternatively, if the reviewer has pushed 12 bullet comments sent by user U1 targeting live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the user ID U1 corresponding to the instant comments reviewed and approved in the last half hour exceeds the second frequency threshold, and the user ID of user U1 is added to the whitelist corresponding to live broadcast room A, so that when the bullet comment sent by user U1 targeting live broadcast room A is received later, the received bullet comment is determined as a bullet comment exempted from review and pushed to the live broadcast client for display.

[0186] Accordingly, if the reviewer has marked 11 bullet comments Y for live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the comment content Y that has not passed the review in the last half hour exceeds the third frequency threshold, and the bullet comment Y is added to the blacklist corresponding to live broadcast room A, so that when the bullet comment Y is received later, the bullet comment is determined as an exempted bullet comment and discarded. Alternatively, if the reviewer has marked 12 bullet comments sent by user U2 for live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the user ID U2 corresponding to the instant comments that have not passed the review in the last half hour exceeds the fourth frequency threshold, and the user ID of user U2 is added to the blacklist corresponding to live broadcast room A, so that when the bullet comment sent by user U2 for live broadcast room A is received later, the received bullet comment is determined as an exempted bullet comment and discarded.

[0187] Identification method three: For any instant comment, input the instant comment into a pre-built sentiment analysis model, and obtain the sentiment prediction result output by the sentiment analysis model, and determine whether the instant comment is an exempted comment or a pending review comment according to the sentiment prediction result. Among them, the sentiment analysis model can be constructed based on a machine learning algorithm, such as a deep learning algorithm can be used to construct and train the machine learning algorithm. This application does not limit the structure and training method of the specific sentiment analysis model. The sentiment analysis model can output the sentiment prediction result of the instant comment. If the situation prediction result is a negative sentiment, the instant comment is determined to be an exempted comment and the instant comment is discarded. Specifically, the sentiment analysis model can be a negative sentiment analysis model. If the output result of the negative sentiment analysis model is yes, the instant comment is determined to be an exempted comment; the sentiment analysis model can be a positive sentiment analysis model. If the output result of the positive sentiment analysis model is no, the instant comment is determined to be an exempted comment.

[0188] Identification method 4: Build a sensitive word library. For any instant comment, use the NLP (Natural Language Processing) algorithm to determine whether the instant comment contains sensitive words. If so, the instant comment is determined to be exempted from review and discarded.

[0189] Step S430, sorting the instant comments in the comment pool in order of the comment sending time from the latest to the earliest, and extracting at least one target instant comment from the comment pool according to the sorting result.

[0190] Step S440, displaying the target instant comment in the review interface for reviewing the target instant comment.

[0191] It can be seen that after receiving the instant comments sent by the user, the embodiment of the present application identifies the exempted comments and / or the awaited comments from the instant comments sent by the user, and stores the awaited comments in the comment pool, so that only some of the awaited comments need to be displayed in the review interface, thereby reducing the amount of comments that need to be reviewed, saving review resources, and further improving the matching degree between the instant comments displayed on the user end and the electronic resource screen.

[0192] Figure 5 A flow chart of another comment processing method provided in an embodiment of the present application is shown. The comment processing method provided in this embodiment is specifically used for processing instant comments.

[0193] Specifically, Figure 5 As shown, the method comprises the following steps:

[0194] Step S510, obtaining any instant comment sent by the user.

[0195] In an optional implementation, after obtaining any instant comment sent by a user, the preset review exemption rule can be used to determine whether the instant comment is a review exemption comment; if the instant comment is not a review exemption comment, the negative comment model can be used to determine whether the instant comment is a negative comment; if the instant comment is a review exemption comment, the instant comment is used as the instant comment displayed to the audience user, thereby improving the review efficiency of the instant comment.

