Multi-point matching method for OTT quality detection mode matching system

Abstract

The invention discloses a multi-point matching method for an OTT quality detection mode matching system, and relates to the technical field of OTT quality detection, and the method comprises the following specific steps: dynamic weight distribution, adaptive abnormal threshold setting, and multi-point matching detection and feedback and adjustment. According to the method, the quality condition of the OTT service can be evaluated more accurately, the quality problem can be found and processed in time, meanwhile, the method further has the capability of automatic adjustment according to different network environments, content types and user preferences, the adaptability and flexibility of the system are improved, and continuous feedback adjustment and optimization improvement are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of OTT quality detection, and in particular to a multi-point matching method for an OTT quality detection pattern matching system. Background Art

[0002] With the rapid development of Internet technology, OTT (Over The Top) services, as a video transmission and playback method based on the Internet, have gradually become an indispensable part of people's daily lives. OTT services cover multiple fields such as video on demand, live broadcast, interactive entertainment, etc., providing users with a rich variety of content choices and convenient viewing experience.

[0003] However, with the popularization of OTT services and the continuous improvement of user needs, the quality requirements for OTT services are also getting higher and higher. Traditional OTT quality detection methods mostly rely on a single indicator or a fixed threshold for evaluation. This method often cannot fully and accurately reflect the overall quality of OTT services. For example, only focusing on a single indicator such as video freeze frequency or loading time may ignore other equally important quality factors, such as audio quality, picture clarity, etc. In addition, due to the diversity and complexity of OTT service application scenarios and user groups, different network environments, content types and user preferences will affect the quality of OTT services. Therefore, traditional fixed threshold detection methods are difficult to adapt to such diverse scenario requirements. In actual applications, false alarms or missed alarms often occur, causing the accuracy and reliability of quality detection results to be questioned.

[0004] In order to solve the above problems, the industry has begun to explore more comprehensive, accurate and adaptable OTT quality detection methods. Among them, the multi-point matching method, as an emerging detection technology, can more accurately evaluate the quality of OTT services and adapt to different scenario requirements by comprehensively considering multiple indicators and dynamically adjusting thresholds. However, the current multi-point matching method still has some shortcomings, such as unreasonable weight distribution and inaccurate abnormal threshold setting, which need further improvement and optimization. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the prior art and provide a multi-point matching method for an OTT quality detection pattern matching system. It can comprehensively consider multiple indicators and dynamically adjust weights and thresholds. The present invention can more accurately evaluate the quality status of OTT services, and promptly discover and handle quality problems. At the same time, the method also has the ability to automatically adjust according to different network environments, content types and user preferences, thereby improving the adaptability and flexibility of the system. Through continuous feedback adjustment and optimization and improvement, the present invention can provide users with a more stable and smooth OTT viewing experience, and improve the performance and reliability of the system.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a multi-point matching method for an OTT quality detection pattern matching system, the method comprising the following specific steps:

[0007] Dynamic weight allocation: Dynamic weights are allocated to each indicator based on its impact on OTT quality, determining the importance of different indicators in the overall quality assessment.

[0008] Adaptive abnormal threshold setting: The weights of each indicator obtained by dynamic weight allocation are standardized, and the indicators are divided into three importance levels: high, medium, and low. The formula for determining the importance level is: Among them, L i represents the importance level of the index, W i is the weight of the indicator. By analyzing different scenario factors, a corresponding indicator abnormal threshold adjustment model is established for each scenario factor. The calculation formula is: in represents the abnormal threshold of the ith indicator after adjustment in the jth scenario, T ij is the threshold before adjustment, β ij is the adjustment coefficient of the ith indicator in the jth scenario, C j It is a scenario characteristic factor. It takes scenario factors as input and outputs the abnormal threshold adjustment coefficient of each indicator. The initial abnormal threshold of each indicator is adjusted according to the adjustment coefficient. When an indicator is determined to be of high importance, a more sophisticated threshold division method is used to monitor the high-weight indicator in real time.

