Direct broadcasting room data processing method based on artificial intelligence AI

Abstract

The invention discloses a live broadcast room data processing method based on artificial intelligence (AI), and relates to the technical field of live broadcast room data processing. The carousel monitoring unit monitors carousel videos of the target live broadcasting room in real time to obtain carousel monitoring data of a corresponding carousel period when the carousel state is changed into the carousel state, and the carousel analysis unit analyzes the carousel monitoring data to determine a carousel drainage evaluation index of each carousel video; then, a carousel optimization unit performs optimization judgment on the carousel mode of the current target live broadcasting room based on the carousel drainage evaluation index of each carousel video, and in the process of performing optimization judgment on the carousel mode, different optimization judgment rules are adopted for performing optimization judgment according to different carousel modes; in this way, the accuracy of optimizing the carousel modes of the target live broadcasting room in different carousel modes is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of live broadcast room data processing, and in particular to a live broadcast room data processing method based on artificial intelligence (AI). Background Art

[0002] With the rapid development of Internet technology, the live broadcast industry is becoming increasingly prosperous. The live broadcast room has become an important place for people to obtain information, entertainment and interaction. During the live broadcast process, a large amount of data will be generated, including the number of viewers, interaction status, number of gifts, etc. These data are of great significance for evaluating the live broadcast effect, optimizing the live broadcast content and improving the user experience.

[0003] At present, after the live broadcast room is closed, with the authorization of the live broadcast platform, the anchor will often arrange a carousel to attract traffic and ensure the stability of traffic. At present, there are two main carousel modes provided by the platform for anchors, namely random carousel and weighted carousel. In the random carousel mode, the anchor only needs to upload several videos to be rotated, and the live broadcast platform will randomly control the playback of the carousel videos. In the weighted carousel mode, the anchor needs to set the carousel weights for several videos to be rotated. Usually, the anchor will set the weights of the carousel videos based on factors such as the number of viewers, interaction, and the number of gifts.

[0004] However, these two carousel modes do not fully consider the live broadcast traffic diversion in the non-live broadcast mode, and have certain limitations. In actual applications, this carousel mode may cause the carousel content to not match the audience's needs, and fail to maximize the carousel's traffic diversion effect;

[0005] In order to solve the above problems, the present invention proposes a solution. Summary of the invention

[0006] The purpose of the present invention is to provide a live broadcast room data processing method based on artificial intelligence AI, in order to solve the problems raised in the above background technology;

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] The live broadcast room data processing method based on artificial intelligence AI includes the following steps:

[0009] Step 1: When the live broadcast state of the target live broadcast room is monitored to be changed to the carousel state, the carousel monitoring unit monitors the carousel video played in the target live broadcast room in real time to obtain the carousel monitoring data of a carousel cycle, and transmits the carousel monitoring data to the carousel optimization unit for storage;

[0010] Step 2: At each carousel analysis period, the carousel analysis unit analyzes all the carousel monitoring data stored in the carousel analysis period according to a preset carousel analysis rule to generate carousel analysis data corresponding to the carousel analysis period;

[0011] Step three: After each carousel analysis cycle's carousel analysis data is generated, the carousel optimization unit will make an optimization decision on the carousel mode of the current target live broadcast room according to the preset optimization decision rules. When it is determined that optimization is required, the corresponding carousel analysis data will be transmitted to the carousel activation unit for update and storage, and the carousel mode of the current target live broadcast room will be changed to smart carousel.

[0012] Furthermore, before proceeding to step 1, the following steps need to be performed:

[0013] SS31: When the live broadcast state of the target live broadcast room is detected to be changed from the broadcast state to the non-broadcast state, the live broadcast state monitoring module generates a carousel activation instruction and transmits the carousel activation instruction to the carousel activation unit;

[0014] SS32: After receiving the transmitted carousel activation instruction, the carousel activation unit first changes the live broadcast state of the target live broadcast room to the carousel state, and then obtains the carousel mode of the current target live broadcast room, wherein the above-mentioned carousel modes are divided into three types, namely, random carousel, weighted carousel and intelligent carousel;

[0015] Furthermore, the live broadcast status of the target live broadcast room is divided into three types, namely, the non-broadcast status, the carousel status and the broadcast status.