[0196] Specifically, in the process of judging whether the instant comment is an exempted comment: determine the electronic resource targeted by the instant comment, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, then determine that the instant comment is an exempted comment. In this identification method, each electronic resource has a whitelist corresponding to it, for example, each live broadcast room has a corresponding whitelist. The whitelist contains at least one piece of information, which can be: at least one comment field and / or at least one user identifier. When matching the information of the instant comment with the information in the whitelist corresponding to the electronic resource, specifically, match the comment content of the instant comment with each comment field in the whitelist, if the comment content of the instant comment matches a certain comment field, then determine that the instant comment is an exempted comment; and / or, match the user identifier of the user corresponding to the instant comment with each user identifier in the whitelist, if the user identifier of the user corresponding to the instant comment matches a certain user identifier in the whitelist, then determine that the instant comment is an exempted comment.

[0197] In another optional implementation, after obtaining any instant comment sent by the user, a preset filtering rule is used to determine whether the instant comment is a filtered comment; if the instant comment is not a filtered comment, a negative comment model is used to determine whether the instant comment is a negative comment; if the instant comment is a filtered comment, the instant comment is discarded, thereby improving the review efficiency of the instant comment.

[0198] Specifically, in the process of judging whether the instant comment is a process comment: determine the electronic resource targeted by the instant comment, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, then determine that the instant comment is a filtered comment. In this identification method, each electronic resource has a blacklist corresponding to it, for example, each live broadcast room has a corresponding blacklist. The blacklist contains at least one piece of information, which can be: at least one comment field and / or at least one user identifier. When matching the information of the instant comment with the information in the blacklist corresponding to the electronic resource, specifically, match the comment content of the instant comment with each comment field in the blacklist, if the comment content of the instant comment matches a certain comment field, then determine that the instant comment is a filtered comment; and / or, match the user identifier of the user corresponding to the instant comment with each user identifier in the blacklist, if the user identifier of the user corresponding to the instant comment matches a certain user identifier in the blacklist, then determine that the instant comment is a filtered comment.

[0199] Further optionally, in order to improve the recognition accuracy of exempted review comments and / or filtered comments, the present application further obtains the review results of the instant comments for the electronic resource that have been reviewed by the manual review platform, and updates the preset exempted review rules and / or preset filtering rules according to the review results. Specifically, the above-mentioned whitelist and / or blacklist is updated according to the review results of the instant comments. Specifically, the first-category target instant comments for electronic resources that have been reviewed and approved in the historical period are obtained, and the target instant comments for electronic resources that have been reviewed and approved in the historical period are the first-category target instant comments. The comment content of the first-category target instant comments is obtained, and the frequency of the comment content is counted. If the frequency of occurrence of a certain comment content is greater than the first frequency threshold, the comment content is added to the whitelist of the electronic resource; and / or, the user identifier corresponding to the first-category target instant comment is obtained, and the frequency of the user identifier is counted. If the frequency of occurrence of a certain user identifier is greater than the second frequency threshold, the user identifier is added to the whitelist of the electronic resource.

[0200] Correspondingly, the second-category target instant comments on the electronic resources that have not passed the review in the historical period are obtained. The target instant comments on the electronic resources that have not passed the review in the historical period are the second-category target instant comments. The comment content of the second-category target instant comments is obtained, and the frequency of the comment content is counted. If the occurrence frequency of a certain comment content is greater than the third frequency threshold, the comment content is added to the blacklist of the electronic resource; and / or, the user identifier corresponding to the second-category target instant comments is obtained, and the frequency of the user identifier is counted. If the occurrence frequency of a certain user identifier is greater than the fourth frequency threshold, the user identifier is added to the blacklist of the electronic resource.

[0201] For example, if the reviewer has pushed 11 bullet comments X targeting live broadcast room A through the manual review platform in the last half hour, it means that the frequency of occurrence of the reviewed and approved comment content X in the last half hour exceeds the first frequency threshold, and X is added to the whitelist corresponding to live broadcast room A, so that when the X bullet comment is received later, the bullet comment is determined as a bullet comment exempted from review and pushed to the live broadcast client for display. Alternatively, if the reviewer has pushed 12 bullet comments sent by user U1 targeting live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the user ID U1 corresponding to the instant comments reviewed and approved in the last half hour exceeds the second frequency threshold, and the user ID of user U1 is added to the whitelist corresponding to live broadcast room A, so that when the bullet comment sent by user U1 targeting live broadcast room A is received later, the received bullet comment is determined as a bullet comment exempted from review and pushed to the live broadcast client for display.

