Scheduling management method and system based on live broadcast data

Abstract

The invention relates to the technical field of scheduling management, in particular to a scheduling management method and system based on live broadcast data, and the method comprises the following steps: based on live broadcast activity key data, including the number of viewers, comment frequency, like and sharing times and performance indexes of live broadcast personnel, the indexes include an interaction quality score, audience feedback information and a conversion rate; and executing data cleaning and integration processing to obtain a live broadcast data integration set. According to the invention, high-frequency updating and real-time analysis are carried out through live broadcast activity key data, so that the scheduling can be dynamically adjusted to meet the actual work requirements, and the ARIMA model is adopted, so that the prediction accuracy is enhanced, the scheduling decision has perspectiveness, the human resource allocation is optimized, and the scheduling efficiency is improved. And the clustering analysis helps to accurately identify the performance modes of different live broadcast personnel, and optimizes the working schedule according to the modes, thereby ensuring that the most suitable personnel are deployed in the time period with the highest audience activeness, and improving the employee satisfaction degree and the audience service quality.

Description

Technical Field

[0001] The present invention relates to the technical field of shift management, and in particular to a shift management method and system based on live broadcast data. Background Art

[0002] Scheduling management technology focuses on how to effectively arrange and optimize employees' working hours and tasks to improve efficiency and meet the needs of different work scenarios. It covers everything from simple manual scheduling methods to complex computer-assisted scheduling systems. With the development of technology, scheduling management has begun to adopt more data analysis and artificial intelligence algorithms to predict demand and optimize human resource allocation, and handle emergencies.

[0003] Among them, the scheduling management method of live broadcast data is to use live broadcast or real-time data to dynamically adjust and optimize the scheduling. It can instantly collect and analyze data from live broadcasts, such as customer flow, number of service requests or workload changes, and adjust employees' working hours and task allocations in real time based on the data. Its main purpose is to be able to respond quickly to on-site changes, improve service efficiency and employee satisfaction, and is especially suitable for industries with large demand fluctuations and frequent customer interactions, such as catering services, retail, and medical emergencies.

[0004] Existing technologies rely on preset working modes and less updated data, and are slow to respond when demand changes dramatically. For example, when traffic suddenly increases, fixed schedules cannot be adjusted in time, resulting in service delays and audience dissatisfaction. In addition, the scheduling in existing technologies fails to effectively predict peak and trough periods, resulting in uneven staffing. Employees will be overworked during periods of low demand and understaffed during periods of high demand, which not only reduces service efficiency but also increases employee turnover. Summary of the invention

[0005] In order to solve the technical problems that the existing technology relies on preset working modes and less updated data, and is slow to respond when demand changes sharply, for example, when traffic suddenly increases, the fixed schedule cannot be adjusted in time, resulting in service delays and audience dissatisfaction. In addition, the schedule in the existing technology fails to effectively predict peak and trough periods, resulting in uneven staffing, employees will be overworked during low-demand periods, and insufficient staff during high-demand periods, which not only reduces service efficiency but also increases employee turnover, the embodiment of the present invention provides a scheduling management method and system based on live broadcast data. The technical solution is as follows:

[0006] On the one hand, a scheduling management method based on live broadcast data is provided, comprising the following steps:

[0007] S1: Based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcaster, including the interaction quality score, audience feedback information and conversion rate, perform data cleaning and integration processing to obtain the live broadcast data integration set;

[0008] S2: Based on the live broadcast data integration set, using the ARIMA model, analyze the periodicity and trend characteristics of the live broadcast activity, identify the peak time period of audience interaction, and obtain audience interaction prediction information;

[0009] S3: Based on the audience interaction prediction information, the random forest algorithm is used to process the real-time interaction data, and the needs of the live broadcast personnel in the future time period are predicted to obtain future demand prediction data;

[0010] S4: Based on the future demand forecast data, cluster analysis is performed to identify the performance patterns of different live broadcast personnel, and the work schedule of the live broadcast personnel is refined according to the identification results to match the time period with the best audience activity to obtain the optimal scheduling configuration;

[0011] S5: Based on the optimized scheduling configuration, the performance trend of the live broadcast personnel in multiple time periods is predicted, and the prediction results are used to formulate a scheduling plan for the live broadcast platform, determine the personnel configuration that matches the optimal audience service and interaction, and obtain the live broadcast scheduling management result.

[0012] On the other hand, the live broadcast data integration set includes audience behavior characteristics, live broadcast personnel interaction index, and key performance indicators. The audience interaction prediction information includes peak time periods, activity cycles, and interaction trends. The future demand forecast data includes demand forecast cycles, demand change trends, and key demand indicators. The scheduling optimization configuration includes time period matching information, personnel deployment configuration, and efficiency optimization results. The live broadcast scheduling management results include determined schedules, key personnel configurations, and interaction efficiency optimization results.