[0016] Furthermore, the carousel analysis rules for generating the carousel analysis data of the current carousel analysis cycle are as follows:

[0017] S11: Mark all received carousel monitoring data in order of time of receipt as A1, A2...Aa, a≥1;

[0018] S12: traverse the carousel monitoring data A1, A2...Aa, obtain all the carousel videos contained therein, deduplicate all the obtained carousel videos, and mark all the remaining carousel videos after deduplication as B1, B2...Bb, where b≥1;

[0019] S13: extracting all the carousel monitoring data including the carousel video B1 from the carousel monitoring data A1, A2...Aa, and re-marking all the extracted carousel monitoring data as C1, C2...Cc in the order of the carousel monitoring data A1, A2...Aa, 1≤c≤a;

[0020] S14: Calculate and obtain the evaluation amount J1 of the carousel traffic diversion of the carousel video B1 based on the concerned user according to the preset first calculation rule;

[0021] S15: Calculate and obtain the evaluation amount R1 of the carousel traffic diversion of the carousel video B1 based on non-following users according to the preset second calculation rule;

[0022] S16: Calculate and obtain the carousel traffic evaluation index T1 of the carousel video B1 using the formula T1=J1×δ1+R1×δ2, where δ1 and δ2 are the first and second preset weights respectively;

[0023] S17: according to S13 to S16, the carousel traffic evaluation indexes of the carousel videos B2, B3, ... Bb are calculated in sequence and combined with the carousel video B1 to generate the carousel analysis data of the current carousel analysis period.

[0024] Beneficial effects of the present invention:

[0025] The present invention monitors the live broadcast state of the target live broadcast room in real time through the live broadcast state monitoring module, and when the state is changed to the carousel state, the carousel monitoring unit monitors the carousel video of the target live broadcast room in real time to obtain the carousel monitoring data of the corresponding carousel cycle, and the carousel analysis unit analyzes the carousel monitoring data to determine the carousel drainage evaluation index of each carousel video, and then the carousel optimization unit optimizes and determines the carousel mode of the current target live broadcast room based on the carousel drainage evaluation index of each carousel video, and in the process of optimizing and determining the carousel mode, different optimization determination rules are used for different carousel modes to optimize and determine, and in this way, the accuracy of the optimization of the carousel mode of the target live broadcast room in different carousel modes is guaranteed;

[0026] In the process of analyzing and obtaining the evaluation index of the carousel diversion of each carousel video, the present invention introduces the browsing frequency of each carousel video by concerned users and non-concerned users, determines the diversion degree of each carousel video in different modes, and judges and changes the carousel mode based on this. In this way, the change judgment is made in combination with the diversion situation of the carousel video in the non-live broadcast mode, ensuring that the carousel time in different carousel time periods of the optimized changed carousel mode plays a diversion role to a certain extent;

[0027] In addition to providing random carousel and weighted carousel, the present invention also provides intelligent carousel, and the intelligent carousel provides automatic change after analysis and judgment, making the carousel more intelligent and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention will be further described below in conjunction with the accompanying drawings.

[0029] Figure 1 is a flow chart of the method of the present invention;

[0030] Figure 2 It is a flow chart of the steps for generating carousel analysis data of the current carousel analysis cycle in the present invention;

[0031] Figure 3 It is a system block diagram of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Embodiment 1, as Figure 1 , 2 As shown, a live broadcast room data processing method based on artificial intelligence AI is performed by a live broadcast room data processing system based on artificial intelligence AI, and the system includes a live broadcast status monitoring module and a carousel management module;

[0034] The live broadcast status monitoring module is used to monitor the live broadcast status of the target live broadcast room in real time, wherein the live broadcast status is divided into three types, namely, the non-broadcasting status, the rotation status and the broadcasting status;

[0035] When the live broadcast status of the target live broadcast room is monitored to be changed from the broadcast status to the non-broadcast status, the live broadcast status monitoring module generates a carousel activation instruction, and transmits the carousel activation instruction to the carousel management module, wherein the live broadcast status of the target live broadcast room is changed from the broadcast status to the non-broadcast status manually by the anchor of the target live broadcast room;

[0036] A carousel management module, used to perform carousel management on the target live broadcast room, the carousel management module includes a carousel activation unit, a carousel monitoring unit, a carousel analysis unit and a carousel optimization unit;

[0037] The carousel activation unit pre-stores a number of carousel videos and their video names that are manually uploaded by the host of the target live broadcast room and have been reviewed by the target live broadcast room platform. In this embodiment, one of the carousel videos is a complete video data obtained by the host of the target live broadcast room by cutting and editing a number of live playback videos;