[0202] Accordingly, if the reviewer has marked 11 bullet comments Y for live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the comment content Y that has not passed the review in the last half hour exceeds the third frequency threshold, and the bullet comment Y is added to the blacklist corresponding to live broadcast room A, so that when the bullet comment Y is received later, it is determined as a filtered bullet comment and discarded. Alternatively, if the reviewer has marked 12 bullet comments sent by user U2 for live broadcast room A through the review interface in the last half hour, it means that the frequency of occurrence of the user ID U2 corresponding to the instant comments that have not passed the review in the last half hour exceeds the fourth frequency threshold, and the user ID of user U2 is added to the blacklist corresponding to live broadcast room A, so that when the bullet comment sent by user U2 for live broadcast room A is received later, the received bullet comment is determined as a filtered bullet comment and discarded.

[0203] Step S520, using a negative comment model to determine whether the instant comment is a negative comment; if not, executing step S530.

[0204] In order to reduce the amount of review by the manual review platform and improve the review efficiency of comments, the embodiment of the present application pre-builds a negative review model, which can identify negative comments. Specifically, the negative review model is used to identify any instant comment obtained. If the negative review model determines that the instant comment is a negative comment, the instant comment is discarded, so that the negative comment is no longer reviewed and displayed in the future; if the negative review model determines that the instant comment is not a negative comment, step S530 is executed to perform the next step of review.

[0205] In an optional implementation, the negative review model may be determined in one or more of the following ways:

[0206] Method 1: If the negative comment model contains sensitive word recognition rules, the negative comment model is used to determine whether the instant comment is a negative comment. Specifically, the sensitive word recognition rules are used to determine whether there are sensitive words in the instant comment; if so, the instant comment is determined to be a negative comment; if not, the instant comment is determined not to be a negative comment. Specifically, a sensitive word library containing at least one sensitive word is constructed, and a corresponding natural language recognition algorithm is used to determine whether there are sensitive words in the instant comment. If so, the instant comment is determined to be a negative comment; if not, the instant comment is determined not to be a negative comment.

[0207] Method 2: The negative comment model is a negative sentiment analysis model generated based on a machine learning algorithm, and the negative comment model is used to determine whether the instant comment is a negative comment. Specifically, the instant comment is input into the negative comment model, and the sentiment analysis result of the instant comment output by the negative comment model is obtained; if the sentiment analysis result is negative sentiment, the instant comment is determined to be a negative comment; if not, the instant comment is determined not to be a negative comment. Among them, this application does not limit the structure of the specific sentiment analysis model and the training method.

[0208] Step S530, using the positive comment model to determine whether the instant comment is a positive comment; if so, executing step S540.

[0209] If the negative comment model determines that the instant comment is not a negative comment, the positive comment model is further used to determine whether the instant comment is a positive comment. If yes, the process proceeds to step S540 for the next step of review. If no, the instant comment is discarded.

[0210] In an optional implementation, the positive review model may be determined in one or more of the following ways:

[0211] Method 1: If the positive comment model contains positive word recognition rules, the positive comment model is used to determine whether the instant comment is a positive comment. Specifically, the positive word recognition rules are used to determine whether there are positive words in the instant comment; if so, the instant comment is determined to be a positive comment; if not, the instant comment is determined not to be a positive comment.

[0212] Method 2: The positive comment model is a positive sentiment analysis model generated based on a machine learning algorithm. The positive comment model is used to determine whether the instant comment is a positive comment. Specifically, the instant comment is input into the positive comment model, and the sentiment analysis result of the instant comment output by the positive comment model is obtained; if the sentiment analysis result is a positive sentiment, the instant comment is determined to be a positive comment; if not, the instant comment is determined not to be a positive comment.

[0213] Step S540: Send the instant comment to a manual review platform.

[0214] If the positive comment model determines that the instant comment is a positive comment, the instant comment will be further sent to the manual review platform, and the reviewer will further review the instant comment.

