A method, device, equipment and storage medium for detecting a playing abnormal factor

Abstract

The application provides a method and device for detecting abnormal factors of playing, equipment and storage medium, which can be applied to the field of Internet of Vehicles or the field of intelligent transportation, and is used to solve the problem of low detection accuracy and detection reliability of abnormal factors causing playing abnormality. The method comprises the following steps: obtaining target probability prediction model, and predicting target abnormal probability of each multimedia file in a target cycle based on obtained target occurrence rate; obtaining target probability prediction model, and determining target root cause probability of each abnormal factor corresponding to each abnormal playing amount in each multimedia file in the target cycle based on the target occurrence rate and the target abnormal probability. The target root cause probability is the conditional probability of the abnormal factor causing the abnormal playing amount under the premise of the abnormal playing amount in each playing amount in the target cycle, which improves the detection accuracy and detection reliability of the abnormal factors causing playing abnormality.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, device, and storage medium for detecting abnormal playback factors. Background Technology

[0002] With the continuous development of technology, more and more devices can play multimedia files. For example, connected car devices can play songs or videos.

[0003] During the playback of multimedia files, various abnormal factors may cause playback anomalies such as a sudden increase or decrease in the playback count of certain multimedia files. For example, when a car networking device is playing a song, a network failure may cause the playback to be interrupted. The target device may then attempt to replay the song multiple times, causing the playback count to suddenly increase.

[0004] If playback abnormalities occur and the factors causing them are not detected and repaired in a timely manner, the playback abnormalities will continue to occur, affecting the stability of the device when playing multimedia files.

[0005] In related technologies, one method for detecting abnormal factors that cause playback abnormalities is to manually count the abnormal factors that occur within a time period during which the playback abnormality occurs, and then take all the abnormal factors that occur as the abnormal factors that cause the playback abnormality.

[0006] However, within a given time period when playback anomalies occur, various abnormal factors may arise, such as push notifications for promotional activities, device system malfunctions, new version releases, network anomalies, or misoperations by the target object. Not every abnormal factor is the cause of playback anomalies. If all abnormal factors occurring within that time period are considered as the cause of playback anomalies, the accuracy of anomaly detection will be low, resulting in staff having to fix a large number of unnecessary anomalies, generating unnecessary workload and costs.

[0007] It is evident that, under the relevant technologies, the detection accuracy and reliability of abnormal factors causing playback anomalies are relatively low. Summary of the Invention

[0008] This application provides a method, apparatus, computer device, and storage medium for detecting abnormal playback factors, which addresses the problem of low detection accuracy and reliability of abnormal factors causing playback abnormalities.

[0009] Firstly, a method for detecting abnormal playback factors is provided, including:

[0010] Obtain the target occurrence rate of each abnormal factor when playing each multimedia file within the target period;

[0011] Using a target probability prediction model, based on the obtained occurrence rate of each target, the target anomaly probability of each multimedia file is predicted when playing each multimedia file within the target period, wherein the target anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file;

[0012] Using a target probability prediction model, based on the occurrence rate of each target and the target anomaly probability, the target root cause probability corresponding to each abnormal factor is determined when there is an abnormal playback volume in the playback volume of each multimedia file within the target period. The target root cause probability represents the probability that the abnormal factor causes the abnormal playback volume.

[0013] Based on the obtained root cause probabilities of each target, the abnormal factors are sorted to generate the detection results of abnormal playback factors.

[0014] Secondly, a device for detecting abnormal playback factors is provided, comprising:

[0015] Acquisition module: used to acquire the target occurrence rate of each abnormal factor when playing each multimedia file within the target period;

[0016] Processing module: Used to use a target probability prediction model to predict the target anomaly probability of each multimedia file when playing each multimedia file within the target period, based on the obtained occurrence rate of each target, wherein the target anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file;

[0017] The processing module is further configured to: adopt a target probability prediction model, based on the occurrence rate of each target and the target anomaly probability, determine the target root cause probability corresponding to each abnormal factor when there is an abnormal playback volume in the playback volume of each multimedia file within the target period, wherein each target root cause probability represents the probability that the abnormal factor causes the abnormal playback volume.

[0018] The processing module is further configured to: sort the abnormal factors based on the obtained root cause probabilities of each target, and generate detection results of abnormal factors in playback.

[0019] Optionally, the target probability prediction model includes multiple target model parameters, and different target model parameters are trained for different anomaly factors; the processing module is specifically used for:

[0020] For each of the aforementioned abnormal factors, perform the following operations respectively:

[0021] The probability of an abnormal factor is determined by multiplying the target occurrence rate of the abnormal factor with the target model parameters corresponding to the abnormal factor. The probability of an abnormal factor represents the probability that there is an abnormal playback volume caused by the abnormal factor in the playback volume of the multimedia file.

[0022] The ratio between the probability of the factor and the probability of the target anomaly is taken as the target root cause probability of the abnormal factor when the abnormal playback volume exists in the playback volume of each multimedia file within the target period.

[0023] Optionally, the processing module is further configured to:

[0024] Obtain sample data corresponding to each of the multiple historical periods, wherein the sample data includes the historical anomaly probability of each multimedia file in the historical period, and the historical occurrence rate of each anomaly factor in the corresponding historical period, and the historical anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of each multimedia file in the historical period.

[0025] Multiple sample data points from each of the obtained sample data points are used as training data to establish a training dataset, and at least one sample data point from each of the sample data points is used as training data to establish a test dataset.