[0038] After receiving the transmitted carousel activation instruction, the carousel management module transmits it to the carousel activation unit. After receiving the transmitted carousel activation instruction, the carousel activation unit first changes the live broadcast state of the target live broadcast room to the carousel state, and then obtains the carousel mode of the current target live broadcast room. The above-mentioned carousel modes are divided into three types, namely random carousel, weighted carousel and intelligent carousel. In this embodiment, the carousel mode is manually set by the anchor of the target live broadcast room. When the carousel mode is set to weighted carousel, it is also necessary to set the carousel weight for all currently stored carousel videos;

[0039] If the carousel mode is random carousel, a number of carousel videos pre-stored in the carousel activation unit are randomly played until the live broadcast state of the target live broadcast room is changed to the start state by the anchor of the target live broadcast room;

[0040] One of the carousel cycles starts from the moment when the live broadcast status of the target live broadcast room changes to the non-live broadcast status and ends when the live broadcast status of the target live broadcast room changes to the live broadcast status;

[0041] If the carousel mode is weighted carousel, all the carousel videos pre-stored in the carousel activation unit are played in sequence according to the carousel weight from large to small, until the live broadcast state of the target live broadcast room is changed to the live broadcast state by the anchor of the target live broadcast room, wherein the carousel video with a larger carousel weight is played first;

[0042] When it is detected that the live broadcast state of the target live broadcast room is changed to the carousel state, the carousel monitoring unit monitors the carousel video played in the target live broadcast room in real time and obtains the carousel monitoring data of a carousel cycle after the live broadcast state of the target live broadcast room is changed to the broadcast state, and transmits the carousel monitoring data of the carousel cycle to the carousel optimization unit for storage;

[0043] The carousel monitoring data includes the video names and flow duration data of several carousel videos; the flow duration data of one carousel video includes all concerned users who browsed the carousel video, the first browsing duration, all non-concerned users and the second browsing duration;

[0044] The first browsing time refers to the average browsing time of all following users browsing the video carousel, and the second browsing time refers to the average browsing time of all non-following users browsing the video carousel;

[0045] Following users refer to users who follow the target live broadcast room, and non-following users refer to users who do not follow the target live broadcast room;

[0046] In this embodiment, the browsing time of the corresponding carousel video by all the following users and non-following users included in the carousel monitoring data is at least P1;

[0047] Embodiment 2 is implemented on the basis of embodiment 1;

[0048] At each interval of a carousel analysis cycle, the carousel analysis unit analyzes all the carousel monitoring data stored in the carousel analysis cycle according to the preset carousel analysis rules to generate carousel parsing data corresponding to the carousel analysis cycle and transmits it to the carousel optimization unit. The interval duration of a carousel analysis cycle is the preset carousel analysis duration P1;

[0049] The carousel analysis rules for generating carousel analysis data for the current carousel analysis cycle are as follows:

[0050] S11: Mark all received carousel monitoring data in order of time of receipt as A1, A2...Aa, a≥1;

[0051] S12: traverse the carousel monitoring data A1, A2...Aa, obtain all the carousel videos contained therein, deduplicate all the obtained carousel videos, and mark all the remaining carousel videos after deduplication as B1, B2...Bb, where b≥1;

[0052] S13: extracting all the carousel monitoring data including the carousel video B1 from the carousel monitoring data A1, A2...Aa, and re-marking all the extracted carousel monitoring data as C1, C2...Cc in the order of the carousel monitoring data A1, A2...Aa, 1≤c≤a;

[0053] S14: Calculate and obtain the evaluation amount J1 of the carousel traffic diversion of the carousel video B1 based on the concerned user according to a preset first calculation rule. The first calculation rule is as follows:

[0054] S141: Acquire all the followed users included in the carousel monitoring data C1, C2...Cc, and remove duplicates from all the obtained followed users, and mark all the remaining followed users after removal of duplicates as D1, D2...Dd, where d≥1;

[0055] S142: Obtain the total number of carousel monitoring data containing the attention user D1 from the carousel monitoring data C1, C2...Cc, and mark the total number as the browsing frequency E1 of the attention user D1. Similarly, the browsing frequencies E2, E3...Ed of the attention users D2, D3...Dd can be obtained in sequence. Combined with the browsing frequency E1, the browsing frequency with the largest number of occurrences is obtained, and the browsing frequency is used as the attention frequency benchmark G1 of the carousel video B1;