[0009] Multi-point matching detection: During the OTT quality detection process, multiple indicators are tested at the same time, and multi-point matching is performed based on the results of dynamic weight allocation and adaptive abnormal threshold setting to determine the quality status of the OTT service;

[0010] Feedback and adjustment: Based on the results of multi-point matching detection, feedback adjustments are made to the dynamic weight allocation and adaptive anomaly threshold setting to continuously optimize the accuracy and efficiency of OTT quality detection.

[0011] Furthermore, in the dynamic weight allocation step, a set of OTT quality-related indicators is first determined, including video freeze frequency, video loading time, and audio quality. By deploying a data acquisition module on the client and server of the OTT platform, the data of each indicator is captured in real time and stored and transmitted. The data acquisition module adopts a distributed architecture and sets a data caching mechanism. At the same time, an initial weight determination formula is used to determine the initial weight of each indicator before adjustment. The calculation formula is: Among them, W i represents the initial weight of the i-th indicator, N is the number of historical data samples, M is the total number of indicators, and I ij represents the value of the i-th indicator in the j-th sample, S j represents the user satisfaction score corresponding to the jth sample.

[0012] Furthermore, in the dynamic weight allocation step, the collected historical data is statistically analyzed, the mean, variance and standard deviation of each indicator are calculated, and an OTT quality assessment model is established. The indicator data is used as input features, and the user viewing experience satisfaction is used as the output label. The model is trained and the weight is adjusted using a dynamic adjustment formula, which is: in, represents the adjusted weight of the i-th indicator, W i is the weight before adjustment, α is the adjustment coefficient, F i It is a feedback factor. After training, the feature weights in the model are the initial weights of each indicator in the overall quality assessment.

[0013] Furthermore, in the step of dynamic weight allocation, a user feedback processing system is established to receive user feedback information and classify and encode it, and the feedback ratio related to each indicator is determined according to the type and content of the feedback information to adjust the indicator weight. The feedback factor calculation formula is: F i =γ×R i +(1-γ)×H i , where F i is the feedback factor of the ith indicator, γ is the weight coefficient, R i is the feedback value determined based on user feedback, H i It is a feedback value determined based on historical data trends.

[0014] Furthermore, in the step of setting the adaptive abnormal threshold, a large amount of OTT quality-related data is statistically analyzed, including user feedback data and historical monitoring data. By analyzing the relationship between different indicators and user satisfaction, it is found that when the weight of a certain indicator exceeds 0.7, the change of the indicator has a very significant impact on user satisfaction. When the indicator weight is between 0.3-0.7, it has a certain impact on user satisfaction, but it is relatively weak and is an indicator of medium importance. When the weight is less than 0.3, the impact on user satisfaction is small and can be regarded as a low-importance indicator. Therefore, it is determined

[0015] Furthermore, in the adaptive abnormal threshold setting step, when an indicator is determined to be at a high importance level, a more sophisticated threshold division method is used to monitor the high-weight indicator in real time. For video freeze frequency, which is determined to be a high-importance indicator, the high-weight indicator monitoring module divides the freeze time into multiple intervals. For audio quality, which is determined to be a high-importance indicator, the high-weight indicator monitoring module divides the intervals according to different parameters of the audio, and sets different abnormal thresholds for different intervals. For audio quality, which is determined to be a high-importance indicator, the high-weight indicator monitoring module divides the intervals according to different parameters of the audio. When entering a specific interval, the system can detect abnormal situations more quickly and take corresponding processing measures.

[0016] Furthermore, in the multi-point matching detection step, a distributed monitoring system is designed, and monitoring tools are deployed on different nodes to monitor different indicators. Multiple monitoring points are set for each indicator to ensure data accuracy.

[0017] Furthermore, in the multi-point matching detection step, the weighted average method is used to comprehensively analyze the detection data of multiple indicators, and the weight value of each indicator is determined according to the dynamic weight allocation result. The detection data of each indicator is compared with the corresponding abnormal threshold to obtain the abnormality degree, and the abnormality degree of each indicator is weighted averaged according to the weight to obtain a comprehensive OTT quality evaluation value. The formula is: Among them, Q represents the comprehensive OTT quality evaluation value, M i is the total number of indicators, W i is the weight of the ith indicator, D i is the abnormality degree of the ith indicator.

