Security mechanism execution method applied to group chat

Abstract

The invention discloses a security mechanism execution method applied to group chat, which comprises the following steps: when group chat members publish information contents, preliminarily checking the information contents published by the members, and if a preliminary checking result is to be manually checked, sending the information contents published by the members to a checker in a working state at a current time point; the accuracy of abnormal content detection is improved by partially combining manual auditing, meanwhile, the current time point serves as a reference point, abnormal data of the group chat members in the set time period are extracted and analyzed, and therefore the abnormal indexes of the group chat members in the set time period are obtained, and the abnormal content detection accuracy is improved. And if the abnormal index of the group chat member in the set time period is greater than the preset threshold index, an early warning is triggered and sent to a group chat manager, so that deep analysis of the group chat member corresponding to the abnormal content is realized, and the comprehensiveness of a safety mechanism is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method for executing a security mechanism applied to group chat. Background Art

[0002] With the rapid development of Internet technology, group chat has become an important way for people to communicate in daily life, especially in the fields of social media, work collaboration and online learning. Group chat provides a convenient platform for multiple people to exchange information, share opinions and collaborate to complete tasks in real time.

[0003] However, with the popularization of group chat, its security issues have become increasingly prominent, and the group chat security mechanism in the prior art still has the following deficiencies:

[0004] The existing group chat security mechanism often directly blocks the execution when abnormal content is detected during the review of group chat content. Although this can play a security role to a certain extent, there are cases of misjudgment, and it does not consider the historical abnormal data of the group chat members corresponding to the abnormal content within the set time period for comprehensive evaluation, and conduct in-depth analysis of the group chat members corresponding to the abnormal content, so as to timely discover and prevent potential abnormal problems;

[0005] At the same time, most of the existing technologies provide targeted solutions when problems occur in group chats. It is impossible to set a fixed evaluation time interval to conduct regular evaluations on group chats, so as to determine the risk level of group chats, and there is one-sidedness in the use of security mechanisms.

[0006] To this end, a security mechanism execution method applied to group chat is introduced. Summary of the invention

[0007] In view of this, the present invention provides a security mechanism execution method applied to group chat to solve the problems raised by the above background technology.

[0008] The purpose of the present invention can be achieved by the following technical solution: A method for executing a security mechanism applied to a group chat, comprising:

[0009] Information audit: When a group chat member posts information, a preliminary review of the information posted by the member is conducted;

[0010] Evaluation and execution: Based on the preliminary review results of the information released, the corresponding steps are executed;

[0011] Audit optimization: Taking the time point when a group chat member posts information as the reference point, when the group chat member posts information again, obtain the abnormal index pt of the group chat member and perform corresponding steps to optimize the audit steps;

[0012] Periodic evaluation: preset the evaluation time interval of the group chat. After the preset estimated time interval is reached, analyze the abnormal status of the group chat during the evaluation time period to obtain the hidden danger index qe of the group chat during the evaluation time period. Based on the hidden danger index qe of the group chat during the evaluation time period, determine the hidden danger level of the group chat during the evaluation time period.

[0013] In some embodiments, a preliminary review is conducted on the information content posted by the member, specifically:

[0014] S1: extract the information content posted by the group chat members to perform extension check, and execute corresponding steps based on the result of the extension check. If the check result indicates that the posted information content is text, execute step S1-101; if the check result indicates that the posted information content is a picture, execute step S1-102; if the check result indicates that the posted information content is video or audio, execute step S1-103;

[0015] S1-101: If the inspection result indicates that the published information content is text, after extracting the text content, use the pre-built sensitive word library to scan the text, and match the text content with the words in the sensitive word library; after the match is successful, mark the position of the sensitive word in the text content, and after the marking is completed, count the number of marked sensitive words as the sensitive number, and at the same time divide the marked sensitive words into sensitivity levels. According to the pre-built sensitive word library, build multiple sensitivity level intervals in the sensitive word library, set each sensitivity level interval to contain multiple groups of sensitive words, and store each group of sensitive words in the corresponding sensitivity level interval according to the corresponding sensitivity level; after dividing the marked sensitive words into sensitivity levels, obtain the sensitivity level of each group of marked sensitive words;