[0013] On the other hand, based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcast personnel, including the interaction quality score, audience feedback information and conversion rate, the data is cleaned and integrated to obtain the integrated set of live broadcast data. The specific steps are as follows:

[0014] S101: Based on the key data of the live broadcast activity, the number of viewers, comment frequency, and number of likes and shares are extracted, non-compliant records are excluded, and the performance indicators of the live broadcast personnel are integrated to obtain a cleaned data set;

[0015] S102: based on the cleaned data set, unify the timestamps of multiple data sources, merge similar data fields, verify the integrity and availability of the data, and obtain an integrated data set;

[0016] S103: Based on the integrated data set, extract key indicators, including interaction quality score, audience feedback information and conversion rate, analyze the indicator data, and calculate the average value and change trend to obtain the integrated live broadcast data set.

[0017] On the other hand, based on the live broadcast data integration set, the ARIMA model is used to analyze the periodicity and trend characteristics of the live broadcast activity, identify the peak time period of audience interaction, and obtain the audience interaction prediction information in the following steps:

[0018] S201: Based on the live broadcast data integration set, using the ARIMA model, identifying repetitive cycles in the data, monitoring natural fluctuations of the data, and separating key periodic patterns to obtain a periodic feature set;

[0019] S202: Based on the periodic feature set, analyzing the growth or decline trend of the data, applying the sliding average method to highlight the key trend, using the difference method to reduce data noise, highlighting the long-term trend and seasonal influence, and obtaining the trend and seasonal feature set;

[0020] S203: Based on the trend and seasonal feature set, the prediction process of the peak time period is refined, the prediction accuracy is optimized, and the time period with active audience interaction is determined to obtain audience interaction prediction information.

[0021] On the other hand, based on the audience interaction prediction information, the random forest algorithm is used to process the real-time interaction data and predict the needs of the live broadcast personnel in the future time period. The steps for obtaining the future demand prediction data are as follows:

[0022] S301: Based on the audience interaction prediction information, current audience interaction data, including the number of viewers and comment frequency, is collected from the live broadcast platform, and the format of the data is standardized, and outliers and duplicate records are removed to obtain a standardized interaction data set;

[0023] S302: Based on the standardized interactive data set, filter the data in the key time period, analyze the user activity in the target time period, identify the peak interactive time period, and predict the personnel demand for future live broadcasts to obtain a personnel demand prediction set;

[0024] S303: Based on the personnel demand prediction set, the prediction data is compared with the historical personnel allocation, the number and type of live broadcast personnel required for multiple time periods in the future are evaluated, and the personnel allocation is optimized to obtain future demand prediction data.

[0025] On the other hand, the user activity during the target period is analyzed according to the formula:

[0026]

[0027] Calculate the average activity index A during the target period t , where V i represents the total number of viewers in the ith subinterval, C i represents the comment frequency in the ith subinterval, λ is the weight factor of the comment frequency, and k is the number of subintervals.

[0028] On the other hand, based on the future demand forecast data, cluster analysis is performed to identify the performance patterns of different live broadcast personnel, and the work schedule of the live broadcast personnel is refined according to the identification results to match the time period with the best audience activity. The specific steps for obtaining the optimal scheduling configuration are as follows:

[0029] S401: Based on the future demand forecast data, historical performance data of live broadcast personnel are classified, personnel groups with different ability levels are analyzed and identified, and working time periods are allocated to multiple groups to obtain personnel level grouping details;

[0030] S401: Based on the personnel hierarchical grouping details, the work schedule is refined, and the personnel allocation of multiple shifts is adjusted according to the personnel efficiency to match the time period with the best audience activity, and the optimal scheduling configuration is obtained by combining the workload and the personnel work preference.

[0031] On the other hand, the historical performance data of the live broadcast personnel are classified according to the formula:

[0032]

[0033] Calculate the performance index P of each live broadcaster i , where R n Represents the number of interactive replies in each live broadcast, W r is the weight of the number of interactive replies, F n Represents audience feedback rating, W f is the weight of the audience feedback score, T n Represents the live broadcast duration, W t is the weight of the live broadcast duration, and N is the number of live broadcasts.

[0034] On the other hand, based on the optimized scheduling configuration, the performance trend of live broadcast personnel in multiple time periods is predicted, and the prediction results are used to formulate a scheduling plan for the live broadcast platform to determine the personnel configuration that matches the optimal audience service and interaction. The specific steps for obtaining the live broadcast scheduling management results are as follows:

[0035] S501: Based on the shift optimization configuration, collect live broadcast data in multiple time periods, including live broadcast personnel performance and audience interaction feedback, and collect performance data in key time periods to obtain performance information in multiple time periods;

[0036] S502: Analyze the performance trend of the live broadcast personnel based on the multi-period performance information, identify time periods with high and low performance consistency, determine key performance parameters, and obtain personnel performance trend data;

[0037] S503: Based on the personnel performance trend data and combined with the peak data of audience interaction, the scheduling plan of the live broadcast personnel is optimized to match the best audience service and interaction, and obtain the live broadcast scheduling management result.