[0018] Furthermore, in the feedback and adjustment step, when the detection results do not match the actual user experience, an adjustment optimization mechanism is initiated, including manual adjustment and automatic adjustment. Manual adjustment is performed by the system administrator to adjust the indicator weights and abnormal thresholds based on the analysis results, and automatic adjustment uses an optimization algorithm to automatically adjust the indicator weights and abnormal thresholds based on historical data and real-time monitoring data.

[0019] Compared with the prior art, the multi-point matching method for the OTT quality detection pattern matching system has the following beneficial effects:

[0020] 1. The present invention avoids the limitation of focusing on a single indicator by comprehensively considering multiple indicators that affect the quality of OTT. At the same time, the dynamic weight allocation mechanism can be flexibly adjusted according to the importance of different indicators, ensuring that each indicator is given a reasonable weight in the overall quality assessment. In addition, the adaptive anomaly threshold setting can be automatically adjusted according to scenario requirements such as network environment, content type and user preference, thereby improving the sensitivity and accuracy of anomaly detection. Under the joint effect, the comprehensiveness and accuracy of OTT quality detection are significantly improved, providing users with a more stable and smooth viewing experience.

[0021] 2. By introducing a feedback and adjustment mechanism, the present invention can continuously optimize and improve the dynamic weight allocation and adaptive abnormal threshold setting according to the results of multi-point matching detection and user feedback, so that the system can continuously adapt to changes in the network environment, updates of content types and changes in user preferences, thereby improving the adaptability and flexibility of the system. At the same time, by continuously collecting real-time monitoring data and user feedback, the system can promptly discover potential quality problems and handle them, thereby reducing the probability of quality problems. This not only improves the performance and reliability of the system, but also provides users with a more high-quality and personalized OTT service experience.

[0022] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 It is a flow chart of a multi-point matching method for an OTT quality detection pattern matching system;

[0025] Figure 2 The present invention is a flow chart of a multi-point matching method for an OTT quality detection pattern matching system. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. 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.

[0027] Example 1

[0028] This embodiment describes in detail an application of a multi-point matching method for an OTT quality detection pattern matching system in the field of online video playback platforms. By introducing the present invention, a more high-quality and stable viewing experience is provided to users.

[0029] First, we need to identify key indicators such as video freeze frequency, video loading time, picture clarity, and audio quality, and carefully deploy data collection modules on the server and client sides of the platform. This module uses advanced technical means to capture the data of each indicator in real time and accurately. The data collection module adopts a highly reliable distributed architecture to ensure stable and efficient operation in various complex network environments and on different types of devices. At the same time, a powerful data caching mechanism is set up to prevent data loss and ensure data integrity and continuity even when the network is unstable. By establishing a precise OTT quality evaluation model, we use the data of each indicator as rich input features and user viewing experience satisfaction as the key output label to conduct in-depth and detailed model training. During the training process, by continuously adjusting the model parameters, the model can predict the user viewing experience satisfaction with a high degree of accuracy. After careful training, the feature weights in the model are the initial weights of each indicator in the overall quality evaluation. For example, let the video freeze frequency be indicator i=1, the video loading time be indicator i=2, the picture clarity be indicator i=3, and the audio quality be indicator i=4. Through the formula Determine the initial weight of each indicator, where N is the number of historical data samples, M is the total number of indicators, and I ij represents the value of the i-th indicator in the j-th sample, S j represents the user satisfaction score corresponding to the jth sample. At the same time, the system can widely receive feedback information submitted by users through various channels and use advanced classification and coding techniques to carefully process the feedback information. According to the type and content of the feedback information, the feedback ratio related to each indicator can be accurately determined. For example, if the video freeze problem accounts for a high proportion in user feedback, the weight of the video freeze frequency indicator is significantly increased, and the adjustment formula is: in, represents the adjusted weight of the i-th indicator, W i is the weight before adjustment, α is the adjustment coefficient, F iis the feedback factor, which can be determined based on user feedback, historical data trends, and other factors. For example, when there are many issues involving the i-th indicator in user feedback, F i Increase, so that the weight of this indicator increases rapidly to better adapt to user needs and actual conditions.