[0056] S143: Compare all the concerned users included in the carousel monitoring data C1 and C2 for consistency, obtain the total number of concerned users included in the carousel monitoring data C1 but not in the carousel monitoring data C2, label the total number as the basic concerned traffic G2 of the carousel video B1, and at the same time obtain the total number of concerned users included in the carousel monitoring data C2 but not in the carousel monitoring data C1, label the total number as the fault-tolerant concerned traffic F1 of the carousel monitoring data C2. Similarly, the fault-tolerant concerned traffic F2, F3... Fc-1 of the carousel monitoring data C3, C4... Cc can be obtained in turn;

[0057] It should be noted here that in the process of calculating the fault-tolerant concerned traffic F2, all the concerned users included in the carousel monitoring data C3 and C1, C2 are compared for consistency, and the total number of concerned users included in the carousel monitoring data C3 but not in the carousel monitoring data C1, C2 is obtained;

[0058] S144: Use the formula to calculate the deviation H1 of the fault-tolerant concerned traffic F1, F2... Fc-1, and compare the sizes of H1 and H, where F is the average value of Ff at this time, and H is the preset standard fault-tolerant concerned traffic deviation comparison value;

[0059] If H1≥H, then delete the largest Fh in descending order of |Fh - F| in turn and calculate the deviation H1 of the remaining Fh, and compare the sizes of H1 and H again until H1 < H. Obtain the average value of all the fault-tolerant concerned traffic participating in the calculation of the remaining deviation H1 when H1 < H at this time, and label the average value as the concerned fault-tolerant traffic benchmark G3 of the carousel video B1;

[0060] S145: Use the formula J1 = Z1×{G1×β1 + [G2×(1 - G3×ɑ1) - 1]×β2} to calculate the carousel drainage evaluation amount J1 of the carousel video B1 based on the concerned users, where ɑ1 is the preset first fault-tolerant traffic adjustment factor, β1 and β2 are the preset first and second concerned balance dimension adjustment factors respectively, and Z1 is the first viewing duration of the carousel video B1;

[0061] S15: Calculate the carousel drainage evaluation amount R1 of the carousel video B1 based on non-concerned users according to the preset second calculation rule. The second calculation rule is as follows:

[0062] S151: Obtain all the non-concerned users included in the carousel monitoring data C1, C2... Cc, and deduplicate all the obtained non-concerned users. Mark the remaining non-concerned users after deduplication as K1, K2... Kk respectively, where k≥1;

[0063] S152: Obtain the total number of carousel monitoring data containing non - concerned user K1 from carousel monitoring data C1, C2... Cc, and label this total number as the browsing frequency L1 of non - concerned user K1. Similarly, the browsing frequencies L2, L3... Lk of non - concerned users K2, K3... Kk can be obtained in turn. Combine the browsing frequency L1 and obtain the browsing frequency that appears the most times among them. Use this browsing frequency as the attention frequency benchmark M1 of the carousel video B1;

[0064] S153: Conduct a consistency comparison on all non - concerned users included in carousel monitoring data C1 and C2, obtain the total number of non - concerned users included in carousel monitoring data C1 but not in carousel monitoring data C2, and label this total number as the basic non - concerned traffic M2 of carousel video B1. At the same time, obtain the total number of non - concerned users included in carousel monitoring data C2 but not in carousel monitoring data C1, and label this total number as the fault - tolerant non - concerned traffic N1 of carousel monitoring data C2. Similarly, the fault - tolerant non - concerned traffic N2, N3... Nc - 1 of carousel monitoring data C3, C4... Cc can be obtained in turn;

[0065] It should be noted here that in the process of calculating the fault - tolerant non - concerned traffic N2, all non - concerned users included in carousel monitoring data C3 and C1, C2 are compared for consistency, and the total number of non - concerned users included in carousel monitoring data C3 but not in carousel monitoring data C1, C2 is obtained;

[0066] S154: Use the formula to calculate the deviation Q1 of the fault - tolerant non - concerned traffic N1, N2... Nc - 1, and compare the sizes of Q1 and Q, where N is the average value of Nq at this time, and Q is a preset standard fault - tolerant non - concerned traffic deviation comparison value;