[0038] On the other hand, a scheduling management system based on live broadcast data is provided, which is applied to a scheduling management method based on live broadcast data, including:

[0039] The data integration module performs data cleaning and integration based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcast personnel, including the interaction quality score, audience feedback information and conversion rate, to obtain the live broadcast data integration set;

[0040] The trend analysis module uses the ARIMA model based on the live broadcast data integration set to analyze the periodicity and trend characteristics of the live broadcast activity, identify the peak time period of audience interaction, and obtain audience interaction prediction information;

[0041] The demand prediction module processes the real-time interaction data using a random forest algorithm based on the audience interaction prediction information, predicts the demand of the live broadcast personnel in the future time period, and obtains future demand prediction data;

[0042] The scheduling optimization module identifies the performance patterns of different live broadcast personnel based on the future demand forecast data, refines the work schedules of the live broadcast personnel according to the identification results, and predicts the performance trends of the live broadcast personnel in multiple time periods, formulates a scheduling plan for the live broadcast platform, determines the staffing that matches the optimal audience service and interaction, and obtains the live broadcast scheduling management results.

[0043] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0044] By using key data of live broadcast activities, such as the number of viewers and interaction frequency, high-frequency updates and real-time analysis can be performed to enable dynamic adjustments in scheduling to meet actual work needs. The use of the ARIMA model not only enhances forecast accuracy, but also makes scheduling decisions forward-looking and optimizes human resource allocation. Cluster analysis helps to accurately identify the performance patterns of different live broadcast personnel and optimize work schedules based on the patterns, ensuring that the most suitable personnel are deployed during periods of highest audience activity, thereby improving employee satisfaction and audience service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0046] Figure 1 It is a main step flow chart of the present invention;

[0047] Figure 2 is a flow chart of the steps of S1 of the present invention;

[0048] Figure 3 is a flow chart of the steps of S2 of the present invention;

[0049] Figure 4 is a flow chart of the steps of S3 of the present invention;

[0050] Figure 5 is a flow chart of the steps of S4 of the present invention;

[0051] Figure 6 is a flow chart of the steps of S5 of the present invention;

[0052] Figure 7 It is a system block diagram of the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0054] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0055] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0056] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0057] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0058] The embodiment of the present invention provides a scheduling management method based on live broadcast data, such as Figure 1 As shown, the following steps are included:

[0059] S1: Based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcaster, including the interaction quality score, audience feedback information and conversion rate, perform data cleaning and integration processing to obtain the live broadcast data integration set;

[0060] S2: Based on the integrated set of live broadcast data, the ARIMA model is used to analyze the periodicity and trend characteristics of live broadcast activities, identify the peak time period of audience interaction, and obtain audience interaction prediction information;

[0061] S3: Based on the audience interaction prediction information, the random forest algorithm is used to process the real-time interaction data and predict the needs of the live broadcast personnel in the future time period to obtain the future demand prediction data;

[0062] S4: Based on the future demand forecast data, cluster analysis is performed to identify the performance patterns of different live broadcast personnel, and the work schedule of the live broadcast personnel is refined according to the identification results to match the time period with the best audience activity and obtain the optimal scheduling configuration;

[0063] S5: Based on the optimized scheduling configuration, the performance trends of live broadcast personnel in multiple time periods are predicted, and the prediction results are used to formulate a scheduling plan for the live broadcast platform, determine the staffing configuration that matches the optimal audience service and interaction, and obtain the live broadcast scheduling management results.

[0064] The integrated live broadcast data set includes audience behavior characteristics, live broadcast personnel interaction index, and key performance indicators. Audience interaction prediction information includes peak time periods, activity cycles, and interaction trends. Future demand forecast data includes demand forecast cycles, demand change trends, and key demand indicators. Scheduling optimization configuration includes time period matching information, personnel deployment configuration, and efficiency optimization results. The live broadcast scheduling management results include determined schedules, key personnel configurations, and interaction efficiency optimization results.

[0065] like Figure 2 As shown in the figure, based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcast personnel, the indicators include the interaction quality score, audience feedback information and conversion rate, the data cleaning and integration processing are performed to obtain the live broadcast data integration set. The specific steps are:

[0066] S101: Based on the key data of the live broadcast activity, the number of viewers, comment frequency, and number of likes and shares are extracted, non-compliant records are excluded, and the performance indicators of the live broadcast personnel are integrated to obtain a cleaned data set;

[0067] The number of viewers, comment frequency, and number of likes and shares are extracted, and the original data is filtered according to preset rules, including excluding records with 0 viewers or comment frequency less than 1 time per minute, and data with likes and shares less than 5% of the total interaction volume. The performance indicators of live broadcasters, such as interaction quality scores, are analyzed and classified, and the scoring basis is refined into multiple sub-indicators such as content relevance, answer timeliness, and audience satisfaction scores. The average value of the indicators is calculated according to each live broadcast activity time period, and outlier detection is performed to eliminate unreasonable data. The integrated data set includes the qualified number of viewers, comment frequency, likes and shares, and sub-item scores of performance indicators.