[0026] Based on the training dataset and the test dataset, the probability prediction model is trained iteratively in multiple rounds to output the trained target probability prediction model.

[0027] Optionally, the processing module is specifically used for:

[0028] In each round of training iterations, perform the following operations:

[0029] Each historical occurrence rate in the training data is used as the input to the probability prediction model, and the historical anomaly probability in the training data is used as the output of the probability prediction model to determine the training model parameters of the probability prediction model corresponding to each anomaly factor.

[0030] Using the probability prediction model that includes the parameters of each of the obtained training models, the test anomaly probability of each multimedia file is predicted based on the historical occurrence rate in the test data, wherein the test anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file within the historical period.

[0031] The probability prediction model is trained based on the error between the obtained test anomaly probability and the historical anomaly probability in the test data.

[0032] Optionally, the processing module is specifically used for:

[0033] For each of the aforementioned historical periods, perform the following operations respectively:

[0034] The playback count of each multimedia file and the historical occurrence rate of each abnormal factor are obtained within the historical period.

[0035] Based on the playback volume of each multimedia file, the historical anomaly probability of each multimedia file within the historical period is determined.

[0036] The obtained historical occurrence rates and historical anomaly probabilities are used as sample data corresponding to the historical period.

[0037] Optionally, the processing module is specifically used for:

[0038] Statistical analysis of the playback volume distribution information for each of the aforementioned multimedia files;

[0039] Based on statistical distribution information, the confidence interval of the playback volume of each multimedia file is determined, wherein the confidence interval of the playback volume represents the range of normal playback volume.

[0040] The historical anomaly probability of each multimedia file within the historical period is determined by the ratio of the weighted sum of abnormal playback counts that are not within the confidence interval of the playback count in each multimedia file to the total number of files in each multimedia file.

[0041] Optionally, the processing module is specifically used for:

[0042] Obtain the number of times each multimedia file has been played within the historical period;

[0043] The sum of the playback counts for each file is taken as the total playback count.

[0044] The ratio between the number of times each file is played and the total number of times it is played is taken as the playback count of the corresponding multimedia file.

[0045] Optionally, the distribution information includes the average number of plays and the standard deviation of plays for each multimedia file; the processing module is specifically used for:

[0046] The minimum value is the difference between the product of the preset weight and the standard deviation of the playback volume and the average playback volume.

[0047] The maximum value is the sum of the product of the preset weight and the standard deviation of the playback volume, and the average playback volume.

[0048] The confidence interval for the number of plays is determined based on the range between the obtained minimum and maximum values.

[0049] Thirdly, a computer program product is provided, including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0050] Fourthly, a computer device is provided, comprising:

[0051] Memory, used to store program instructions;

[0052] A processor is configured to invoke program instructions stored in the memory and execute the method described in the first aspect according to the obtained program instructions.

[0053] Fifthly, a computer-readable storage medium is provided, the computer-readable storage medium storing computer-executable instructions for causing a computer to perform the method as described in the first aspect.

[0054] In this embodiment, the detection of playback anomalies is based on the target occurrence rate of each anomaly when playing each multimedia file within the target period, rather than directly and roughly taking the occurrence rate of each target as the detection result of playback anomalies.

[0055] After obtaining the occurrence rate of each target, a target probability prediction model is used. Based on the occurrence rate of each target, the target anomaly probability of abnormal playback volume in the playback volume of each multimedia file can be predicted. Continuing with the target probability prediction model, based on the occurrence rate and target anomaly probability of each target, the target root cause probability of each anomalous factor causing abnormal playback volume can be determined when abnormal playback volume exists in the playback volume of each multimedia file within the target period. In other words, each target root cause probability is the conditional probability that each anomalous factor causes that abnormal playback volume given that abnormal playback volume exists in the playback volume of each multimedia file within the target period. Determining the detection result of playback anomaly factors through conditional probability, compared to simply using the occurrence rate of each anomalous factor as the detection result, provides a more accurate representation of the probability that each anomalous factor causes abnormal playback volume, thereby improving the accuracy and reliability of detecting anomalous factors causing playback anomalies.

[0056] Furthermore, the detection results of abnormal playback factors are obtained by ranking each abnormal factor based on the probability of each target root cause. The higher the probability of the target root cause, the greater the possibility that the corresponding abnormal factor will cause abnormal playback volume. Therefore, ranking can effectively assist subsequent analysis and repair tasks. For example, by analyzing and repairing each abnormal factor in the order of its occurrence, the abnormal playback volume can be prevented from happening again in a timely manner, thus ensuring playback stability. Attached Figure Description

[0057] Figure 1A This is a schematic diagram illustrating the application field of the method for detecting abnormal playback factors provided in the embodiments of this application;

[0058] Figure 1B This is one application scenario of the method for detecting abnormal playback factors provided in the embodiments of this application;

[0059] Figure 2 A flowchart illustrating a method for detecting abnormal playback factors provided in this application embodiment;

[0060] Figure 3 A schematic diagram illustrating the principle of a method for detecting abnormal playback factors provided in this application embodiment;

[0061] Figure 4 A schematic diagram of the principle of a method for detecting abnormal playback factors provided in this application embodiment. Figure 2 ;

[0062] Figure 5 A flowchart illustrating a method for detecting abnormal playback factors provided in this application embodiment. Figure 2 ;

[0063] Figure 6 A flowchart illustrating a method for detecting abnormal playback factors provided in this application embodiment. Figure 3 ;

[0064] Figure 7 A schematic diagram of the principle of a method for detecting abnormal playback factors provided in this application embodiment. Figure 3 ;

[0065] Figure 8 A schematic diagram of a device for detecting abnormal playback factors provided in an embodiment of this application;

[0066] Figure 9 A schematic diagram of the structure of the device for detecting abnormal playback factors provided in the embodiments of this application. Figure 2 . Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0068] The following explanations of some terms used in the embodiments of this application are provided to facilitate understanding by those skilled in the art.