[0067] If Q1≥Q, then delete the largest NQ in descending order according to |N - N| and calculate the deviation Q1 of the remaining NQ. Compare the sizes of Q1 and Q again until Q1 < Q. Obtain the average value of all fault - tolerant non - concerned traffic participating in the calculation of the remaining deviation Q1 at this time when Q1 < Q, and label this average value as the attention fault - tolerant traffic benchmark M3 of the carousel video B1;

[0068] S155: Use the formula R1 = Z2×{M1×θ1+[M2×(1 - M3×ɑ2)-1]×θ2} to calculate the carousel drainage evaluation amount J1 of the carousel video B1 based on non - concerned users, where ɑ1 is a preset second fault - tolerant traffic adjustment factor, θ1 and θ2 are preset first and second non - concerned balance dimension adjustment factors respectively, and Z2 is the second browsing duration of the carousel video B1;

[0069] S16: Calculate and obtain the carousel traffic evaluation index T1 of the carousel video B1 using the formula T1=J1×δ1+R1×δ2, where δ1 and δ2 are the first and second preset weights respectively;

[0070] S17: according to S13 to S16, the carousel traffic evaluation indexes of the carousel videos B2, B3, ... Bb are calculated in sequence, combined with the carousel video B1 to generate the carousel analysis data of the current carousel analysis period, and the carousel analysis data of the current carousel analysis period is transmitted to the carousel optimization unit;

[0071] Embodiment 3 is implemented on the basis of Embodiments 1 and 2. After receiving the transmitted carousel analysis data of the current carousel analysis period, the carousel optimization unit determines whether to optimize the carousel mode of the current target live broadcast room according to the preset optimization determination rules. The optimization determination rules are as follows:

[0072] S21: marking all the carousel videos in the carousel parsing data as U1, U2, ..., Ub in order according to the carousel traffic evaluation index from large to small;

[0073] The carousel traffic evaluation indicators of the carousel videos U1, U2...Ub in the carousel parsing data are marked as V1, V2...Vb respectively;

[0074] S22: Obtain the carousel mode of the current target live broadcast room. If the carousel mode is random carousel, determine whether the carousel mode of the current target live broadcast room needs to be optimized according to the preset first optimization determination rule. The first optimization determination rule is as follows:

[0075] SS11: Use the formula W1=V1-V2 to calculate the traffic difference W1 between the carousel videos U1 and U2. Similarly, the traffic differences W2, W3, ...Wu-1 between the carousel videos U2 and U3, U3 and U4, ...Uu-1 and Uu can be obtained in sequence.

[0076] SS12: If P3≤x1 / (u-1) is satisfied, it is determined that the current live broadcast room's carousel mode does not need to be optimized, otherwise it is determined that the current live broadcast room's carousel mode needs to be optimized;

[0077] Where x1 is the number of drainage differences among drainage differences W1, W2, W3...Wu-1 whose values ​​are less than or equal to the preset drainage difference standard threshold W, and P3 is the preset random carousel optimization judgment weight threshold;

[0078] S23: If the carousel mode is weighted carousel, whether the carousel mode of the current target live broadcast room needs to be optimized is determined according to the preset second optimization determination rule. The second optimization determination rule is as follows:

[0079] SS21: Get the carousel weights of the carousel videos U1, U2...Ub, and recalibrate the carousel videos U1, U2...Ub in descending order of the carousel weights to Y1, Y2...Yb:

[0080] SS22: Perform consistency comparison on the carousel videos U1, U2...Ub and the carousel videos Y1, Y2...Yb, and obtain the video names of the carousel videos and the total number x2 of the carousel videos whose subscripts are consistent with each other;

[0081] SS23: If P4≤x2 is satisfied, it is determined that the carousel mode of the current live broadcast room does not need to be optimized, otherwise it is determined that the carousel mode of the current live broadcast room needs to be optimized, and the carousel analysis data of the current carousel analysis cycle is transmitted to the carousel activation unit. P4 is the preset weighted carousel optimization determination threshold;

[0082] S24: if the carousel mode is the smart carousel, at this time, the carousel traffic evaluation index of the corresponding carousel video stored in the carousel activation unit is updated according to the carousel traffic evaluation indexes of the several carousel videos contained in the received carousel parsing data;

[0083] If the carousel activation unit receives the transmitted carousel parsing data of the current carousel analysis period, it updates and stores it, and at the same time changes the carousel mode of the current live broadcast room to smart carousel, when the transmitted carousel activation instruction is received again, all the carousel videos and their carousel traffic evaluation indicators contained in the carousel parsing data currently stored in the carousel activation unit are first obtained, and then the corresponding carousel videos are played in descending order according to the carousel traffic evaluation indicators;

[0084] In the description of the specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0085] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the invention or exceed the scope defined by the claims, they shall all fall within the protection scope of the present invention.