[0068] S102: based on the cleaned data set, unify the timestamps of multiple data sources, merge similar data fields, verify the integrity and availability of the data, and obtain an integrated data set;

[0069] Unify the timestamps of different data sources to the same time base and format them in a unified time format such as "YYYY-MM-DDHH:MM:SS". Combine and count the total amount of similar data fields such as likes and shares. Verify the data integrity of each field separately, including checking for missing and duplicate values, filling missing values ​​with the mean, and verifying the logical consistency between fields. For example, the number of viewers cannot be less than the number of comments. The result is an integrated set of data including the number of viewers after integrity verification, interactive behaviors (comments, likes, shares), and performance indicators.

[0070] S103: extracting key indicators based on the integrated data set, including interaction quality score, audience feedback information and conversion rate, analyzing the indicator data, and calculating average values ​​and change trends to obtain an integrated set of live broadcast data;

[0071] The key indicators of interaction quality score, audience feedback information and conversion rate were extracted, and the interaction quality score was decomposed into two parts: interaction timeliness and response accuracy. The average interaction response time and the accuracy of answering audience questions in each live broadcast were calculated respectively. When counting the audience feedback information, the distribution of each level was summarized according to the score level, and the conversion rate of each live broadcast was calculated separately based on the conversion rate calculation formula "conversion rate = actual number of purchasing users / total number of viewers". The live broadcast data integration set was obtained by statistically analyzing the daily average value and change trend of key indicators.

[0072] like Figure 3As shown in the figure, based on the integrated set of live broadcast data, the ARIMA model is used to analyze the periodicity and trend characteristics of live broadcast activities, identify the peak time period of audience interaction, and obtain the audience interaction prediction information in the following steps:

[0073] S201: Based on the live data integration set, use the ARIMA model to identify repetitive cycles in the data, monitor the natural fluctuations of the data, and separate the key periodic patterns to obtain the periodic feature set;

[0074] The ARIMA model is used to preprocess the time series in the data set. The stationarity of the series is evaluated, and the non-stationary series is used to generate a stationary data series through data difference operation. The parameter range of the model is set according to the stationary series. The parameter range usually includes the number of autoregressive terms, the number of differences, and the number of moving average terms. Combined with the characteristics of time series data, correlation analysis is used to evaluate the dependency between data, and the specific values ​​of parameters are adjusted. Then, the optimized model is subjected to data fitting analysis to separate the periodic pattern presented in the time series, and the recurring periodic features are screened out in combination with the fluctuation monitoring results, and the periodic feature set is output.

[0075] S202: Based on the periodic feature set, analyze the growth or decline trend of the data, apply the sliding average method to highlight the key trends, use the difference method to reduce data noise, highlight the long-term trend and seasonal influence, and obtain the trend and seasonal feature set;

[0076] The data is smoothed by sliding average analysis. A time window of fixed length is selected, such as seven days as a unit. The average value within the unit time is calculated and replaced with the original value, thereby reducing the impact of short-term fluctuations on the overall trend. For the smoothed data, the rate of change analysis is further used to identify the growth or decline direction of the trend. At the same time, combined with the changes in seasonal factors, the data is divided into multiple time periods according to a certain time period, such as quarterly or monthly. The change characteristics and differences in each time period are analyzed in turn to form a trend and seasonal feature set including long-term trend information and seasonal fluctuation information.

[0077] S203: Based on the trend and seasonal feature set, refine the prediction process of the peak time period, optimize the prediction accuracy, and determine the time period when the audience interaction is active to obtain audience interaction prediction information;

[0078] First, divide the data in the time series into segments, and associate each segment with its corresponding activity characteristics. Perform regression analysis on the active behavior in the peak time period, select the optimal time segmentation parameters, divide the entire prediction process into multiple time periods for modeling, and adjust the time interval for activity calculation lag based on the activity changes in historical data. Then, gradually optimize the parameters in the time model, analyze the distribution characteristics of audience interaction in hourly data, derive the distribution of time periods when audience interaction is active every day, and generate audience interaction prediction information.

[0079] like Figure 4 As shown in FIG, based on the audience interaction prediction information, the random forest algorithm is used to process the real-time interaction data and predict the needs of the live broadcast personnel in the future time period. The specific steps for obtaining the future demand prediction data are as follows:

[0080] S301: Based on the audience interaction prediction information, current audience interaction data is collected from the live broadcast platform, including the number of viewers and comment frequency, and the data format is standardized to remove outliers and duplicate records to obtain a standardized interaction data set;

[0081] First, the collected data is unified into a standard timestamp format, the validity of each piece of data is checked, and records with zero viewers and outliers with too high a comment frequency (for example, exceeding the normal upper limit of 100 comments per minute) are eliminated. At the same time, redundant records with repeated timestamps are removed. For missing values ​​in the data, data completion is completed by filling in the historical mean. The processed data is then sorted in chronological order, merged and integrated based on the timestamp to obtain a standardized interactive data set with a unified format and no redundancy.