[0030] In the adaptive abnormal threshold setting stage, the weights of each indicator obtained by dynamic weight allocation are strictly standardized so that their value range is accurately between 0 and 1. According to the standardized weight values, the indicators are divided into three importance levels: high, medium, and low. For example, the weight value greater than 0.7 can be set as a high importance level, between 0.3 and 0.7 as a medium importance level, and less than 0.3 as a low importance level. Assuming that the video freeze frequency and picture clarity are determined as high importance level indicators, the video loading time is a medium importance level indicator, and the audio quality is a low importance level indicator, such a division can clearly reflect the importance of each indicator in the overall quality evaluation. The importance of network environment in the assessment is analyzed in depth, including network environment (such as 4G, 5G, Wi-Fi, etc.). For each scenario factor, a corresponding indicator abnormal threshold adjustment model is established. The model uses scenario factors as key inputs and outputs the abnormal threshold adjustment coefficients of each indicator through complex algorithms. For example, for network environment factors, different adjustment models can be established according to different network types. In the model, the network type information is input, and the abnormal threshold adjustment coefficients of each indicator are calculated through a pre-set precise algorithm. Then, the initial abnormal threshold of each indicator is accurately adjusted according to the adjustment coefficient. The adjustment formula combined with scenario requirements is as follows: in represents the abnormal threshold of the ith indicator after adjustment in the jth scenario, T ij is the threshold before adjustment, β ij is the adjustment coefficient of the ith indicator in the jth scenario, C j It is the scene characteristic factor. This formula can flexibly adjust the abnormal threshold according to different scenarios to improve the accuracy and adaptability of quality detection. For high-importance video freeze frequency, the high-weight indicator monitoring module uses advanced technology to divide the freeze time into multiple fine intervals, and sets different abnormal thresholds for different intervals. When the freeze time enters a specific interval, the system can rely on the efficient detection mechanism to detect the abnormal situation more quickly and take corresponding processing measures quickly. For example, for a slight freeze interval, mild monitoring and early warning can be carried out. For a severe freeze interval, emergency processing measures are immediately initiated to ensure that the user's viewing experience is not seriously affected.

[0031] In the multi-point matching detection stage, an advanced distributed monitoring system is designed, and professional monitoring tools are deployed at different key nodes to conduct comprehensive and real-time monitoring of different indicators. Each monitoring tool can use advanced technology to collect data of the corresponding indicator in real time, and quickly send the data to the central server for efficient processing. For each indicator, multiple carefully arranged monitoring points are set to ensure the accuracy and comprehensiveness of the data. For example, for the video freeze frequency indicator, monitoring points can be set at important locations such as the client, server, and network intermediate nodes to comprehensively monitor the freeze situation from multiple angles. The weighted average method is used to conduct in-depth and comprehensive analysis of the detection data of multiple indicators. First, according to the results of dynamic weight allocation, the weight value of each indicator is accurately determined. Then, the detection data of each indicator is strictly compared with the corresponding abnormal threshold to obtain the abnormal degree of each indicator. Finally, the abnormal degree of each indicator is accurately weighted averaged according to the weight to obtain a comprehensive OTT quality evaluation value. The comprehensive quality evaluation formula is: Where Q represents the comprehensive OTT quality evaluation value, M is the total number of indicators, and W i is the weight of the ith indicator, D i is the abnormality degree of the ith indicator. This formula can comprehensively consider the weights and abnormality degrees of multiple indicators, providing a scientific and accurate basis for the overall quality assessment of the platform.