[0069] (1) Conditional probability:

[0070] Conditional probability refers to the probability of event A occurring given that another event B has already occurred. Conditional probability can be expressed as P(A|B), read as "the probability of A given B". Conditional probability can be calculated using decision trees.

[0071] (2) Prior probability:

[0072] Prior probability refers to the probability obtained based on past experience and analysis, such as the law of total probability. It is often used as the probability of the "cause" appearing in the "cause-effect" problem.

[0073] (3) Confidence interval:

[0074] A confidence interval is an interval bounded by the upper and lower confidence limits of a statistic. For a given set of sample data with mean μ and standard deviation σ, the 100(1-α)% confidence interval for the overall data mean is (μ-σ). α / 2 σ, μ+Z α / 2 σ), where α is the area covered by the non-confidence level within the normal distribution, Z α / 2 This is the corresponding standard score.

[0075] This application relates to the fields of Artificial Intelligence (AI) and cloud computing, and can be applied to fields such as smart transportation, smart agriculture, smart healthcare, or mapping.

[0076] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that studies the design principles and implementation methods of various machines, attempting to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence, enabling machines to have perception, reasoning, and decision-making functions.

[0077] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, interactive operating systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, machine learning / deep learning, autonomous driving, and intelligent transportation. With the development and progress of AI, it has been researched and applied in numerous fields, such as smart homes, intelligent customer service, virtual assistants, smart speakers, intelligent marketing, wearable devices, autonomous driving, drones, robots, smart healthcare, vehicle networking, and intelligent transportation. It is believed that with further technological advancements, AI will be applied in even more fields, playing an increasingly important role. The solutions provided in this application's embodiments relate to deep learning and augmented reality technologies in AI.

[0078] Cloud computing refers to the delivery and usage model of IT infrastructure, meaning obtaining necessary resources in an on-demand and easily scalable manner through a network. In a broader sense, cloud computing refers to the delivery and usage model of services, meaning obtaining necessary services in an on-demand and easily scalable manner through a network. These services can be IT and software related, internet-related, or other services. Cloud computing is a product of the development and integration of traditional computer and network technologies such as grid computing, distributed computing, parallel computing, utility computing, network storage technologies, virtualization, and load balancing.

[0079] With the development of the internet, real-time data streams, and the diversification of connected devices, as well as the demands for search services, social networks, mobile commerce, and open collaboration, cloud computing has rapidly developed. Unlike previous parallel distributed computing, cloud computing will fundamentally revolutionize the entire internet model and enterprise management model.

[0080] Intelligent Vehicle Infrastructure Cooperative Systems (IVICS) are a development direction of Intelligent Transportation Systems (ITS). IVICS utilizes advanced wireless communication and next-generation Internet technologies to implement comprehensive, real-time dynamic information exchange between vehicles and infrastructure. Based on the collection and fusion of dynamic traffic information across all times and spaces, it conducts active vehicle safety control and cooperative road management, fully realizing effective collaboration between people, vehicles, and roads. This ensures traffic safety, improves traffic efficiency, and ultimately forms a safe, efficient, and environmentally friendly road traffic system.

[0081] It should be noted that the embodiments of this application involve data such as appearance rate or play count. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0082] The application areas of the method for detecting abnormal playback factors provided in the embodiments of this application will be briefly introduced below.

[0083] With the continuous development of technology, more and more devices can play multimedia files. For example, please refer to... Figure 1A Different applications can be used in vehicle-to-everything (V2X) devices to play songs or videos.

[0084] During the playback of multimedia files, various abnormal factors may cause playback anomalies such as a sudden increase or decrease in the playback count of certain multimedia files. For example, when a car networking device is playing a song, a network failure may cause the playback to be interrupted. The target device may then attempt to replay the song multiple times, causing the playback count to suddenly increase.

[0085] For example, when a vehicle-to-everything (V2X) device is playing a video, the target vehicle may try to replay the video multiple times due to parking, reversing, or other reasons, causing the video's playback count to suddenly increase.

[0086] For example, when a connected car device is playing an audiobook, a system malfunction may cause multiple audiobooks to fail to display properly. As a result, the target audience will not play these audiobooks that cannot be displayed properly, causing a sudden drop in the number of plays for these audiobooks.

[0087] If abnormal playback counts occur and the underlying causes are not promptly detected and addressed, the abnormal playback counts will continue to occur, affecting the stability of the device's playback of multimedia files.

[0088] In related technologies, one method for detecting abnormal factors that cause playback abnormalities is to statistically analyze the abnormal factors that occur within a time period during which the playback abnormality occurs, and then manually check each abnormal factor one by one, or directly take all the abnormal factors that occur as the abnormal factors that cause the playback abnormality.

[0089] However, within a given timeframe where playback anomalies occur, various factors may arise, such as promotional campaigns, device system malfunctions, new version releases, network issues, or misoperation by the target audience. Not every factor is responsible for abnormal playback numbers. Treating all anomalies occurring within that timeframe as the cause of playback anomalies would lower the accuracy of anomaly detection, requiring staff to fix numerous unnecessary anomalies, resulting in unnecessary workload and costs. Manually checking each anomaly individually would also incur significant manpower costs and be extremely time-consuming.