[0016] Different sensitivity levels are set to correspond to a sensitivity base value, and the sensitivity base value is converted based on the sensitivity level of each group of marked sensitive words. After the conversion is completed, the sensitivity base values ​​of each group are accumulated to obtain the total sensitivity value;

[0017] Preset weight coefficients corresponding to the sensitive quantity and the total sensitive value, multiply the sensitive quantity and the total sensitive value by the corresponding weight coefficients respectively, and then sum them up to obtain the sensitivity index of the information content released by the member;

[0018] A reference threshold range of the preset sensitivity index is used to compare the sensitivity index of the information content published by the member with the preset reference threshold range. If the sensitivity index of the information content published by the member is within the preset reference threshold range, the information content published by the member is marked as pending manual review; if the sensitivity index of the information content published by the member is higher than the preset reference threshold range, the information content published by the member is marked as abnormal publication.

[0019] In some embodiments, the preliminary review of the information content posted by the member further includes:

[0020] S1-102: If the check result indicates that the published information content is a picture, the picture is first preprocessed; after the preprocessing is completed, the text in the picture is detected and extracted, and then step S1-101 is executed;

[0021] S1-103: If the check result indicates that the published information content is video or audio, the video content or audio content is converted into text using a transcription tool; after conversion into text, the conversion errors in the converted text content are searched, and after the search is completed, step S1-101 is executed.

[0022] In some embodiments, based on the preliminary review results of the published information content, corresponding steps are performed, specifically:

[0023] M1: If the preliminary review result is pending manual review, the information content published by the member is sent to the auditor who is currently working. After the auditor reviews, if the information content published by the member is found to be abnormal, the preliminary review result of the information content published by the member is modified to abnormal release, and step M2 is executed;

[0024] M2: If the preliminary review result is abnormal release, the group chat member corresponding to the abnormal distribution is marked, and the abnormal data of the group chat member within the set time period is extracted and analyzed with the current time point as the reference point, so as to obtain the abnormal index pt of the group chat member within the set time period;

[0025] M3: preset the threshold index of the abnormal index pt. If the abnormal index pt of the group chat member within the set time period is greater than the preset threshold index, an early warning is triggered and sent to the administrator of the group chat.

[0026] In some embodiments, the abnormal index pt of the group chat member within a set time period is obtained, specifically:

[0027] M2-101: At the current time point, extract the number of pending manual review and abnormal postings of the group chat member within the set time period, accumulate the number of pending manual review and abnormal postings, obtain the number of suspicious times of the group chat member within the set time period, and calculate with the set time period, that is, divide the suspicious number by the set time period to obtain the suspicious frequency of the group chat member, recorded as k1;

[0028] M2-102: Preset weight coefficients corresponding to pending manual review and abnormal release, respectively, multiply the number of pending manual review and abnormal release of the group chat member by the corresponding preset weight coefficients, and then sum them up to obtain the suspicious cumulative value of the group chat member; based on the preset conversion rule, convert the suspicious cumulative value of the group chat member into a suspicious score, which is recorded as k2;

[0029] M2-103: Preset each group of cumulative value ranges of suspicious cumulative values, and set each group of cumulative value ranges to correspond to a suspicious score; match the suspicious cumulative value of the group chat member in the set time period with the preset groups of cumulative value ranges, so as to obtain the suspicious score of the group chat member in the set time period;

[0030] M2-104: Obtain the time the group chat member has been in the group chat; the time the member has been in the group chat is counted from the time the member has entered the group chat, and the time difference with the current time is calculated; according to the time the member has been in the group chat, obtain the loyalty index of the member, and record it as k3; based on the preset conversion rule, that is, when the time the member has been in the group chat reaches the corresponding time range, automatically upgrade the group chat level of the member, and set each group chat level to correspond to a loyalty index;

[0031] M2-105: Based on the formula The suspicious frequency k1, suspicious score k2 and loyalty index k3 of the group chat member within the set time period are weightedly calculated to obtain the abnormal index pt of the group chat member within the set time period, where a1, a2 and a3 are the influencing weight factors of the suspicious frequency k1, suspicious score k2 and loyalty index k3 respectively.