[0032] In the feedback and adjustment stage, a powerful detection result analysis module is established. This module conducts a comprehensive comparative analysis of the results obtained from multi-point matching detection with the actual user experience. By widely collecting user feedback information and behavioral data, advanced analysis technology is used to evaluate the accuracy of the system detection results. For example, the degree of consistency between the system detection results and the actual user experience can be judged by analyzing the user's viewing time, exit rate and other behavioral data, as well as the user's feedback and evaluation. When the detection results do not match the actual user experience, the adjustment and optimization mechanism is quickly initiated. This mechanism includes manual adjustment and automatic adjustment. Manual adjustment: experienced system administrators make fine manual adjustments to indicator weights and abnormal thresholds based on the analysis results. During the adjustment process, the system provides real-time Data analysis and visualization help administrators make more accurate and informed decisions, and automatic adjustment: using optimization algorithms, automatically adjust indicator weights and abnormal thresholds based on historical data and real-time monitoring data. The optimization algorithm takes improving the accuracy of OTT quality detection as its core goal. Through continuous iterative optimization, the optimal combination of weights and thresholds is found, so that the system can better adapt to different situations and user needs. At the same time, a complete feedback loop mechanism is established to continuously collect user feedback and real-time monitoring data. These rich data are used as input to regularly conduct comprehensive evaluations and in-depth improvements on multi-point matching methods. For example, the system can be evaluated once a week or month, and system parameters can be adjusted and optimized in a timely manner based on the evaluation results to continuously improve system performance and user experience.

[0033] In summary, this embodiment provides an efficient, accurate and intelligent OTT quality detection solution for an online video playback platform, thereby improving user experience and platform competitiveness.

[0034] Example 2

[0035] This embodiment describes in detail an application of a multi-point matching method for an OTT quality detection pattern matching system in the field of live broadcast platforms. By introducing the present invention, the high quality and smoothness of the live broadcast are ensured, bringing users an excellent viewing experience.

[0036] First, determine key indicators such as live broadcast delay, picture clarity, sound synchronization, and live broadcast stability. Using advanced data acquisition technology, deploy data acquisition modules on the host and audience ends respectively to collect various indicator data during the live broadcast process in real time. For example, monitor the transmission of live broadcast signals on the host end, including video encoding parameters, audio quality, etc., and record the live broadcast delay time, subjective evaluation of picture clarity, and perception of sound synchronization on the audience end. The data acquisition module adopts an efficient communication protocol to ensure real-time transmission and accuracy of data. At the same time, set up a data storage mechanism for subsequent analysis and processing.

[0037] In the dynamic weight allocation stage, considering the real-time characteristics of live broadcast, live broadcast delay and live broadcast stability are given higher weights in the initial stage. By analyzing the historical live broadcast data and combining the user feedback on different indicators, the initial weight of each indicator is determined. For example, the statistical analysis method is used to calculate the impact of live broadcast delay, picture clarity, sound synchronization and live broadcast stability on user satisfaction in historical live broadcasts. The initial weight is determined according to the impact. Assuming that the initial weight of live broadcast delay is W1, picture clarity is W2, sound synchronization is W3, and live broadcast stability is W4, the formula is used. Determine the initial weight of each indicator, where N is the number of historical data samples, M is the total number of indicators, and I ij represents the value of the i-th indicator in the j-th sample, S j It represents the user satisfaction score corresponding to the jth sample. With the changes of different live broadcast contents (such as sports live broadcast, music live broadcast, game live broadcast, etc.) and user needs, the weights of various indicators are dynamically adjusted. For example, in sports live broadcast, picture clarity and live broadcast delay are more important, and the weights of these two indicators are increased. In music live broadcast, the weight of sound synchronization is increased. A real-time monitoring system is established to monitor factors such as user feedback, network conditions, and live broadcast content characteristics during the live broadcast process. When a specific situation is monitored, the indicator weight is automatically adjusted according to the preset rules. For example, when network congestion causes an increase in live broadcast delay, the weight of live broadcast stability is appropriately reduced, and the weights of live broadcast delay and picture clarity are increased to give priority to ensuring the smoothness and image quality of the live broadcast. The adjustment formula is: in, represents the adjusted weight of the i-th indicator, W i is the weight before adjustment, α is the adjustment coefficient, F i It is a feedback factor. The feedback factor can be determined based on factors such as user feedback, network conditions, and live content characteristics. For example, when users feedback that the live delay is too high, F1 increases, thereby increasing the weight of the live delay indicator.