[0215] In an optional implementation, when sending instant comments to the manual review platform, they can be sent in a last-in-first-out order. Specifically, the instant comments are stored in a comment pool, and the comment sending time of the instant comments is recorded; the instant comments in the comment pool are sorted in order from the latest to the earliest comment sending time, and at least one target instant comment is extracted from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool; the target instant comment is sent to the manual review platform, so that the target instant comment is displayed in the review interface of the manual review platform. Thus, the latest instant comment is sent to the manual review platform each time.

[0216] The following Figure 6 Take the following example to explain the implementation process of the present application in detail:

[0217] In a live broadcast, after the user sends a barrage, this application first performs repeated rule filtering for this broadcast to determine the exemption rules and filtering rules applicable to this broadcast, and removes the repeated rules. Further screening of the broadcast rules is performed. If the barrage hits the exemption rule, the barrage is sent successfully, so that the live broadcast user can see the barrage in time; if the barrage hits the filtering rule, the sending is unsuccessful, so that the live broadcast user cannot see the barrage; if the barrage neither hits the exemption rule nor the filtering rule, it further enters the negative model filtering, and the negative comment model determines whether the barrage is a negative comment. If the barrage hits the negative model, that is, the barrage is a negative comment, the barrage is sent unsuccessfully; if the barrage does not hit the negative model, that is, the barrage is not a negative comment, the barrage enters the positive model recognition, and the positive comment model determines whether the barrage is not a positive comment. If the barrage does not hit the positive model, it means that the barrage is not a positive comment, and the barrage is not sent successfully; if the barrage hits the positive model, it means that the barrage is a positive comment, and the barrage enters the manual review platform. The reviewer conducts manual review on the manual review platform. If the reviewer pushes the barrage, the barrage is sent successfully; if the reviewer marks the barrage or ignores it, the barrage is not sent successfully. In addition, the rules are further updated in real time based on the review results of the barrage on the manual review platform.

[0218] It can be seen that the embodiment of the present application adopts a negative comment model, a positive comment model and a manual review platform in turn to review instant comments, thereby reducing the number of instant comments entering the manual review platform, simplifying the manual review process and improving review efficiency.

[0219] Figure 7 The schematic diagram of the structure of a comment processing device provided in an embodiment of the present application is shown. The device is located in the instant comment review platform. Figure 7 As shown, the device 700 includes: a comment processing module 710 , a comment extraction module 720 and a display module 730 .

[0220] The comment processing module 710 is used to store the instant comments sent by the user in the comment pool and record the comment sending time of the instant comments; sort the instant comments in the comment pool in the order of comment sending time from the latest to the earliest;

[0221] A comment extraction module 720 is used to extract at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool;

[0222] The display module 730 is used to display the target instant comment in the review interface for reviewing the target instant comment.

[0223] In an optional embodiment, the device also includes: a judgment module, which is used to judge whether the display time of the target instant comment in the review interface reaches a first preset time after the target instant comment is displayed in the review interface; if so, executing the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface to update the review interface.

[0224] In an optional implementation, the device further includes: a duration determination module, configured to obtain a first number of target instant comments reviewed through the review interface within a first historical time window;

[0225] The first preset duration is determined according to a first window length of the first historical time window and the first quantity.

[0226] In an optional implementation, the device further includes: a duration determination module, configured to obtain a second number of instant comments stored in the comment pool within a second historical time window;

[0227] The first preset duration is determined according to a second window length of the second historical time window and the second quantity.

[0228] In an optional embodiment, the device also includes: a detection module, which is used to detect the review interface update operation, execute the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface to update the review interface.

[0229] In an optional implementation, the comment processing module is used to re-store the target instant comments that have not been reviewed in the review interface before the update in the comment pool.

[0230] In an optional embodiment, the comment extraction module is used to: extract at least one target instant comment from the comment pool based on the sorting result and the time difference between the sending time of the instant comment and the current time; wherein the time difference between the sending time of the target instant comment and the current time is less than a second preset duration.