[0032] In some embodiments, corresponding steps are performed to optimize the auditing steps, specifically:

[0033] When the group chat member publishes information content again, the time of the last information content published by the group chat member from the current time point is obtained, and the time difference is calculated with the current time point to obtain the publishing time difference of the group chat member;

[0034] Extract the abnormal index pt of the group chat member, preset the index value ranges of each group of the abnormal index pt, and set each group of index value ranges to correspond to an allowed time difference;

[0035] The abnormal index pt of the group chat member is matched with the preset range of index values ​​of each group to obtain the allowed time difference of the group chat member; the publishing time difference of the group chat member is compared with the allowed time difference. If the publishing time difference of the group chat member is less than the allowed time difference, the information content published again by the group chat member is directly approved.

[0036] In some embodiments, the hidden danger index qe of the group chat in the evaluation time period is obtained, specifically:

[0037] Extract the number of times that members of the group chat trigger warnings during the evaluation period, and record it as the number of warnings f1; at the same time, obtain the number of new members of the group chat during the evaluation period as the evaluation number; extract the number of new members of the group chat during the five evaluation periods before the current evaluation period, and calculate the average, so as to obtain the average number of the group chat in the five evaluation periods before the current evaluation period; calculate the ratio of the evaluation number and the average number, that is, through the evaluation number / average number; use the calculated ratio as the abnormal value of the group chat during the evaluation period f2; count the number of complaints against the group chat during the evaluation period and record it as f3;

[0038] According to the formula The number of warnings f1, personnel anomalies f2 and number of complaints f3 of group chat in the evaluation period are weightedly calculated to obtain the hidden danger index qe of group chat in the evaluation period; b1, b2 and b3 are the influence weight factors of the number of warnings f1, personnel anomalies f2 and number of complaints f3 respectively; z1 represents the maximum allowable value of the number of warnings f1 in the evaluation period, z2 represents the maximum allowable value of the number of personnel anomalies f2 in the evaluation period, and z3 represents the maximum allowable value of the number of complaints in the evaluation period.

[0039] In some embodiments, based on the hidden danger index qe of the group chat in the evaluation time period, the hidden danger level of the group chat in the evaluation time period is determined, specifically:

[0040] Three groups of index value ranges of the hidden danger index qe are preset, and each group of index value ranges is set to correspond to the hidden danger level of a group chat; the hidden danger levels include general hidden dangers, high hidden dangers and serious hidden dangers;

[0041] The hidden danger index qe of the group chat in the evaluation period is matched with the preset three groups of index value ranges, so as to obtain the hidden danger level of the group chat in the evaluation period and send it to the administrator of the group chat.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] When a group chat member publishes information content, the present invention performs a preliminary review of the information content published by the member. If the preliminary review result is to be manually reviewed, the information content published by the member is sent to the auditor who is in working state at the current time point. The accuracy of abnormal content detection is improved by partially combining manual review. At the same time, taking the current time point as the reference point, the abnormal data of the group chat member in the set time period is extracted and analyzed to obtain the abnormal index of the group chat member in the set time period. If the abnormal index of the group chat member in the set time period is greater than the preset threshold index, an early warning is triggered and sent to the administrator of the group chat, so as to achieve in-depth analysis of the group chat member corresponding to the abnormal content, and improve the comprehensiveness of the security mechanism.

[0044] The present invention presets an evaluation time interval for group chats, and after reaching the preset estimated time interval, analyzes the abnormal state of the group chat within the evaluation time period, thereby obtaining a hidden danger index for the group chat within the evaluation time period, and based on the hidden danger index for the group chat within the evaluation time period, determines the hidden danger level of the group chat within the evaluation time period, thereby solving the problem that most of the prior arts only carry out targeted solutions when problems occur in the group chat, and cannot set a fixed evaluation time interval to perform regular evaluation on the group chat, thereby determining the hidden danger level of the group chat, and realizes active security evaluation of the security mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Further details, features and advantages of the present application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0046] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0047] Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete, and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

[0048] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and / or this specification, and will not be interpreted in an idealized or overly formal sense, unless explicitly defined as such herein.