[0082] S302: Based on the standardized interactive data set, filter the data in the key time period, analyze the user activity in the target time period, identify the peak interactive time period, and predict the personnel demand for future live broadcasts to obtain a personnel demand prediction set;

[0083] The time period is divided into sub-intervals of hours, and the audience activity data of each sub-interval is extracted, including the cumulative values ​​of the total number of viewers and comment frequency. The average number of viewers and comment density are calculated for each time period, and the high-activity time periods are screened out by setting an activity threshold (for example, the average comment frequency is greater than 10 times per minute). Combined with the activity distribution law, the audience behavior in key time periods is analyzed using historical data. Further, based on the activity fluctuations and the audience interaction volume, the personnel demand required for the live broadcast event is predicted, including the required number of personnel and the interaction pressure index, and a personnel demand prediction set is generated.

[0084] Analyze the user activity during the target period according to the formula:

[0085]

[0086] Calculate the average activity index A during the target period t , where V i represents the total number of viewers in the ith subinterval, C i represents the comment frequency in the ith subinterval, λ is the weight factor of the comment frequency, and k is the number of subintervals;

[0087] Activity Index A t It aims to measure the interactive activity of the audience in the live broadcast, among which the number of viewers V i It directly reflects the ability to attract audiences, while the comment frequency C i It reflects the audience's participation. In order to balance the two effects, the influence of the comments is adjusted by the weight factor λ. The following data is available. In a specific period of time, it is divided into three sub-intervals. The data of each interval are as follows:

[0088] Sub-interval 1: Number of viewers V1 = 200, comment frequency C1 = 15 times / hour;

[0089] Sub-interval 2: Number of viewers V2 = 180, comment frequency C2 = 20 times / hour;

[0090] Sub-interval 3: Number of viewers V3 = 220, comment frequency C3 = 25 times / hour;

[0091] If the weight factor of comment frequency λ = 0.5, the calculation is as follows:

[0092]

[0093] The results show that during this analysis period, the average activity index was 210, which is a measure of the average level of interaction among viewers. t The value indicates that there is high user activity in a specific time period, which helps to predict which time periods require more participation of live broadcasters, thereby optimizing staffing.

[0094] S303: Based on the personnel demand forecast set, the forecast data is compared with the historical personnel allocation, the number and type of live broadcast personnel required in multiple time periods in the future are evaluated, and the personnel allocation is optimized to obtain future demand forecast data;

[0095] The predicted future demand data is compared item by item with the historical live broadcast personnel configuration records, and the difference between the personnel distribution in different time periods under the historical configuration and the actual demand is analyzed. By comparing the personnel load records of live broadcast activities, the work efficiency and interaction indicators of the live broadcast personnel in each time period are evaluated. The distribution of the number of personnel in multiple time periods is readjusted according to the predicted personnel demand, and the specific types of live broadcast personnel in different time periods are classified and labeled, such as hosting, technical or customer service personnel. The configuration is optimized according to the labeled type and the number of personnel to generate a predicted data set covering the personnel demand in multiple time periods of future live broadcast activities.

[0096] like Figure 5 As shown in the figure, based on the future demand forecast data, cluster analysis is performed to identify the performance patterns of different live broadcast personnel, and the work schedule of the live broadcast personnel is refined according to the identification results to match the time period with the best audience activity. The specific steps for obtaining the optimal scheduling configuration are as follows:

[0097] S401: Based on the future demand forecast data, the historical performance data of the live broadcast personnel is classified, and personnel groups with different ability levels are analyzed and identified, and working time periods are allocated to multiple groups to obtain the personnel level grouping details;

[0098] First, we collected historical data of each live broadcaster, including the number of interactive replies, audience feedback scores, and live broadcast duration. We then subdivided the performance indicators of each live broadcaster by time period, divided the interactive quality scores into three levels: high, medium, and low according to the score range, analyzed the emotional tendency based on the text content in the audience feedback information, labeled the feedback data into three types: positive, negative, and neutral. After classifying the performance data by level, we used the average performance data of each person in different time periods as the basis for classification. We combined the conversion rate and interaction frequency indicators to identify groups of people with similar performances, subdivided them into three categories: high performance, medium performance, and basic performance according to their ability levels, and assigned corresponding working time periods to each group to obtain the details of the hierarchical grouping of people.