[0090] It is evident that, under the relevant technologies, the detection accuracy and reliability of abnormal factors causing playback anomalies are relatively low.

[0091] To address the issue of low detection accuracy and reliability in trained object detection models, this application proposes a method for detecting playback anomalies. This method involves obtaining the target occurrence rate of each anomaly factor when playing each multimedia file within a target period. Then, a target probability prediction model is used to predict the target anomaly probability of each multimedia file within the target period, based on the obtained target occurrence rates. Here, the target anomaly probability represents the probability that anomalies exist in the playback count of each multimedia file. Continuing with the target probability prediction model, based on the target occurrence rates and target anomaly probabilities, the target root cause probability of each anomaly factor is determined when anomalies exist in the playback count of each multimedia file within the target period. Here, each target root cause probability represents the probability that the corresponding anomaly factor causes abnormal playback counts. Based on the obtained target root cause probabilities, the anomalies are ranked to generate the detection results for playback anomalies.

[0092] In this embodiment, the detection of playback anomalies is based on the target occurrence rate of each anomaly when playing each multimedia file within the target period, rather than directly and roughly taking the occurrence rate of each target as the detection result of playback anomalies.

[0093] After obtaining the occurrence rate of each target, a target probability prediction model is used. Based on the occurrence rate of each target, the target anomaly probability of abnormal playback volume in the playback volume of each multimedia file can be predicted. Continuing with the target probability prediction model, based on the occurrence rate and target anomaly probability of each target, the target root cause probability of each anomalous factor causing abnormal playback volume can be determined when abnormal playback volume exists in the playback volume of each multimedia file within the target period. In other words, each target root cause probability is the conditional probability that each anomalous factor causes that abnormal playback volume given that abnormal playback volume exists in the playback volume of each multimedia file within the target period. Determining the detection result of playback anomaly factors through conditional probability, compared to simply using the occurrence rate of each anomalous factor as the detection result, provides a more accurate representation of the probability that each anomalous factor causes abnormal playback volume, thereby improving the accuracy and reliability of detecting anomalous factors causing playback anomalies.

[0094] Furthermore, the detection results of abnormal playback factors are obtained by ranking each abnormal factor based on the probability of each target root cause. The higher the probability of the target root cause, the greater the possibility that the corresponding abnormal factor will cause abnormal playback volume. Therefore, ranking can effectively assist subsequent analysis and repair tasks. For example, by analyzing and repairing each abnormal factor in the order of its occurrence, the abnormal playback volume can be prevented from happening again in a timely manner, thus ensuring playback stability.

[0095] The following describes the application scenarios of the method for detecting abnormal playback factors provided in this application.

[0096] Please refer to Figure 1B This diagram illustrates an application scenario of the method for detecting abnormal playback factors provided in this application. The application scenario includes a client 101 and a server 102. The client 101 and the server 102 can communicate with each other. The communication method can be wired, such as through a network cable or serial cable; or wireless, such as through Bluetooth or Wi-Fi. No specific limitation is imposed.

[0097] Client 101 generally refers to devices capable of playing multimedia files, such as terminal devices, third-party applications accessible by terminal devices, or web pages accessible by terminal devices. Terminal devices include, but are not limited to, mobile phones, computers, smart medical devices, smart home appliances, in-vehicle terminals, or aircraft. Server 102 generally refers to devices capable of detecting abnormal playback factors, such as terminal devices or servers. Servers include, but are not limited to, cloud servers, local servers, or associated third-party servers. Both client 101 and server 102 can utilize cloud computing to reduce the consumption of local computing resources; similarly, they can also utilize cloud storage to reduce the consumption of local storage resources.

[0098] As one embodiment, the client 101 and the server 102 can be the same device, and there is no specific limitation. In this embodiment, the client 101 and the server 102 are described as different devices.

[0099] The following is based on Figure 1B Using server 102 as the server and the server as the main body, this application provides a detailed description of the method for detecting abnormal playback factors provided in its embodiments. Please refer to... Figure 2 This is a flowchart illustrating a method for detecting abnormal playback factors provided in an embodiment of this application.

[0100] S201, obtain the target occurrence rate of each abnormal factor when playing each multimedia file within the target period.

[0101] The target period can be a duration with the current time as the end time and a preset duration as the period. The preset duration can be in units of seconds, minutes, hours, or days, etc., with no specific restrictions. For example, when the multimedia file is a song, a target period of 5 minutes with the current time as the end time can be used; or when the multimedia file is a video, a target period of 2 hours with the current time as the end time can be used, and so on.

[0102] During the target period, several abnormal factors may occur simultaneously when playing various multimedia files, such as push notifications for promotional activities, system malfunctions of devices or applications, release of new versions of devices or applications, network anomalies, or accidental operations by the target object. Therefore, it is possible to obtain the target occurrence rate of each abnormal factor when playing various multimedia files within the target period, and use the target occurrence rate as a basis to detect abnormal factors during playback.

[0103] One method to determine the target occurrence rate for each anomalous factor is to first determine the target occurrence count for each anomalous factor within the target period. Then, based on the sum of the target occurrence counts, determine the total occurrence count. The ratio of the target occurrence count for each anomalous factor to the total occurrence count is then used as the target occurrence rate for that anomalous factor.