[0049] See also Figure 1As shown, the security mechanism execution method applied to group chat includes:

[0050] Permission configuration: Use the backend server to set permissions for group chat members; the administrator sets permissions based on the roles and needs of different members in the group chat; permissions include but are not limited to speaking permissions, information publishing permissions, and management permissions;

[0051] Permission configuration is implemented through a friendly member interface, and administrators can easily adjust and save settings to ensure the security and order of group chats;

[0052] Permission verification: When a group chat member attempts to initiate an operation request, such as speaking or publishing information, the group chat member's permission is automatically verified. If the permission verification passes, the operation request of the group chat member is allowed and executed. If the permission verification fails, the operation request of the group chat member is rejected and the reason is fed back to the member, such as lack of corresponding permission.

[0053] Information audit: When a group chat member posts information, a preliminary review of the information posted by the member is conducted;

[0054] Specifically:

[0055] S1: extract the information content posted by the group chat members to perform extension check, and execute corresponding steps based on the result of the extension check. If the check result indicates that the posted information content is text, execute step S1-101; if the check result indicates that the posted information content is a picture, execute step S1-102; if the check result indicates that the posted information content is video or audio, execute step S1-103;

[0056] S1-101: If the inspection result indicates that the published information content is text, after extracting the text content, use the pre-built sensitive word library to scan the text and match the text content with the words in the sensitive word library; use a string matching algorithm, such as the KMP algorithm, the Boyer-Moore algorithm or the Trie tree structure, to efficiently find sensitive words in the text; after a successful match, mark the position of the sensitive word in the text content, and after marking, count the number of marked sensitive words as the sensitive number, and divide the marked sensitive words into sensitivity levels. According to the pre-built sensitive word library, multiple sensitivity level intervals are constructed in the sensitive word library, and each sensitivity level interval is set to contain multiple groups of sensitive words, and each group of sensitive words is stored in the corresponding sensitivity level interval according to the corresponding sensitivity level; after dividing the marked sensitive words into sensitivity levels, the sensitivity level of each group of marked sensitive words is obtained;

[0057] The higher the sensitivity level, the higher the sensitivity of the group of sensitive words, and the higher the corresponding sensitivity base value;

[0058] Different sensitivity levels are set to correspond to a sensitivity base value, and the sensitivity base value is converted based on the sensitivity level of each group of marked sensitive words. After the conversion is completed, the sensitivity base values ​​of each group are accumulated to obtain the total sensitivity value;

[0059] Preset weight coefficients corresponding to the sensitive quantity and the total sensitive value, multiply the sensitive quantity and the total sensitive value by the corresponding weight coefficients respectively, and then sum them up to obtain the sensitivity index of the information content released by the member;

[0060] A reference threshold range of a preset sensitivity index is set, and the sensitivity index of the information content published by the member is compared with the preset reference threshold range. If the sensitivity index of the information content published by the member is within the preset reference threshold range, the information content published by the member is marked as pending manual review; if the sensitivity index of the information content published by the member is higher than the preset reference threshold range, the information content published by the member is marked as abnormal publication;

[0061] S1-102: If the check result indicates that the published information content is a picture, the picture is first preprocessed; the preprocessing includes steps such as denoising, grayscale, binarization, de-skew and correction; after the preprocessing is completed, the text in the picture is detected and extracted using OCR technology, and then step S1-101 is executed;

[0062] S1-103: If the check result indicates that the published information content is video or audio, the video content or audio content is converted into text using a transcription tool; the transcription tool is selected appropriately according to the needs, such as Scr i beBuddy, Deepgram, Video To Text AI, Videotowords AI, TurboScr i be, Yescr i be.ai, etc.; after conversion to text, the conversion errors in the converted text content are searched, and after the search is completed, step S1-101 is executed;

[0063] Translation error retrieval specifically uses text correction tools or algorithms to detect errors in the translated text; these tools may include rule-based systems, machine learning models, or deep learning models that can identify spelling, grammatical, or semantic errors in the text;

[0064] It should be noted that, in summary, by combining automation technology, effective management of group chat content is achieved, and the efficiency and accuracy of content review are improved.