[0086] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A live broadcast room data processing method based on artificial intelligence AI, characterized in that: The following steps are involved: Step 1: When the live broadcast state of the target live broadcast room is monitored to be changed to the carousel state, the carousel monitoring unit monitors the carousel video played in the target live broadcast room in real time to obtain the carousel monitoring data of a carousel cycle, and transmits the carousel monitoring data to the carousel optimization unit for storage; Step 2: At each carousel analysis period, the carousel analysis unit analyzes all the carousel monitoring data stored in the carousel analysis period according to a preset carousel analysis rule to generate carousel analysis data corresponding to the carousel analysis period; Step three: After each carousel analysis cycle's carousel analysis data is generated, the carousel optimization unit will make an optimization decision on the carousel mode of the current target live broadcast room according to the preset optimization decision rules. When it is determined that optimization is required, the corresponding carousel analysis data will be transmitted to the carousel activation unit for update and storage, and the carousel mode of the current target live broadcast room will be changed to smart carousel.

2. The method for processing live broadcast room data based on artificial intelligence AI according to claim 1, characterized in that: Before proceeding to step 1, you need to do the following: SS31: When the live broadcast state of the target live broadcast room is detected to be changed from the broadcast state to the non-broadcast state, the live broadcast state monitoring module generates a carousel activation instruction and transmits the carousel activation instruction to the carousel activation unit; SS32: After receiving the transmitted carousel activation instruction, the carousel activation unit first changes the live broadcast state of the target live broadcast room to the carousel state, and then obtains the carousel mode of the current target live broadcast room, wherein the above-mentioned carousel modes are divided into three types, namely random carousel, weighted carousel and intelligent carousel.

3. The method for processing live broadcast room data based on artificial intelligence AI according to claim 2 is characterized in that: In SS32, if the carousel mode is random carousel, a number of carousel videos pre-stored in the carousel activation unit are randomly played until the live broadcast state of the target live broadcast room is changed to the start state by the anchor of the target live broadcast room; If the carousel mode is weighted carousel, all the carousel videos pre-stored in the carousel activation unit are played in sequence according to the carousel weight from large to small, until the live broadcast status of the target live broadcast room is changed to the start status by the anchor of the target live broadcast room; If the carousel mode is smart carousel, all carousel videos pre-stored in the carousel activation unit are played in sequence according to the carousel traffic evaluation indicators from large to small, until the live broadcast status of the target live broadcast room is changed to the live broadcast status by the anchor of the target live broadcast room.

4. The method for processing live broadcast room data based on artificial intelligence AI according to claim 1, characterized in that: There are three live broadcast statuses in the target live broadcast room: not yet broadcast, carousel, and broadcast.

5. The method for processing live broadcast room data based on artificial intelligence AI according to claim 1, characterized in that: The carousel monitoring data includes the video names and flow duration data of several carousel videos; the flow duration data of one of the carousel videos includes all the concerned users who browsed the carousel video, the first browsing duration, all the non-concerned users and the second browsing duration.

6. The method for processing live broadcast room data based on artificial intelligence AI according to claim 1, characterized in that: The carousel analysis rules for generating carousel analysis data for the current carousel analysis cycle are as follows: S11: Mark all received carousel monitoring data in order of time of receipt as A1, A2...Aa, a≥1; S12: traverse the carousel monitoring data A1, A2...Aa, obtain all the carousel videos contained therein, deduplicate all the obtained carousel videos, and mark all the remaining carousel videos after deduplication as B1, B2...Bb, where b≥1; S13: extracting all the carousel monitoring data including the carousel video B1 from the carousel monitoring data A1, A2...Aa, and re-marking all the extracted carousel monitoring data as C1, C2...Cc in the order of the carousel monitoring data A1, A2...Aa, 1≤c≤a; S14: Calculate and obtain the evaluation amount J1 of the carousel traffic diversion of the carousel video B1 based on the concerned user according to the preset first calculation rule; S15: Calculate and obtain the evaluation amount R1 of the carousel traffic diversion of the carousel video B1 based on non-following users according to the preset second calculation rule; S16: Calculate and obtain the carousel traffic evaluation index T1 of the carousel video B1 using the formula T1=J1×δ1+R1×δ2, where δ1 and δ2 are the first and second preset weights respectively; S17: according to S13 to S16, the carousel traffic evaluation indexes of the carousel videos B2, B3, ... Bb are calculated in sequence and combined with the carousel video B1 to generate the carousel analysis data of the current carousel analysis period.