[0104] Let the target period be denoted as period t+1, then the number of times each abnormal factor's target occurs is denoted as {y}. (t+1)i |i=1,……,m}, where i represents the i-th anomalous factor, m represents the total number of anomalous factors, and y (t+1)i The number of times an abnormal factor occurs within the target period.

[0105] So, the total number of occurrences y t+1 Please refer to formula (1).

[0106]

[0107] After obtaining the occurrence count of each target and the total occurrence count, the target occurrence rate p(c) of each anomaly factor can be determined. (t+1)i Please refer to formula (2).

[0108]

[0109] S202 uses a target probability prediction model to predict the target anomaly probability of each multimedia file when playing each multimedia file within the target period, based on the obtained occurrence rate of each target.

[0110] The target anomaly probability represents the probability that there are abnormal playback counts in the playback counts of each multimedia file.

[0111] Since the playback count of each multimedia file during the target period may be either abnormal or normal, it is impossible to accurately determine whether the playback count of a multimedia file is abnormal or normal simply by looking at the playback count of each file individually. Therefore, this application provides a target probability prediction model. This model can predict the probability of abnormal playback counts in each multimedia file during the target period based on the target occurrence rate of each abnormal factor. Please refer to... Figure 3 Each set of anomalies includes m anomalies. Using a target probability prediction model, based on the target occurrence rate of each of the m anomalies, the model can predict the target probability of abnormal playback counts for each multimedia file within a target period. The higher the target occurrence rate of each anomaly, the greater the probability of abnormal playback counts for each multimedia file; conversely, the lower the target occurrence rate of each anomaly, the lower the probability of abnormal playback counts for each multimedia file. Therefore, the target probability prediction model can more accurately determine the probability of abnormal playback counts for each multimedia file within a target period.

[0112] Continuing with the example of a target period of t+1, the target anomaly probability of each multimedia file is... Please refer to formula (3).

[0113]

[0114] Where i represents the i-th anomalous factor, m represents the total number of anomalous factors, and p(c (t+1)i ) represents the target occurrence rate of each i-th abnormal factor. The target model parameters characterize the target prediction model.

[0115] As one example, the target prediction model can have multiple target model parameters, meaning the target prediction model contains multiple target model parameters, and different target model parameters are trained for different anomalies. So, what is the target anomaly probability of each multimedia file? Please refer to formula (4).

[0116]

[0117] S203 uses a target probability prediction model to determine the target root cause probability of each abnormal factor when there is an abnormal number of plays in the play count of each multimedia file within the target period, based on the occurrence rate and abnormal probability of each target.

[0118] Each target root cause probability represents the probability that the corresponding abnormal factor will cause abnormal playback volume.

[0119] Since each multimedia file may have abnormal playback counts or all normal playback counts, the root cause probability of each abnormal factor when abnormal playback counts exist within the target period is equivalent to the conditional probability of each abnormal factor causing that abnormal playback count, assuming the event of abnormal playback counts occurs within the target period. By using conditional probabilities, the likelihood of each abnormal factor causing abnormal playback counts can be compared under the same benchmark, thus accurately identifying the most significant abnormal factor when abnormal playback counts occur, aiding in the analysis and remediation of such situations.

[0120] The target probability prediction model involved in the embodiments of this application can determine the target root cause probability corresponding to each abnormal factor based on the occurrence rate of each target and the target anomaly probability.

[0121] As an example, if the target probability prediction model contains multiple target model parameters, and different target model parameters are trained for different anomalous factors, then the method of using the target probability prediction model to determine the target root cause probability corresponding to each anomalous factor based on the occurrence rate and the target anomalous probability of each target can be: based on the target model parameters contained in the target probability prediction model, and the operation between the occurrence rate and the target anomalous probability of each target, the target root cause probability corresponding to each anomalous factor can be determined separately.

[0122] The following example demonstrates the calculation process for the root cause probability of an anomalous factor. The calculation process for the root cause probability of other anomalous factors is similar and will not be repeated here.

[0123] Please refer to Figure 4 The probability of an anomaly is determined by multiplying the target occurrence rate of the anomaly with the target model parameters corresponding to the anomaly. The factor probability represents the probability that an anomaly caused by the anomaly exists in the playback volume of each multimedia file. The ratio between the factor probability and the target anomaly probability is used as the target root cause probability corresponding to the anomaly when an anomaly exists in the playback volume of each multimedia file within the target period.

[0124] The root cause probability P[c] corresponding to the i-th anomalous factor (t+1)i |A t+1 Please refer to formula (5).

[0125]

[0126] in, p(c) represents the target model parameters corresponding to the i-th anomaly. (t+1)i ) represents the target occurrence rate of the i-th abnormal factor. c represents the probability of target anomaly. (t+1)i The value of is in the range [1, m].

[0127] S204. Based on the obtained root cause probabilities of each target, sort the abnormal factors to generate the detection results of abnormal factors in playback.

[0128] After obtaining the probabilities of each target root cause, the abnormal factors can be sorted based on these probabilities. For example, they can be sorted in descending order of target root cause probability; or in ascending order of target root cause probability; or, for abnormal factors with target root cause probabilities greater than a preset threshold, they can be sorted in descending order of target root cause probability, and so on. There are no specific restrictions. This allows analysts to prioritize and investigate each abnormal factor, thus promptly identifying the cause of abnormal playback and effectively assisting analysts in finding the root cause.

[0129] As one embodiment, the above-mentioned target probability prediction model can be trained based on the following method. The process of training the probability prediction model to obtain the target probability prediction model is described below. Please refer to [link / reference]. Figure 5 .