[0038] In the adaptive abnormal threshold setting stage, the weights of each indicator obtained by dynamic weight allocation are standardized so that their value range is between 0 and 1. According to the standardized weight values, the indicators are divided into three importance levels: high, medium and low. For example, a weight value greater than 0.7 can be set as a high importance level, between 0.3 and 0.7 as a medium importance level, and less than 0.3 as a low importance level. Assuming that in a specific case, live broadcast delay is determined as a high importance level indicator, picture clarity is a medium importance level indicator, and sound synchronization and live broadcast stability are low importance level indicators, different scenario factors are analyzed, including network conditions (such as network congestion, network stability, etc.), live broadcast content types (such as sports live broadcast, music live broadcast, game live broadcast, etc.), etc. For each scenario factor, a corresponding indicator abnormal threshold adjustment model is established. The model takes the scenario factor as input and outputs the abnormal threshold adjustment coefficient of each indicator. For example, for the scenario of network congestion, the abnormal threshold of live broadcast stability is appropriately relaxed, and the abnormal threshold requirements of picture clarity and sound synchronization are reduced to adapt to changes in network conditions. The adjustment formula combined with scenario requirements is: in represents the abnormal threshold of the ith indicator after adjustment in the jth scenario, T ij is the threshold before adjustment, β ij is the adjustment coefficient of the ith indicator in the jth scenario, C j is the scene characteristic factor, which can be determined based on factors such as network conditions and live content type. For example, in a scenario with network congestion, C j The value can be adjusted according to the degree of network congestion. For high-importance live broadcast delay indicators, the high-weight indicator monitoring module divides the live broadcast delay into different urgency intervals. For example, the live broadcast delay is divided into slight delay (0-1 second), medium delay (1-3 seconds), severe delay (3-5 seconds) and extremely severe delay (more than 5 seconds). Different abnormal thresholds and processing measures are set for different intervals. When the live broadcast delay enters the severe delay interval (3-5 seconds), the system issues an early warning and tries to optimize the network connection or adjust the live broadcast parameters. When the live broadcast delay enters the extremely severe delay interval (more than 5 seconds), the system immediately takes emergency measures such as switching the live broadcast source or reducing the video bit rate to ensure the smoothness of the live broadcast.

[0039] In the multi-point matching detection stage, a distributed monitoring system is used to deploy monitoring tools on different nodes such as the host end, the audience end and the server end to monitor different indicators in real time. Each monitoring tool can accurately collect data of the corresponding indicator and send the data to the central server for processing. For each indicator, multiple monitoring points are set to ensure the accuracy and comprehensiveness of the data. For example, for the live broadcast delay indicator, monitoring points can be set on the host end, the audience end and the server end respectively to monitor the sending time, receiving time and processing time of the live broadcast signal, so as to calculate the live broadcast delay. The weighted average method is used to comprehensively analyze the detection data of multiple indicators. First, according to the result of dynamic weight allocation, the weight value of each indicator is determined. Then, the detection data of each indicator is compared with the corresponding abnormal threshold to obtain the abnormal degree of each indicator. Finally, the abnormal degree of each indicator is weighted averaged according to the weight to obtain a comprehensive OTT quality evaluation value. The comprehensive quality evaluation formula is Q=∑ i＝1 W i ×D i , where Q represents the comprehensive OTT quality evaluation value, M i is the total number of indicators, W i is the weight of the ith indicator, D i is the abnormal degree of the i-th indicator. Through comprehensive analysis, we can fully understand the quality of the live broadcast, discover problems in time and take corresponding measures.