[0231] In an optional implementation, the comment processing module is used to: for any instant comment sent, store the instant comment in a comment pool, and add an index record of the instant comment in an index table corresponding to the comment pool; wherein the index record includes a comment identifier and a comment sending time; after adding the index record of the instant comment to the index table corresponding to the comment pool, sort the index records according to the order of comment sending time in the index record from latest to earliest, so as to update the index table;

[0232] A comment extraction module is used to search the first N index records in the index table, determine the instant comment corresponding to the comment identifier in the first N index records as the target instant comment, extract the target instant comment from the comment pool, and delete the index record of the target instant comment in the index table; wherein N is a positive integer greater than or equal to 1.

[0233] In an optional implementation, the comment processing module is used to: store the instant comments sent by the user for any electronic resource in a comment pool corresponding to the electronic resource;

[0234] The display module is used to display the target instant comment and the playback screen of the electronic resource corresponding to the target instant comment in the review interface.

[0235] In an optional implementation, the instant comment is an instant comment on a live stream;

[0236] The comment extraction module is used to: extract at least one target instant comment from the comment pool according to the sorting result during the live streaming process;

[0237] The device also includes: a playback review module, which is used to extract instant comments from the comment pool after the live stream is turned off, and display the extracted instant comments in the playback review interface for reviewing the instant comments, and display the instant comments that have passed the review in the playback interface of the live stream.

[0238] In an optional implementation, the comment processing module is used to: receive instant comments sent by users;

[0239] Identify the exempted comments from the instant comments sent by the users, and store the instant comments other than the exempted comments in the comment pool.

[0240] In an optional implementation, the comment processing module is used to: for any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, determine that the instant comment is an exempted comment;

[0241] And / or, for any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine that the instant comment is an exempted comment.

[0242] In an optional embodiment, the device also includes: a list update module, which is used to obtain the review results of the target instant comments on the electronic resources that have been reviewed through the review interface within the third historical period, and update the whitelist and / or blacklist corresponding to the electronic resource according to the review results.

[0243] In an optional embodiment, the comment processing module is used to: for any instant comment, input the instant comment into a pre-built sentiment analysis model, obtain the sentiment prediction result output by the sentiment analysis model, and identify the exempted comments based on the sentiment prediction result.

[0244] It can be seen that the embodiment of the present application stores the instant comments in a comment pool, and sorts the instant comments in the comment pool in order from late to early comment sending time, and extracts the target instant comments with late comment sending time for display, so that the latest instant comments can be reviewed, the display delay of the instant comments can be reduced, the matching degree of the displayed instant comments and the played electronic resources can be improved, and the user experience can be improved.

[0245] Figure 8 FIG. 1 is a schematic diagram showing the structure of a comment processing system provided by an embodiment of the present application. Figure 8 As shown, the system 800 includes: an acquisition module 810 , a negative comment model 820 , a positive comment model 830 , a sending module 840 , and a manual review platform 850 .

[0246] The acquisition module 810 is used to acquire any instant comment sent by the user;

[0247] Negative comment model 820, used to determine whether the instant comment is a negative comment;

[0248] The positive comment model 830 is used to determine whether the instant comment is a positive comment if the negative comment model determines that the instant comment is not a negative comment;

[0249] A sending module 840 is used to send the instant comment to a manual review platform if the positive comment model determines that the instant comment is a positive comment;

[0250] The manual review platform 850 is used to manually review the instant comments.

[0251] In an optional implementation, the negative comment model includes a sensitive word recognition rule, and the negative comment model is used to: use the sensitive word recognition rule to determine whether there is a sensitive word in the instant comment; if so, determine that the instant comment is a negative comment; if not, determine that the instant comment is not a negative comment;

[0252] And / or, the negative comment model is a negative sentiment analysis model generated based on a machine learning algorithm, and the negative comment model is used to: input the instant comment into the negative comment model, and obtain the sentiment analysis result of the instant comment output by the negative comment model; if the sentiment analysis result is a negative sentiment, then the instant comment is determined to be a negative comment; if not, then the instant comment is determined not to be a negative comment.

[0253] In an optional implementation, the positive comment model includes a positive word recognition rule, and the positive comment model is used to: use the positive word recognition rule to determine whether there is a positive word in the instant comment; if so, determine that the instant comment is a positive comment; if not, determine that the instant comment is not a positive comment;

[0254] And / or, the positive comment model is a positive sentiment analysis model generated based on a machine learning algorithm, and the positive comment model is used to: input the instant comment into the positive comment model, and obtain the sentiment analysis result of the instant comment output by the positive comment model; if the sentiment analysis result is a positive sentiment, then the instant comment is determined to be a positive comment; if not, then the instant comment is determined not to be a positive comment.