7. The method for processing live broadcast room data based on artificial intelligence AI according to claim 1, characterized in that: After generating the carousel analysis data of the current carousel analysis cycle, the carousel optimization unit performs an optimization judgment on the carousel mode of the current target live broadcast room. The optimization judgment rules are as follows: S21: marking all the carousel videos in the carousel parsing data as U1, U2, ..., Ub in order according to the carousel traffic evaluation index from large to small; The carousel traffic evaluation indicators of the carousel videos U1, U2...Ub in the carousel parsing data are marked as V1, V2...Vb respectively; S22: Acquire the carousel mode of the current target live broadcast room. If the carousel mode is a random carousel, determine whether the carousel mode of the current target live broadcast room needs to be optimized according to a preset first optimization determination rule. If it is determined that the carousel mode of the current live broadcast room needs to be optimized, transmit the carousel analysis data of the current carousel analysis period to the carousel activation unit; S23: If the carousel mode is a weighted carousel, whether the carousel mode of the current target live broadcast room needs to be optimized is determined according to a preset second optimization determination rule. If it is determined that the carousel mode of the current live broadcast room needs to be optimized, the carousel parsing data of the current carousel analysis period is transmitted to the carousel activation unit; S24: If the carousel mode is smart carousel, the carousel traffic evaluation index of the corresponding carousel video stored in the carousel activation unit is updated according to the carousel traffic evaluation indexes of the several carousel videos contained in the received carousel parsing data.

8. The method for processing live broadcast room data based on artificial intelligence AI according to claim 7, characterized in that: In S22, the first optimization determination rule for determining whether the carousel mode of the current target live broadcast room needs to be optimized is as follows: SS11: Use the formula W1=V1-V2 to calculate the traffic difference W1 between the carousel videos U1 and U2. Similarly, the traffic differences W2, W3, ...Wu-1 between the carousel videos U2 and U3, U3 and U4, ...Uu-1 and Uu can be obtained in sequence. SS12: If P3≤x1 / (u-1) is satisfied, it is determined that the current live broadcast room's carousel mode does not need to be optimized, otherwise it is determined that the current live broadcast room's carousel mode needs to be optimized; Among them, x1 is the number of drainage differences among the drainage differences W1, W2, W3...Wu-1 whose values ​​are less than or equal to the preset drainage difference standard threshold W, and P3 is the preset random carousel optimization judgment weight threshold.

9. The method for processing live broadcast room data based on artificial intelligence AI according to claim 7, characterized in that: In S23, the second optimization determination rule for determining whether the carousel mode of the current target live broadcast room needs to be optimized is as follows: SS21: Get the carousel weights of the carousel videos U1, U2...Ub, and recalibrate the carousel videos U1, U2...Ub in descending order of the carousel weights to Y1, Y2...Yb: SS22: Perform consistency comparison on the carousel videos U1, U2...Ub and the carousel videos Y1, Y2...Yb, and obtain the video names of the carousel videos and the total number x2 of the carousel videos whose subscripts are consistent with each other; SS23: If P4≤x2 is satisfied, it is determined that the carousel mode of the current live broadcast room does not need to be optimized. Otherwise, it is determined that the carousel mode of the current live broadcast room needs to be optimized. The carousel analysis data of the current carousel analysis cycle is transmitted to the carousel activation unit. P4 is the preset weighted carousel optimization judgment threshold.

10. The method for processing live broadcast room data based on artificial intelligence AI according to claim 1, characterized in that: If the carousel activation unit receives the transmitted carousel parsing data of the current carousel analysis period, it updates and stores it, and at the same time changes the carousel mode of the current live broadcast room to smart carousel, when the transmitted carousel activation instruction is received again, it first obtains all the carousel videos and their carousel traffic evaluation indicators contained in the carousel parsing data currently stored in the carousel activation unit, and then plays the corresponding carousel videos in order from large to small according to the carousel traffic evaluation indicators.

Images