[0040] In the feedback and adjustment stage, a detection result analysis module is established to compare and analyze the results obtained from multi-point matching detection with the actual user experience. By collecting user feedback information, viewing behavior data, etc., the accuracy of the system detection results is evaluated. For example, the user's complaint content, viewing time, exit rate and other data are analyzed to determine whether the problems detected by the system are consistent with the user's actual experience. When the detection results are inconsistent with the actual user experience, the adjustment and optimization mechanism is activated. This mechanism includes manual adjustment and automatic adjustment. Manual adjustment: The system administrator manually adjusts the indicator weight and abnormal threshold according to the analysis results. During the adjustment process, the system provides real-time data analysis. Analysis and visualization help administrators make more accurate decisions and automatic adjustments: Utilize optimization algorithms to automatically adjust indicator weights and abnormal thresholds based on historical data and real-time monitoring data. The optimization algorithm aims to improve the accuracy of OTT quality detection. Through continuous iterative optimization, the optimal combination of weights and thresholds is found. A feedback loop mechanism is established to continuously collect user feedback and real-time monitoring data. These data are used as input to regularly evaluate and improve the multi-point matching method. For example, a comprehensive evaluation of the system is conducted every week or month, and system parameters are adjusted based on the evaluation results. The algorithm model is optimized to continuously improve the OTT quality detection and service quality of the live broadcast platform.

[0041] In summary, this embodiment can effectively implement OTT quality detection and optimization on the live broadcast platform, providing users with high-quality and smooth live broadcast viewing experience.

[0042] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A multi-point matching method for an OTT quality detection pattern matching system, characterized in that: The method comprises the following specific steps: Dynamic weight allocation: Dynamic weights are allocated to each indicator based on its impact on OTT quality, determining the importance of different indicators in the overall quality assessment. Adaptive abnormal threshold setting: The weights of each indicator obtained by dynamic weight allocation are standardized, and the indicators are divided into three importance levels: high, medium, and low. The formula for determining the importance level is: Among them, L i represents the importance level of the index, W i is the weight of the indicator. By analyzing different scenario factors, a corresponding indicator abnormal threshold adjustment model is established for each scenario factor. The calculation formula is: in represents the abnormal threshold of the ith indicator after adjustment in the jth scenario, T ij is the threshold before adjustment, β ij is the adjustment coefficient of the ith indicator in the jth scenario, C j It is a scenario characteristic factor. It takes scenario factors as input and outputs the abnormal threshold adjustment coefficient of each indicator. The initial abnormal threshold of each indicator is adjusted according to the adjustment coefficient. When an indicator is determined to be of high importance, a more sophisticated threshold division method is used to monitor the high-weight indicator in real time. Multi-point matching detection: During the OTT quality detection process, multiple indicators are tested at the same time, and multi-point matching is performed based on the results of dynamic weight allocation and adaptive abnormal threshold setting to determine the quality status of the OTT service; Feedback and adjustment: Based on the results of multi-point matching detection, feedback adjustments are made to the dynamic weight allocation and adaptive anomaly threshold setting to continuously optimize the accuracy and efficiency of OTT quality detection.

2. According to claim 1, a multi-point matching method for an OTT quality detection pattern matching system is characterized in that: In the dynamic weight allocation step, a set of OTT quality-related indicators is first determined, including video freeze frequency, video loading time, and audio quality. Data acquisition modules are deployed on the client and server of the OTT platform to capture, store, and transmit data of each indicator in real time. The data acquisition module adopts a distributed architecture and sets a data cache mechanism. At the same time, an initial weight determination formula is used to determine the initial weight of each indicator before adjustment. The calculation formula is: Among them, W i represents the initial weight of the i-th indicator, N is the number of historical data samples, M is the total number of indicators, and I ij represents the value of the i-th indicator in the j-th sample, S j represents the user satisfaction score corresponding to the jth sample.