[0255] In an optional implementation, the system further includes: a session rule module, which is used to use a preset review-free rule to determine whether any instant comment sent by the user is a review-free comment after obtaining the instant comment;

[0256] The negative comment model is used to: if the instant comment is not an exempted comment, the negative comment model is used to determine whether the instant comment is a negative comment.

[0257] In an optional implementation, the session rule module is used to: determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, determine that the instant comment is an exempted comment.

[0258] In an optional implementation, the system further includes: a session rule module, configured to use a preset filtering rule to determine whether the instant comment is a filtered comment;

[0259] The negative comment model is used to: if the instant comment is not a filtered comment, the negative comment model is used to determine whether the instant comment is a negative comment.

[0260] In an optional implementation, the session rule module is used to: determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine that the instant comment is a filtered comment.

[0261] In an optional embodiment, the system further includes: a rule updating module for obtaining the review results of the instant comments on the electronic resources that have been reviewed by the manual review platform, and updating the preset exemption rules and / or preset filtering rules according to the review results.

[0262] In an optional implementation, the sending module is used to: store the instant comment in a comment pool, and record the comment sending time of the instant comment;

[0263] Sorting the instant comments in the comment pool in order of comment sending time from latest to earliest, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool;

[0264] The target instant comments are sent to the manual review platform so as to be displayed in the review interface of the manual review platform.

[0265] It can be seen that the embodiment of the present application adopts a negative comment model, a positive comment model and a manual review platform in turn to review instant comments, thereby reducing the number of instant comments entering the manual review platform, simplifying the manual review process and improving review efficiency.

[0266] Fig. 9 FIG. 1 shows a schematic diagram of a comment processing architecture provided by an embodiment of the present application. Fig. 9 As shown, the architecture includes a terminal layer, a business layer, and a data layer.

[0267] The terminal layer includes at least one user terminal, which can be a PC, Web, IOS and / or Android, etc. The user terminal in the terminal layer can send bullet comments.

[0268] The business layer receives the barrage sent by the end layer. The business layer includes the barrage service, the internal background, and the barrage push job. Among them, the barrage service includes negative rules, positive rules, and session rules, wherein the negative rule is used to determine whether the barrage is a negative comment, the positive rule user determines whether the barrage is a positive comment, and the session rule includes the exemption rule and / or the filtering rule. Through the rules in the barrage service, some barrages can be filtered out, thereby reducing the amount of barrages entering the selected workbench. The internal background further includes the selected workbench (also known as the manual review platform) and the selected configuration, wherein the selected workbench can present the review page to the reviewer for the reviewer to review the barrage, and the selected configuration includes the relevant configuration data of the selected workbench, such as the amount of barrage displayed per page, the page refresh frequency, etc. The barrage push job is used to push the barrage to the end layer, so that the barrage that has passed the review is displayed on the user end. In addition, the operation of the selected workbench is fed back to the barrage service, so that the rules in the barrage service can be updated.

[0269] The data layer includes the pending review queue, the passed queue, the push statistics and the marking statistics modules. Among them, after the barrage is filtered by the barrage service in the business layer, the barrage that needs to be reviewed by the selected workbench will be added to the pending review queue of the pending review pool; the reviewer can obtain the barrage in the pending review queue from the data layer through the selected workbench for review, and determine the barrage that has passed the review based on the review operation, and then add the barrage that has passed the review to the passed queue, and further obtain the barrage push Job that generates the passed barrage through the barrage, and then push the passed barrage to the user end of the end layer. In addition, the push statistics in the data layer include relevant information of the pushed barrage (such as the number of pushed barrages, the pushed barrage ID, etc.), as well as relevant information of the marked barrage (such as the number of marked barrages, the marked barrage ID, etc.).