[0130] S501, obtain sample data corresponding to each of the multiple historical periods.

[0131] Each sample data includes the historical anomaly probability of each multimedia file within the corresponding historical period, as well as the historical occurrence rate of each anomaly factor within the corresponding historical period. Each historical anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of each multimedia file within the corresponding historical period.

[0132] Continuing with the example of the target period being the (t+1)th period, the historical period can be represented as the tkth period, where k = 0, ..., n. Therefore, multiple historical periods contain a total of n+1 periods. The probability of historical anomalies can be expressed as {P(A... t-k The historical occurrence rate of the i-th anomalous factor can be expressed as {p(c) | k = 0, ..., n}. (t-k)i )|i=1,...,m;k=0,...,n}.

[0133] The sample dataset S, composed of all the sample data, can be represented in matrix form, as shown in formula (6).

[0134] S = {P(A t-k ),],p(c (t-k)1 ), ..., p(c (t-k)m (6) |k=0,……,n}

[0135] As one example, the process of obtaining sample data within a historical period will be described below. The process of obtaining sample data within other historical periods is similar and will not be repeated here.

[0136] Obtain the playback count of each multimedia file and the historical occurrence rate of each anomaly factor within the historical period. Based on the playback count of each multimedia file, determine the historical anomaly probability of each multimedia file within the historical period. Use the obtained historical occurrence rates and historical anomaly probabilities as sample data corresponding to the historical period.

[0137] As one embodiment, the playback count of a multimedia file can be the number of times the multimedia file has been played within a historical period, or the playback rate of the multimedia file within a historical period, etc., without any specific limitation. Taking the playback count of a multimedia file as the playback rate of the multimedia file within a historical period as an example, after obtaining the number of times each multimedia file has been played within the historical period, the sum of the number of times each file has been played can be used as the total number of plays. The ratio between the number of times each file has been played and the total number of plays is used as the playback count r of the corresponding multimedia file. (t-k)h Please refer to formula (7).

[0138]

[0139] Where, x (t-k)h x represents the number of times the h-th multimedia file is played within the tk-th period. t-k Represents the total number of playbacks within the tk-th period. (h = 1, ..., M) t-k ), (k = 0, ..., n), M t-k It represents the total number of multimedia files in the tk-th period.

[0140] As one embodiment, when determining the historical anomaly probability of each multimedia file within a historical period based on its individual playback count, the distribution information of the playback count of each multimedia file can be statistically analyzed first. Based on the statistical distribution information, a confidence interval for the playback count of each multimedia file is determined, where the confidence interval represents the range of normal playback counts. The historical anomaly probability of each multimedia file within a historical period is determined by the ratio of the weighted sum of the abnormal playback counts that are not within the confidence intervals to the total number of files in each multimedia file.

[0141] As one embodiment, the distribution information includes the average number of plays and the standard deviation of plays for each multimedia file. The average number of plays can represent the average number of plays within a historical period; it can also represent the average number of plays for each multimedia file across all historical periods. The following example uses the average number of plays to represent the average number of plays for each multimedia file across all historical periods. Please refer to formula (8).

[0142]

[0143] Furthermore, the standard deviation of play counts can be calculated based on the average play counts, please refer to formula (9).

[0144]

[0145] The minimum value is the difference between the product of a preset weight and the standard deviation of the playback volume, and the average playback volume. The maximum value is the sum of the product of the preset weight and the standard deviation of the playback volume, and the average playback volume. Based on the range between the obtained minimum and maximum values, a confidence interval for the playback volume of each multimedia file is determined. For example, the preset weight is... α is the area covered by the non-confidence level within the normal distribution. That is, the corresponding standard score.

[0146] For the confidence interval of the play count, please refer to formula (10).

[0147]

[0148] Therefore, based on formula (10) of the confidence interval of play count, the historical anomaly probability within the tk-th historical period is... It can be expressed using formula (11).

[0149]

[0150] S502, use multiple sample data from each of the obtained sample data as training data to establish a training dataset, and use at least one sample data from each of the sample data as training data to establish a test dataset.

[0151] After obtaining the sample data, multiple samples from each sample can be used as training data to create a training dataset, and at least one sample from each sample can be used as training data to create a test dataset. The sample data can be divided into training and test datasets proportionally or randomly. For example, 'a' percent of the sample data can be used as training data, and '(1-a)' percent as test data; according to general experience, 'a' can be a value such as 8.

[0152] S503 performs multiple rounds of iterative training on the probability prediction model based on the training and test datasets, and outputs the trained target probability prediction model.

[0153] After obtaining the training and test datasets, the probabilistic prediction model can be trained iteratively in multiple rounds based on the training and test datasets to output the trained target probabilistic prediction model.

[0154] For the probability prediction model, please refer to formula (12).

[0155]

[0156] Where, ε t-k Let represent the residual sequence, which characterizes the conditional probability that there is an abnormal playback volume in the playback volume of each multimedia file under the premise that the i-th abnormal factor occurs.

[0157] As an example, the following description uses one round of iterative training as an example. The process of each round of iterative training is similar and will not be repeated here.

[0158] Each historical occurrence rate in the training data is used as the input to the probability prediction model, and the historical anomaly probability in the training data is used as the output of the probability prediction model to determine the training model parameters for each anomaly factor. Please refer to formula (12) for further details. The training model parameters for the probability prediction model corresponding to the i-th anomaly factor can be expressed as follows:

[0159] A probabilistic prediction model incorporating the parameters of each trained model is used to predict the test anomaly probability of each multimedia file based on the historical occurrence rates in the test data. The test anomaly probability represents the probability that abnormal playback volume exists in the playback volume of each multimedia file within the corresponding historical period. Please refer to formula (13).