[0099] Classify the historical performance data of live broadcasters according to the formula:

[0100]

[0101] Calculate the performance index P of each live broadcaster i , where R n Represents the number of interactive replies in each live broadcast, W r is the weight of the number of interactive replies, F n Represents audience feedback rating, W f is the weight of the audience feedback score, T n Represents the live broadcast duration, W t is the weight of the live broadcast duration, and N is the number of live broadcasts;

[0102] To calculate the performance index P of the live broadcaster i First, determine the weight of each indicator, the weight of the interactive reply W r =0.3, the weight of audience feedback rating W f =0.5, weight of live broadcast duration W t =0.2. The weight is determined based on the importance of the live broadcast activity and the impact on the audience, to ensure that each indicator contributes to the overall performance according to its importance. The data of a live broadcaster in the past N = 5 live broadcasts are as follows:

[0103] Interaction replies (R n ): [150, 120, 130, 160, 140];

[0104] Audience feedback score (F n ): [8.0, 8.5, 9.0, 7.5, 8.0];

[0105] Live broadcast duration - hours (T n ): [2, 1.5, 2.5, 2, 2];

[0106] Calculate the contribution score for each live broadcast:

[0107] S1=150×0.3+8.0×0.5+2×0.2=45+4+0.4=49.4;

[0108] S2=120×0.3+8.5×0.5+1.5×0.2=36+4.25+0.3=40.55;

[0109] S3=130×0.3+9.0×0.5+2.5×0.2=39+4.5+0.5=44;

[0110] S4=160×0.3+7.5×0.5+2×0.2=48+3.75+0.4=52.15;

[0111] S5=140×0.3+8.0×0.5+2×0.2=42+4+0.4=46.4;

[0112] Finally, calculate P i The average value of:

[0113]

[0114] The result shows that the average comprehensive performance index of the live broadcast staff is 46.5, which reflects their performance in multiple dimensions and is used to distinguish personnel at different ability levels, and further guide the allocation of corresponding work time periods. Based on this value, the priority of personnel level and work time allocation can be determined.

[0115] S401: Based on the details of the personnel hierarchical grouping, the work schedule is refined, and the staffing of multiple shifts is adjusted according to the staff efficiency to match the time period with the best audience activity, and the optimal shift configuration is obtained by combining the workload and the staff work preference;

[0116] Each day is divided into several fixed time periods, such as early morning, daytime, and evening peak. The audience activity in each time period is ranked, and high-performing personnel are preferentially assigned to the evening peak period with higher audience activity. Medium-performing personnel are assigned to daytime periods with medium audience activity. Basic-performing personnel are assigned to early morning and other time periods with lower audience activity. During the schedule allocation process, the total amount of audience interaction in each time period is counted and matched with the working hours and efficiency data of each person. By comparing indicators, multi-shift staffing adjustments are made, and the allocation results are corrected in combination with each person's workload data and preference information, to ultimately obtain an optimized scheduling configuration that best matches the audience activity.

[0117] like Figure 6 As shown in the figure, based on the optimized scheduling configuration, the performance trend of live broadcast personnel in multiple time periods is predicted, and the prediction results are used to formulate a scheduling plan for the live broadcast platform to determine the optimal staffing configuration that matches the audience service and interaction. The specific steps to obtain the live broadcast scheduling management results are as follows:

[0118] S501: Based on the optimized scheduling configuration, live broadcast data of multiple time periods are collected, including the performance of live broadcast personnel and audience interaction feedback, and performance data of key time periods are collected to obtain performance information of multiple time periods;

[0119] First, the live broadcast data in different time periods are split into hourly granularity. The collected performance of the live broadcast personnel includes the number of interactive replies, the number of audience praises and the problem solving rate. The audience interaction feedback data includes the sentiment analysis of the comment content, and the number of likes and shares. Then, the data is time-stamped and classified by time period. The audience interaction volume is cross-mapped with the performance indicators of the live broadcast personnel. The live broadcast performance data of key time periods are identified and sorted according to the data density of each time period, and integrated to form performance information for multiple time periods.

[0120] S502: Analyze the performance trend of the live broadcast personnel based on the multi-period performance information, identify the time periods with high and low performance consistency, determine key performance parameters, and obtain personnel performance trend data;

[0121] The performance data within each time period is normalized and aggregated using standardized indicators such as interaction efficiency, audience satisfaction score, and performance stability score. The fluctuation range of personnel performance within different time periods is calculated, and the time periods are divided into high consistency and low consistency according to the fluctuation range. At the same time, key indicators within different time periods are counted, such as the time interval with high interaction efficiency. Combined with the positive evaluation ratio involved in the audience feedback, the time periods with high and low performance consistency are finally identified, the key performance parameters of the live broadcast event are extracted, and personnel performance trend data is formed.