[0065] Evaluation and execution: Based on the preliminary review results of the information released, the corresponding steps are executed;

[0066] Specifically:

[0067] M1: If the preliminary review result is pending manual review, the information content published by the member is sent to the auditor who is currently working. After the auditor reviews, if the information content published by the member is found to be abnormal, the preliminary review result of the information content published by the member is modified to abnormal release, and step M2 is executed;

[0068] M2: If the preliminary review result is abnormal release, the group chat member corresponding to the abnormal distribution is marked, and the abnormal data of the group chat member within the set time period is extracted and analyzed with the current time point as the reference point, so as to obtain the abnormal index pt of the group chat member within the set time period;

[0069] M2-101: At the current time point, extract the number of pending manual review and abnormal postings of the group chat member within the set time period, accumulate the number of pending manual review and abnormal postings, obtain the number of suspicious times of the group chat member within the set time period, and calculate with the set time period, that is, divide the suspicious number by the set time period to obtain the suspicious frequency of the group chat member, recorded as k1;

[0070] M2-102: Preset weight coefficients corresponding to pending manual review and abnormal release, respectively, multiply the number of pending manual review and abnormal release of the group chat member by the corresponding preset weight coefficients, and then sum them up to obtain the suspicious cumulative value of the group chat member; based on the preset conversion rule, convert the suspicious cumulative value of the group chat member into a suspicious score, which is recorded as k2;

[0071] M2-103: Preset each group of cumulative value ranges of suspicious cumulative values, and set each group of cumulative value ranges to correspond to a suspicious score; match the suspicious cumulative value of the group chat member in the set time period with the preset groups of cumulative value ranges, so as to obtain the suspicious score of the group chat member in the set time period;

[0072] The suspicious score range is set between 1 and 10, and the higher the suspicious cumulative value, the higher the suspicious score obtained by the corresponding match;

[0073] M2-104: Obtain the time the group chat member has been in the group chat; the time the member has been in the group chat is counted from the time the member has entered the group chat, and the time difference with the current time is calculated; according to the time the member has been in the group chat, obtain the loyalty index of the member and record it as k3; based on the preset conversion rule, that is, when the time the member has been in the group chat reaches the corresponding time range, automatically upgrade the group chat level of the member, and set a loyalty index corresponding to each group chat level; the loyalty index range is set to 1-5, and it gradually increases according to the time the member has been in the group chat;

[0074] M2-105: Based on the formula The suspicious frequency k1, suspicious score k2 and loyalty index k3 of the group chat member in the set time period are weighted to obtain the abnormal index pt of the group chat member in the set time period, where a1, a2 and a3 are the influencing weight factors of the suspicious frequency k1, suspicious score k2 and loyalty index k3 respectively;

[0075] M3: preset threshold index of abnormal index pt. If the abnormal index pt of the group chat member within the set time period is greater than the preset threshold index, an early warning is triggered and sent to the administrator of the group chat;

[0076] Audit optimization: Taking the time point when a group chat member posts information as the reference point, when the group chat member posts information again, obtain the abnormal index pt of the group chat member and perform corresponding steps to optimize the audit steps;

[0077] Specifically:

[0078] When the group chat member publishes information content again, the time of the last information content published by the group chat member from the current time point is obtained, and the time difference is calculated with the current time point to obtain the publishing time difference of the group chat member;

[0079] Extract the abnormal index pt of the group chat member, preset the index value ranges of each group of the abnormal index pt, and set each group of index value ranges to correspond to an allowed time difference; the larger the abnormal index pt, the shorter the allowed time difference of the corresponding match;

[0080] The abnormal index pt of the group chat member is matched with the preset index value ranges of each group, so as to obtain the allowed time difference of the group chat member; the publishing time difference of the group chat member is compared with the allowed time difference, and if the publishing time difference of the group chat member is less than the allowed time difference, the information content re-published by the group chat member is directly approved;

[0081] It should be noted that, in summary, when the group chat member publishes information content again, the abnormal index pt of the group chat member is obtained, and the publishing time difference and the allowed time difference of the group chat member are obtained, so as to determine whether the time interval for the group chat member to publish information content again is in a safe state. If the publishing time difference of the group chat member is less than the allowed time difference, the information content published again by the group chat member is directly approved, which improves the degree of intelligence and optimizes the review process.