[0160]

[0161] The probabilistic prediction model is trained based on the error between the obtained test anomaly probability and the historical anomaly probability in the test data. This error can be the mean squared error (MSE).

[0162] When the error reaches its minimum, the probability prediction model containing the parameters of each of the current training models will be used as the target probability prediction model; otherwise, the next round of iterative training will begin.

[0163] The following example illustrates the method for detecting abnormal playback factors provided in this application. Please refer to the example of playing a song. Figure 6 .

[0164] S601, through data collection, obtains the play count of each song and the occurrence rate of each abnormal factor within multiple historical periods.

[0165] Collect the number of times each song is played within the tn-th historical period, the (t-n+1)-th historical period, ..., and the t-th historical period, {x (t-k)h |h=1,……,M t-k ;k = 0, ..., n}, h represents the h-th song.

[0166] Collect the occurrence count of each abnormal factor in the tn-th historical period, the (t-n+1)-th historical period, ..., and the t-th historical period, {y (t-k)i |i=1,……,m;k=0,……,n}, where i represents the i-th abnormal factor.

[0167] For each historical period, the total number of plays for each song file is calculated by summing the number of plays for each file. For each historical period, the total number of occurrences of each abnormal factor is calculated by summing the occurrences.

[0168] For each historical period, the ratio of the number of times the h-th song file was played to the total number of plays is taken as the play count of the h-th song. (t-k)h =x (t-k)h / x t-k For each historical period, the ratio between the occurrence frequency of the i-th anomalous factor and the total occurrence frequency is taken as the occurrence rate of the i-th anomalous factor, p(c (t-k)i )=y (t-k)i / y t-k .

[0169] S602 determines the historical anomaly probability of each multimedia file within each historical period by constructing prior probabilities.

[0170] Based on the play count of each song within each historical period, determine the average play count of each song across all historical periods. Further, based on the average number of plays for each song, the standard deviation of the play count is determined.

[0171] Based on the determined average play count and standard deviation of play count, a confidence interval for the play count of each song is determined.

[0172] Based on the confidence interval of each song's play count, determine the prior probability of abnormal play counts for each song within the tk-th historical period, i.e., the historical anomaly probability.

[0173] For example, please refer to Figure 7Three songs were played in each of the three historical periods. In the first historical period, the first song had 0.1 plays, the second song had 0.5 plays, and the third song had 0.4 plays; in the second historical period, the first song had 0.4 plays, the second song had 0.4 plays, and the third song had 0.2 plays; in the third historical period, the first song had 0.4 plays, the second song had 0.1 plays, and the third song had 0.5 plays.

[0174] The average number of plays for the first song is 0.3, for the second song it is 0.33, and for the third song it is 0.36.

[0175] The standard deviation of the play count for the first song is 0.016, for the second song it is 0.029, and for the third song it is 0.016.

[0176] Therefore, the confidence interval for the play count of the first song is (0.3-0.016z). α / 2 0.3 + 0.016z α / 2 The confidence interval for the play count of the second song is (0.33-0.029z). α / 2 , 0.33+0.029z α / 2 The confidence interval for the play count of the third song is (0.36-0.016z). α / 2 0.36 + 0.016z α / 2 ).

[0177] Then the historical anomaly probability corresponding to the first historical period is: The historical anomaly probability corresponding to the second historical cycle is The historical anomaly probability corresponding to the third historical cycle is

[0178] S603, Construct a probability prediction model. Please refer to formula (12) for the probability prediction model.

[0179] S604, based on the historical anomaly probability of each song within each historical period, and the historical occurrence rate of each anomaly factor within each historical period, construct each sample data, [P(A t-k ), p(c (t-k)1 ), ..., p(c (t-k)m )).

[0180] Eighty percent of each sample data is used as training data, and the remaining twenty percent is used as test data.

[0181] S605 trains a probability prediction model based on various training and test data.

[0182] By feeding training data into a probabilistic prediction model and using the gradient descent algorithm, the training parameters of the probabilistic prediction model corresponding to each anomalous factor are determined. in,

[0183] The historical occurrence rates of each outlier in a test dataset are then fed into a probability prediction model containing the parameters of the trained model described above. Obtain the test anomaly probability output by the probability prediction model.

[0184] Based on the squared error between the test anomaly probability and the historical anomaly probability in the test data, determine whether the squared error has reached the minimum value. If it has reached the minimum value, then the probability prediction model containing the above training model parameters will be used as the target probability prediction model; if it has not reached the minimum value, then proceed to the next round of training.

[0185] After obtaining the target probability prediction model, multiple trained target model parameters can be obtained.

[0186] S606 uses a target probability prediction model to predict the target anomaly probability of each song within a target period.

[0187] Obtain the number of times each song file is played within the target period, i.e., the (t+1)th period, as well as the occurrence frequency of each anomaly. Based on the number of times each song file is played, determine the total play count for each song. Based on the occurrence frequency of each anomaly, determine the occurrence rate of each anomaly.

[0188] The occurrence rate of each anomalous factor is used as input to the target probability prediction model to predict the target anomalous probability of each song within the target period.

[0189] S607, using the target probability prediction model, determine the target root cause probability of each abnormal factor when there is an abnormal number of plays in the play count of each song within the target period. Please refer to formula (5).