3. The multi-point matching method for an OTT quality detection pattern matching system according to claim 1, characterized in that: In the dynamic weight allocation step, the collected historical data is statistically analyzed to calculate the mean, variance and standard deviation of each indicator, and an OTT quality assessment model is established. The indicator data is used as input features, and the user viewing experience satisfaction is used as the output label. The model is trained and the weight is adjusted using a dynamic adjustment formula, which is: in, represents the adjusted weight of the i-th indicator, W i is the weight before adjustment, α is the adjustment coefficient, F i It is a feedback factor. After training, the feature weights in the model are the initial weights of each indicator in the overall quality assessment.

4. The multi-point matching method for an OTT quality detection pattern matching system according to claim 3, characterized in that: In the dynamic weight allocation step, a user feedback processing system is established to receive user feedback information and classify and encode it. The feedback ratio related to each indicator is determined according to the type and content of the feedback information to adjust the indicator weight. The feedback factor calculation formula is: F i =γ×R i +(1-γ)×H i , where F i is the feedback factor of the ith indicator, γ is the weight coefficient, R i is the feedback value determined based on user feedback, H i It is a feedback value determined based on historical data trends.

5. The multi-point matching method for an OTT quality detection pattern matching system according to claim 1, characterized in that: In the step of setting the adaptive abnormal threshold, a large amount of OTT quality-related data is statistically analyzed, including user feedback data and historical monitoring data. By analyzing the relationship between different indicators and user satisfaction, it is found that when the weight of a certain indicator exceeds 0.7, the change of the indicator has a very significant impact on user satisfaction. When the indicator weight is between 0.3-0.7, it has a certain impact on user satisfaction, but it is relatively weak and belongs to a medium-important indicator. When the weight is less than 0.3, the impact on user satisfaction is small and can be regarded as a low-importance indicator. Therefore, it is determined 6. The multi-point matching method for an OTT quality detection pattern matching system according to claim 1, characterized in that: In the adaptive abnormal threshold setting step, when an indicator is determined to be at a high importance level, a more sophisticated threshold division method is used to monitor the high-weight indicator in real time. For video freeze frequency, which is determined to be a high-importance indicator, the high-weight indicator monitoring module divides the freeze time into multiple intervals. For audio quality, which is determined to be a high-importance indicator, the high-weight indicator monitoring module divides the intervals according to different parameters of the audio, and sets different abnormal thresholds for different intervals. For audio quality, which is determined to be a high-importance indicator, the high-weight indicator monitoring module divides the intervals according to different parameters of the audio. When entering a specific interval, the system can detect abnormal situations more quickly and take corresponding processing measures.

7. The multi-point matching method for an OTT quality detection pattern matching system according to claim 1, characterized in that: In the multi-point matching detection step, a distributed monitoring system is designed, and monitoring tools are deployed on different nodes to monitor different indicators. Multiple monitoring points are set for each indicator to ensure data accuracy.

8. A multi-point matching method for an OTT quality detection pattern matching system according to claim 7, characterized in that In the multi-point matching detection step, the weighted average method is used to comprehensively analyze the detection data of multiple indicators, and the weight value of each indicator is determined according to the dynamic weight allocation result. The detection data of each indicator is compared with the corresponding abnormal threshold to obtain the abnormal degree, and the abnormal degree of each indicator is weighted averaged according to the weight to obtain a comprehensive OTT quality evaluation value. The formula is: Among them, Q represents the comprehensive OTT quality evaluation value, M i is the total number of indicators, W i is the weight of the ith indicator, D i is the abnormality degree of the ith indicator.

9. A multi-point matching method for an OTT quality detection pattern matching system according to claim 1, characterized in that ,In the feedback and adjustment steps, when the detection results do not match the actual ,user experience, the adjustment and optimization mechanism is initiated, including manual ,adjustment and automatic adjustment. The system administrator adjusts the ,index weight and abnormal threshold based on the analysis results. The ,automatic adjustment uses the optimization algorithm to automatically adjust the ,index weight and abnormal threshold based on historical data and ,real-time monitoring data.

Images