[0082] For example,

[0083] There are two members in the group chat, and the information is as follows:

[0084] Member A: The abnormal index pt is lower because member A has a long group chat duration, good historical behavior, and rarely posts illegal content. The system matches A with a longer allowed time difference, such as 1 hour. This means that when A posts new information within 1 hour, the system will automatically pass it without additional review.

[0085] Member B: The abnormal index pt is high because he has posted many illegal contents in the past. The system matches B with a shorter allowed time difference, such as 10 minutes. When B posts information again, the system will calculate the posting time difference. If the time difference is greater than 10 minutes, the information will enter the review process again.

[0086] Periodic evaluation: preset the evaluation time interval of the group chat. After the preset estimated time interval is reached, analyze the abnormal status of the group chat in the evaluation time period to obtain the hidden danger index qe of the group chat in the evaluation time period. Based on the hidden danger index qe of the group chat in the evaluation time period, determine the hidden danger level of the group chat in the evaluation time period.

[0087] Specifically:

[0088] Extract the number of times group chat members trigger warnings during the evaluation period, recorded as the number of warnings f1; at the same time, obtain the number of new members in the group chat during the evaluation period as the evaluation number; evaluate the rate of new members joining in a specific period of time. An abnormal joining rate indicates the risk of bulk registration or malicious intrusion;

[0089] Extract the number of new members in the group chat in the five evaluation time periods before the current evaluation time period, and calculate the average value, so as to obtain the average number of group chats in the five evaluation time periods before the current evaluation time period;

[0090] The ratio of the evaluation quantity to the mean quantity is calculated, that is, the evaluation quantity / mean quantity; the calculated ratio is used as the personnel outlier f2 of the group chat during the evaluation period;

[0091] The number of complaints about the group chat during the evaluation period is counted and recorded as f3; group chat members file complaints about the group chat environment;

[0092] According to the formula The number of warnings f1, personnel abnormal values ​​f2, and number of complaints f3 of the group chat in the evaluation period are weighted and calculated to obtain the hidden danger index qe of the group chat in the evaluation period; b1, b2, and b3 are the influence weight factors of the number of warnings f1, personnel abnormal values ​​f2, and number of complaints f3, respectively; z1 represents the maximum allowable value of the number of warnings f1 in the evaluation period, z2 represents the maximum allowable value of the number of abnormal values ​​f2 in the evaluation period, and z3 represents the maximum allowable value of the number of complaints in the evaluation period;

[0093] Three groups of index value ranges of the hidden danger index qe are preset, and each group of index value ranges is set to correspond to the hidden danger level of a group chat; the hidden danger levels include general hidden dangers, high hidden dangers and serious hidden dangers;

[0094] The hidden danger index qe of the group chat in the evaluation period is matched with the preset three groups of index value ranges, so as to obtain the hidden danger level of the group chat in the evaluation period and send it to the administrator of the group chat;

[0095] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for executing a security mechanism applied to a group chat, characterized in that: include: Information audit: When a group chat member posts information, a preliminary review of the information posted by the member is conducted; Evaluation and execution: Based on the preliminary review results of the information released, the corresponding steps are executed; Audit optimization: Taking the time point when a group chat member posts information as the reference point, when the group chat member posts information again, obtain the abnormal index pt of the group chat member and perform corresponding steps to optimize the audit steps; Periodic evaluation: preset the evaluation time interval of the group chat. After the preset estimated time interval is reached, analyze the abnormal status of the group chat during the evaluation time period to obtain the hidden danger index qe of the group chat during the evaluation time period. Based on the hidden danger index qe of the group chat during the evaluation time period, determine the hidden danger level of the group chat during the evaluation time period.