[0122] S503: Based on the personnel performance trend data and the peak data of audience interaction, the scheduling plan of the live broadcast personnel is optimized to match the best audience service and interaction, and the live broadcast scheduling management result is obtained;

[0123] First, we overlap and analyze the audience interaction time periods in the peak data with the personnel performance trends, screen out the live broadcast personnel most suitable for peak time configuration, group them according to the workload and interaction needs of each peak time, allocate time periods with heavy workloads to personnel with high performance and high consistency, and allocate time periods with low workloads to basic personnel groups with relatively stable performance. We adjust the live broadcast scheduling plan based on the needs of all time periods, and through continuous adjustment and comparison of the differences between actual needs and personnel allocation, we match the optimal audience service and interaction configuration to form a live broadcast scheduling management result.

[0124] like Figure 7 As shown in the figure, the scheduling management system based on live broadcast data includes:

[0125] The data integration module performs data cleaning and integration based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcast personnel, including the interaction quality score, audience feedback information and conversion rate, to obtain the live broadcast data integration set;

[0126] The trend analysis module uses the ARIMA model based on the live broadcast data integration set to analyze the periodicity and trend characteristics of live broadcast activities, identify the peak time period of audience interaction, and obtain audience interaction prediction information;

[0127] The demand forecasting module uses the random forest algorithm to process the real-time interactive data based on the audience interaction forecast information, predict the demand of the live broadcast staff in the future time period, and obtain the future demand forecast data;

[0128] The scheduling optimization module identifies the performance patterns of different live broadcast personnel based on future demand forecast data, refines the work schedules of live broadcast personnel based on the identification results, and predicts the performance trends of live broadcast personnel in multiple time periods. It formulates scheduling plans for the live broadcast platform, determines the staffing that matches the optimal audience service and interaction, and obtains the live broadcast scheduling management results.

[0129] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0130] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0131] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0132] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0134] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0137] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0138] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A scheduling management method based on live broadcast data, characterized in that: The method comprises: Based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcaster, including the interaction quality score, audience feedback information and conversion rate, data cleaning and integration processing are performed to obtain the live broadcast data integration set; Based on the live broadcast data integration set, the ARIMA model is used to analyze the periodicity and trend characteristics of the live broadcast activity, identify the peak time period of audience interaction, and obtain audience interaction prediction information; Based on the audience interaction prediction information, the random forest algorithm is used to process the real-time interaction data, and the needs of the live broadcast personnel in the future time period are predicted to obtain future demand prediction data; Based on the future demand forecast data, cluster analysis is performed to identify the performance patterns of different live broadcast personnel, and the work schedules of the live broadcast personnel are refined according to the identification results to match the time periods with the best audience activity to obtain the optimal scheduling configuration; Based on the scheduling optimization configuration, the performance trends of live broadcast personnel in multiple time periods are predicted, and the prediction results are used to formulate a scheduling plan for the live broadcast platform, determine the personnel configuration that matches the optimal audience service and interaction, and obtain the live broadcast scheduling management result.

2. The scheduling management method based on live broadcast data according to claim 1 is characterized in that: The integrated live broadcast data set includes audience behavior characteristics, live broadcast personnel interaction index, and key performance indicators; the audience interaction prediction information includes peak time periods, activity cycles, and interaction trends; the future demand forecast data includes demand forecast cycles, demand change trends, and key demand indicators; the scheduling optimization configuration includes time period matching information, personnel deployment configuration, and efficiency optimization results; the live broadcast scheduling management results include determined schedules, key personnel configurations, and interaction efficiency optimization results.

3. The scheduling management method based on live broadcast data according to claim 1 is characterized in that: Based on the key data of live broadcast activities, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcasters, including the interaction quality score, audience feedback information and conversion rate, the steps for performing data cleaning and integration processing to obtain the integrated set of live broadcast data are as follows: Based on the key data of live broadcast activities, the number of viewers, comment frequency, and number of likes and shares were extracted, non-compliant records were excluded, and the performance indicators of the live broadcasters were integrated to obtain a cleaned data set; Based on the cleaned data set, unify the timestamps of multiple data sources, merge similar data fields, verify the integrity and availability of the data, and obtain an integrated data set; Based on the integrated data set, key indicators are extracted, including interaction quality score, audience feedback information and conversion rate, the indicator data are analyzed, and the average value and change trend are statistically calculated to obtain the integrated set of live broadcast data.

4. The scheduling management method based on live broadcast data according to claim 1 is characterized in that: Based on the live broadcast data integration set, the ARIMA model is used to analyze the periodicity and trend characteristics of the live broadcast activity, identify the peak time period of audience interaction, and obtain the audience interaction prediction information in the following steps: Based on the live data integration set, the ARIMA model is used to identify repetitive cycles in the data, monitor the natural fluctuations of the data, and separate key periodic patterns to obtain a periodic feature set; Based on the periodic feature set, the growth or decline trend of the data is analyzed, the sliding average method is applied to highlight the key trends, the difference method is used to reduce data noise, highlight the long-term trend and seasonal influence, and obtain the trend and seasonal feature set; Based on the trend and seasonal feature set, the prediction process of the peak time period is refined, the prediction accuracy is optimized, and the time period with active audience interaction is determined to obtain audience interaction prediction information.