[0270] Fig.10 The schematic diagram of the structure of a computing device provided by the embodiment of the present application is shown. The specific embodiment of the present application does not limit the specific implementation of the computing device.

[0271] like Fig.10 As shown, the computing device may include: a processor (processor) 1002 , a communication interface (Communications Interface) 1004 , a memory (memory) 1006 , and a communication bus 1008 .

[0272] The processor 1002, the communication interface 1004, and the memory 1006 communicate with each other via the communication bus 1008. The communication interface 1004 is used to communicate with other devices such as a client or other server network elements. The processor 1002 is used to execute the program 1010, which can specifically execute the relevant steps in the above-mentioned embodiment of the comment processing method.

[0273] Specifically, the program 1010 may include program codes, which include computer operation instructions.

[0274] The processor 1002 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0275] The memory 1006 is used to store the program 1010. The memory 1006 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory. The program 1010 can be specifically used to enable the processor 1002 to execute any of the above method embodiments.

[0276] An embodiment of the present application provides a non-volatile computer storage medium, wherein the computer storage medium stores at least one executable instruction, and the computer executable instruction can execute the comment processing method in any of the above method embodiments.

[0277] The algorithm or display provided here are not inherently related to any specific computer, virtual system or other equipment. Various general systems can also be used together with the teaching based on this. According to the above description, it is obvious to construct the structure required for this type of system. In addition, the present application embodiment is not directed to any specific programming language yet. It should be understood that various programming languages ​​can be utilized to realize the content of the present application embodiment described here, and the above description of specific languages ​​is to disclose the best mode of implementation of the present application embodiment.

[0278] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application embodiments can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0279] Similarly, it should be understood that in order to streamline the embodiments of the present application and help understand one or more of the various application aspects, in the above description of the exemplary embodiments of the present application, the various features of the embodiments of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following intention: the embodiments of the present application claimed for protection require more features than the features clearly stated in each claim. More specifically, as reflected in the claims below, the application aspects are less than all the features of the single embodiment disclosed above. Therefore, the claims following the specific embodiment are hereby explicitly incorporated into the specific embodiment, wherein each claim itself is a separate embodiment of the embodiment of the present application.

[0280] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and in addition they may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0281] In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the embodiments of the present application and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.

[0282] The various component embodiments of the embodiments of the present application can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It should be understood by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all functions of some or all components according to the embodiments of the present application. The embodiments of the present application can also be implemented as a device or apparatus program (e.g., computer program and computer program product) for executing part or all of the methods described herein. Such a program implementing the embodiments of the present application can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0283] It should be noted that the above embodiments illustrate rather than limit the embodiments of the present application, and that those skilled in the art may design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be constructed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of multiple such elements. The embodiments of the present application may be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A comment processing method, characterized in that: include: Store the instant comments sent by users in the comment pool and record the time when the instant comments were sent; Sorting the instant comments in the comment pool in order of comment sending time from latest to earliest, and extracting at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool; The target instant comments are displayed in the review interface for reviewing the target instant comments.

2. The method according to claim 1, characterized in that After displaying the target instant comment in the review interface, the method further includes: Determine whether the display time of the target instant comment on the review interface reaches a first preset time; If so, the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface are performed to update the review interface.

3. The method according to claim 2, characterized in that Before determining whether the display time of the target instant comment on the review interface reaches a first preset time, the method further includes: Obtain a first number of target instant comments that have been reviewed through the review interface within a first historical time window; The first preset duration is determined according to a first window length of the first historical time window and the first quantity.

4. The method according to claim 2, characterized in that: Before determining whether the display time of the target instant comment on the review interface reaches a first preset time, the method further includes: Obtain a second number of instant comments stored in the comment pool within a second historical time window; The first preset duration is determined according to a second window length of the second historical time window and the second quantity.

5. The method according to claim 1, characterized in that After displaying the target instant comment in the review interface, the method further includes: When the review interface update operation is detected, the steps of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface are performed to update the review interface.

6. The method according to any one of claims 2 to 5, characterized in that: After the review interface is updated, the method further includes: re-storing the target instant comments that have not been reviewed in the review interface before the update in the comment pool.