2. The method for executing a security mechanism applied to group chat according to claim 1, characterized in that: Conduct a preliminary review of the information posted by the member, specifically: S1: extract the information content posted by the group chat members to perform extension check, and execute corresponding steps based on the result of the extension check. If the check result indicates that the posted information content is text, execute step S1-101; if the check result indicates that the posted information content is a picture, execute step S1-102; if the check result indicates that the posted information content is video or audio, execute step S1-103; S1-101: If the check result indicates that the published information content is text, after extracting the text content, use the pre-built sensitive word library to scan the text and match the text content with the words in the sensitive word library; after the match is successful, mark the position of the sensitive word in the text content, and after the marking is completed, count the number of marked sensitive words as the sensitive number, and divide the marked sensitive words into sensitivity levels. According to the pre-built sensitive word library, build multiple sensitivity level intervals in the sensitive word library, set each sensitivity level interval to contain multiple groups of sensitive words, and store each group of sensitive words in the corresponding sensitivity level interval according to the corresponding sensitivity level; After classifying the sensitive words into different sensitivity levels, the sensitivity level of each group of sensitive words is obtained; Different sensitivity levels are set to correspond to a sensitivity base value, and the sensitivity base value is converted based on the sensitivity level of each group of marked sensitive words. After the conversion is completed, the sensitivity base values ​​of each group are accumulated to obtain the total sensitivity value; Preset weight coefficients corresponding to the sensitive quantity and the total sensitive value, multiply the sensitive quantity and the total sensitive value by the corresponding weight coefficients respectively, and then sum them up to obtain the sensitivity index of the information content released by the member; A reference threshold range of the preset sensitivity index is used to compare the sensitivity index of the information content published by the member with the preset reference threshold range. If the sensitivity index of the information content published by the member is within the preset reference threshold range, the information content published by the member is marked as pending manual review; if the sensitivity index of the information content published by the member is higher than the preset reference threshold range, the information content published by the member is marked as abnormal publication.

3. The method for executing a security mechanism applied to group chat according to claim 2, characterized in that: Conduct a preliminary review of the information posted by the member, including: S1-102: If the check result indicates that the published information content is a picture, the picture is first preprocessed; after the preprocessing is completed, the text in the picture is detected and extracted, and then step S1-101 is executed; S1-103: If the check result indicates that the published information content is video or audio, the video content or audio content is converted into text using a transcription tool; after conversion into text, the conversion errors in the converted text content are searched, and after the search is completed, step S1-101 is executed.

4. The method for executing a security mechanism applied to group chat according to claim 3, characterized in that: Based on the preliminary review results of the published information content, the corresponding steps are carried out, specifically: M1: If the preliminary review result is pending manual review, the information content published by the member is sent to the auditor who is currently working. After the auditor reviews, if the information content published by the member is found to be abnormal, the preliminary review result of the information content published by the member is modified to abnormal release, and step M2 is executed; M2: If the preliminary review result is abnormal release, the group chat member corresponding to the abnormal distribution is marked, and the abnormal data of the group chat member within the set time period is extracted and analyzed with the current time point as the reference point, so as to obtain the abnormal index pt of the group chat member within the set time period; M3: preset the threshold index of the abnormal index pt. If the abnormal index pt of the group chat member within the set time period is greater than the preset threshold index, an early warning is triggered and sent to the administrator of the group chat.