5. The scheduling management method based on live broadcast data according to claim 1 is characterized in that: Based on the audience interaction prediction information, the random forest algorithm is used to process the real-time interaction data and predict the needs of the live broadcast personnel in the future time period. The specific steps for obtaining the future demand prediction data are as follows: Based on the audience interaction prediction information, current audience interaction data is collected from the live broadcast platform, including the number of viewers and comment frequency, and the data format is standardized to remove outliers and duplicate records to obtain a standardized interaction data set; Based on the standardized interactive data set, data in a key time period is screened, user activity in the target time period is analyzed, the peak interactive time period is identified, and the personnel demand for future live broadcasts is predicted to obtain a personnel demand prediction set; Based on the personnel demand forecast set, the forecast data is compared with the historical staffing configuration, the number and type of live broadcast personnel required for multiple time periods in the future are evaluated, and the staffing configuration is optimized to obtain future demand forecast data.

6. The scheduling management method based on live broadcast data according to claim 5 is characterized in that: Analyze the user activity during the target period, according to the formula: Calculate the average activity index A during the target period t , where V i represents the total number of viewers in the ith subinterval, C i represents the comment frequency in the ith subinterval, λ is the weight factor of the comment frequency, and k is the number of subintervals.

7. The scheduling management method based on live broadcast data according to claim 1 is characterized in that: Based on the future demand forecast data, cluster analysis is performed to identify the performance patterns of different live broadcast personnel, and the work schedule of the live broadcast personnel is refined according to the identification results to match the time period with the best audience activity. The specific steps for obtaining the optimal scheduling configuration are as follows: Based on the future demand forecast data, the historical performance data of the live broadcast personnel is classified, and personnel groups with different ability levels are analyzed and identified, and working time periods are allocated to multiple groups to obtain the details of the personnel hierarchical grouping; Based on the personnel hierarchical grouping details, the work schedule is refined, and the staffing of multiple shifts is adjusted according to staff efficiency to match the time period with the best audience activity, and the optimal scheduling configuration is obtained by combining the workload and staff work preferences.

8. The scheduling management method based on live broadcast data according to claim 7 is characterized in that: The historical performance data of the live broadcast personnel are classified according to the formula: Calculate the performance index P of each live broadcaster i , where R n Represents the number of interactive replies in each live broadcast, W r is the weight of the number of interactive replies, F n Represents audience feedback rating, W f is the weight of the audience feedback score, T n Represents the live broadcast duration, W t is the weight of the live broadcast duration, and N is the number of live broadcasts.

9. The scheduling management method based on live broadcast data according to claim 1 is characterized in that: Based on the above-mentioned optimized shift configuration, the performance trend of live broadcast personnel in multiple time periods is predicted, and the prediction results are used to formulate a shift plan for the live broadcast platform to determine the personnel configuration that matches the optimal audience service and interaction. The specific steps for obtaining the live broadcast shift management results are as follows: Based on the optimized scheduling configuration, live broadcast data in multiple time periods are collected, including live broadcast personnel performance and audience interaction feedback, and performance data in key time periods are collected to obtain performance information in multiple time periods; Based on the multi-period performance information, analyzing the performance trends of the live broadcast personnel, identifying time periods with high and low performance consistency, determining key performance parameters, and obtaining personnel performance trend data; Based on the personnel performance trend data and combined with the peak data of audience interaction, the scheduling plan of the live broadcast personnel is optimized to match the optimal audience service and interaction, and the live broadcast scheduling management result is obtained.

10. A shift scheduling management system based on live broadcast data, wherein the shift scheduling management system based on live broadcast data is used to implement the shift scheduling management method based on live broadcast data as claimed in any one of claims 1 to 9, characterized in that: The system comprises: The data integration module performs data cleaning and integration based on the key data of the live broadcast activity, including the number of viewers, comment frequency, number of likes and shares, and the performance indicators of the live broadcast personnel, including the interaction quality score, audience feedback information and conversion rate, to obtain the live broadcast data integration set; The trend analysis module uses the ARIMA model based on the live broadcast data integration set to analyze the periodicity and trend characteristics of the live broadcast activity, identify the peak time period of audience interaction, and obtain audience interaction prediction information; The demand prediction module processes the real-time interaction data using a random forest algorithm based on the audience interaction prediction information, predicts the demand of the live broadcast personnel in the future time period, and obtains future demand prediction data; The scheduling optimization module identifies the performance patterns of different live broadcast personnel based on the future demand forecast data, refines the work schedules of the live broadcast personnel according to the identification results, and predicts the performance trends of the live broadcast personnel in multiple time periods, formulates a scheduling plan for the live broadcast platform, determines the staffing that matches the optimal audience service and interaction, and obtains the live broadcast scheduling management results.

Images