7. The method according to any one of claims 1 to 5, characterized in that The extracting at least one target instant comment from the comment pool according to the sorting result further comprises: At least one target instant comment is extracted from the comment pool according to the sorting result and the time difference between the sending time of the instant comment and the current time; wherein the time difference between the sending time of the target instant comment and the current time is less than a second preset time length.

8. The method according to any one of claims 1 to 5, characterized in that The storing of the instant comments sent by the user in the comment pool and recording the comment sending time of the instant comments further includes: for any instant comment sent, storing the instant comment in the comment pool, and adding an index record of the instant comment in the index table corresponding to the comment pool; wherein the index record includes the comment identifier and the comment sending time; The step of sorting the instant comments in the comment pool in the order of the comment sending time from the latest to the earliest further comprises: after adding the index record of the instant comment in the index table corresponding to the comment pool, sorting the index records in the order of the comment sending time from the latest to the earliest in the index record to update the index table; The step of extracting at least one target instant comment from the comment pool according to the sorting result further includes: searching the first N index records in the index table, determining the instant comment corresponding to the comment identifier in the first N index records as the target instant comment, extracting the target instant comment from the comment pool, and deleting the index record of the target instant comment in the index table; wherein N is a positive integer greater than or equal to 1.

9. The method according to any one of claims 1 to 5, characterized in that The storing of the instant comments sent by the user in the comment pool further includes: storing the instant comments sent by the user for any electronic resource in the comment pool corresponding to the electronic resource; The displaying of the target instant comment in the review interface further includes: displaying the target instant comment and a playback screen of an electronic resource corresponding to the target instant comment in the review interface.

10. The method according to any one of claims 1 to 5, characterized in that The instant comments are instant comments on the live stream; Then the step of extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface further comprises: during the live stream broadcasting process, extracting at least one target instant comment from the comment pool according to the sorting result and displaying the target instant comment in the review interface; The method also includes: after the live stream is turned off, extracting instant comments from the comment pool, and displaying the extracted instant comments in a playback review interface for reviewing the instant comments, and displaying the instant comments that have passed the review in the playback interface of the live stream.

11. The method according to any one of claims 1 to 5, characterized in that The storing of the instant comments sent by the user in the comment pool further includes: Receive instant comments from users; Identify the exempted comments from the instant comments sent by the users, and store the instant comments other than the exempted comments in the comment pool.

12. The method according to claim 11, characterized in that The step of identifying the review-free comments from the instant comments sent by the user further includes: For any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the whitelist corresponding to the electronic resource, and if the information of the instant comment matches any information in the whitelist, determine that the instant comment is an exempted comment; And / or, for any instant comment, determine the electronic resource to which the instant comment is directed, match the information of the instant comment with the information in the blacklist corresponding to the electronic resource, and if the information of the instant comment matches any information in the blacklist, determine that the instant comment is an exempted comment.

13. The method according to claim 12, characterized in that The method further comprises: The review results of the target instant comments on the electronic resource that have been reviewed through the review interface within the third historical period are obtained, and the white list and / or black list corresponding to the electronic resource are updated according to the review results.

14. The method according to claim 11, characterized in that The step of identifying the review-free comments from the instant comments sent by the user further includes: For any instant comment, the instant comment is input into a pre-built sentiment analysis model, and the sentiment prediction result output by the sentiment analysis model is obtained, and the review-free comments are identified according to the sentiment prediction result.

15. A comment processing device, characterized in that: include: The comment processing module is used to store the instant comments sent by the user in the comment pool and record the comment sending time of the instant comments; Sorting the instant comments in the comment pool in order of the latest comment sending time; A comment extraction module, used to extract at least one target instant comment from the comment pool according to the sorting result; wherein the comment sending time of the target instant comment is later than the comment sending time of each non-target instant comment in the comment pool; The display module is used to display the target instant comments in the review interface for reviewing the target instant comments.

16. A computing device, characterized in that include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform an operation corresponding to the comment processing method according to any one of claims 1-14.

17. A computer storage medium, characterized in that: The storage medium stores at least one executable instruction, and the executable instruction enables the processor to execute an operation corresponding to the comment processing method according to any one of claims 1 to 14.

Images