5. The method for executing a security mechanism applied to group chat according to claim 4, characterized in that: Get the abnormal index pt of group chat members within the set time period, specifically: M2-101: At the current time point, extract the number of pending manual review and abnormal postings of the group chat member within the set time period, accumulate the number of pending manual review and abnormal postings, obtain the number of suspicious times of the group chat member within the set time period, and calculate with the set time period, that is, divide the suspicious number by the set time period to obtain the suspicious frequency of the group chat member, recorded as k1; M2-102: Preset weight coefficients corresponding to pending manual review and abnormal release, respectively, multiply the number of pending manual review and abnormal release of the group chat member by the corresponding preset weight coefficients, and then sum them up to obtain the suspicious cumulative value of the group chat member; based on the preset conversion rule, convert the suspicious cumulative value of the group chat member into a suspicious score, which is recorded as k2; M2-103: Preset each group of cumulative value ranges of suspicious cumulative values, and set each group of cumulative value ranges to correspond to a suspicious score; match the suspicious cumulative value of the group chat member in the set time period with the preset groups of cumulative value ranges, so as to obtain the suspicious score of the group chat member in the set time period; M2-104: Obtain the time the group chat member has been in the group chat; the time the member has been in the group chat is counted from the time the member has entered the group chat, and the time difference with the current time is calculated; according to the time the member has been in the group chat, obtain the loyalty index of the member, and record it as k3; based on the preset conversion rule, that is, when the time the member has been in the group chat reaches the corresponding time range, automatically upgrade the group chat level of the member, and set each group chat level to correspond to a loyalty index; M2-105: Based on the formula The suspicious frequency k1, suspicious score k2 and loyalty index k3 of the group chat member within the set time period are weightedly calculated to obtain the abnormal index pt of the group chat member within the set time period, where a1, a2 and a3 are the influencing weight factors of the suspicious frequency k1, suspicious score k2 and loyalty index k3 respectively.

6. The method for executing a security mechanism applied to group chat according to claim 5, characterized in that: Perform the corresponding steps to optimize the audit process, specifically: When the group chat member publishes information content again, the time of the last information content published by the group chat member from the current time point is obtained, and the time difference is calculated with the current time point to obtain the publishing time difference of the group chat member; Extract the abnormal index pt of the group chat member, preset the index value ranges of each group of the abnormal index pt, and set each group of index value ranges to correspond to an allowed time difference; The abnormal index pt of the group chat member is matched with the preset range of index values ​​of each group to obtain the allowed time difference of the group chat member; the publishing time difference of the group chat member is compared with the allowed time difference. If the publishing time difference of the group chat member is less than the allowed time difference, the information content published again by the group chat member is directly approved.

7. The method for executing a security mechanism applied to group chat according to claim 6, characterized in that: The hidden danger index qe of the group chat in the evaluation period is obtained, which is: Extract the number of times that members of the group chat trigger warnings during the evaluation period, and record it as the number of warnings f1; at the same time, obtain the number of new members of the group chat during the evaluation period as the evaluation number; extract the number of new members of the group chat during the five evaluation periods before the current evaluation period, and calculate the average, so as to obtain the average number of group chats in the five evaluation periods before the current evaluation period; calculate the ratio of the evaluation number and the average number, that is, through the evaluation number / average number; The calculated ratio is used as the outlier value f2 of the group chat during the evaluation period; Count the number of complaints received by the group chat during the evaluation period and record it as f3; According to the formula The number of warnings f1, personnel abnormality f2, and number of complaints f3 of the group chat in the evaluation period are weighted to obtain the hidden danger index qe of the group chat in the evaluation period; Among them, b1, b2 and b3 are the influencing weight factors of the number of warnings f1, the number of personnel abnormal values ​​f2 and the number of complaints f3 respectively; z1 represents the maximum allowable value of the number of warnings f1 during the evaluation period, z2 represents the maximum allowable value of the number of personnel abnormal values ​​f2 during the evaluation period, and z3 represents the maximum allowable value of the number of complaints during the evaluation period.

8. The method for executing a security mechanism applied to group chat according to claim 7, characterized in that: Based on the hidden danger index qe of the group chat during the evaluation period, the hidden danger level of the group chat during the evaluation period is determined as follows: Three groups of index value ranges of the hidden danger index qe are preset, and each group of index value ranges is set to correspond to the hidden danger level of a group chat; the hidden danger levels include general hidden dangers, high hidden dangers and serious hidden dangers; The hidden danger index qe of the group chat in the evaluation period is matched with the preset three groups of index value ranges, so as to obtain the hidden danger level of the group chat in the evaluation period and send it to the administrator of the group chat.

Images