[0190] S608, based on the obtained root cause probabilities of each target, sorts each abnormal factor in descending order to generate the detection results of abnormal factors in playback.

[0191] In this embodiment of the application, a target probability prediction model is used to calculate and analyze various abnormal factors that lead to abnormal playback volume. By calculating the conditional probability value of each abnormal factor under the condition of abnormal playback volume, the abnormal factors that lead to abnormal playback volume are found, which effectively assists in manual investigation.

[0192] By inputting the play counts of each song within multiple historical periods, a confidence interval for the play count of each song is constructed. This determines whether there are abnormal play counts for each song. If the play count is not within the confidence interval, the play count is considered abnormal; otherwise, it is considered normal.

[0193] The target probability prediction model is built and trained by inputting the play count of each song and the occurrence rate of each abnormal factor in multiple historical periods, so as to obtain the conditional probability of abnormal play count under the condition that each abnormal factor occurs.

[0194] By sorting the abnormal factors from largest to smallest based on the root cause probability of each target, analysts can quickly identify the causes of abnormal playback volume in sequence.

[0195] The embodiments of this application can be applied to analysis solutions for abnormal situations in all system environments and business scenarios.

[0196] Based on the same inventive concept, embodiments of this application provide a device for detecting abnormal playback factors, capable of achieving the functions corresponding to the aforementioned method for detecting abnormal playback factors. Please refer to... Figure 8 The device includes an acquisition module 801 and a processing module 802, wherein:

[0197] Acquisition module 801: used to acquire the target occurrence rate of each abnormal factor when playing each multimedia file within the target period;

[0198] Processing module 802: Used to use a target probability prediction model to predict the target anomaly probability of each multimedia file when playing each multimedia file within the target period, based on the obtained occurrence rate of each target. The target anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file.

[0199] The processing module 802 is also used to: adopt a target probability prediction model, based on the occurrence rate of each target and the target anomaly probability, determine the target root cause probability corresponding to each abnormal factor when there is an abnormal playback volume in the playback volume of each multimedia file within the target period, wherein the target root cause probability represents the probability that the abnormal factor causes the abnormal playback volume.

[0200] The processing module 802 is also used to: sort the abnormal factors based on the obtained root cause probabilities of each target, and generate the detection results of abnormal factors in playback.

[0201] In one possible embodiment, the target probability prediction model includes multiple target model parameters, which are trained for different anomaly factors; the processing module 802 is specifically used for:

[0202] For each abnormal factor, perform the following operations respectively:

[0203] The probability of an abnormal factor is determined by multiplying the target occurrence rate of the abnormal factor with the target model parameters corresponding to the abnormal factor. The probability of an abnormal factor represents the probability that there is an abnormal playback volume caused by an abnormal factor in the playback volume of the multimedia file.

[0204] The ratio between the factor probability and the target anomaly probability is used as the target root cause probability when there are abnormal playback volumes in the playback volumes of each multimedia file within the target period.

[0205] In one possible embodiment, the processing module 802 is further configured to:

[0206] Obtain sample data corresponding to multiple historical periods. The sample data includes the historical anomaly probability of each multimedia file in the historical period, and the historical occurrence rate of each anomaly factor in the corresponding historical period. The historical anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of multimedia files in the historical period.

[0207] Multiple samples from each of the obtained sample data are used as training data to establish a training dataset, and at least one sample from each of the sample data is used as training data to establish a test dataset.

[0208] Based on the training and test datasets, the probability prediction model is trained iteratively in multiple rounds to output the trained target probability prediction model.

[0209] In one possible embodiment, the processing module 802 is specifically used for:

[0210] In each round of training iterations, perform the following operations:

[0211] The historical occurrence rates in the training data are used as inputs to the probability prediction model, and the historical anomaly probabilities in the training data are used as outputs to determine the training model parameters of the probability prediction model corresponding to each anomaly factor.

[0212] Using a probabilistic prediction model that includes the parameters of each trained model, the probability of test anomalies for each multimedia file is predicted based on the historical occurrence rates in the test data. The probability of test anomalies represents the probability that there are abnormal playback volumes in the playback volume of multimedia files within the historical period.

[0213] The probability prediction model is trained based on the error between the obtained test anomaly probability and the historical anomaly probability in the test data.

[0214] In one possible embodiment, the processing module 802 is specifically used for:

[0215] Perform the following operations for each of the multiple historical periods:

[0216] Get the playback count of each multimedia file and the historical occurrence rate of each abnormal factor within the historical period;

[0217] Based on the playback volume of each multimedia file, the historical anomaly probability of each multimedia file within the historical period is determined.

[0218] The obtained historical occurrence rates and historical anomaly probabilities are used as sample data corresponding to the historical cycles.

[0219] In one possible embodiment, the processing module 802 is specifically used for:

[0220] Statistics on the distribution of playback volume for each multimedia file;

[0221] Based on statistical distribution information, the confidence interval of playback volume for each multimedia file is determined, where the confidence interval of playback volume represents the range of normal playback volume.

[0222] The historical anomaly probability of each multimedia file within a historical period is determined by the ratio of the weighted sum of abnormal playback counts that are not within the playback count confidence interval in each multimedia file's playback count to the total number of multimedia files.

[0223] In one possible embodiment, the processing module 802 is specifically used for:

[0224] Get the number of times each multimedia file has been played within a historical period;

[0225] The sum of the playback counts for each file is taken as the total playback count.

[0226] The ratio of the number of times each file is played to the total number of times it is played is used as the playback count of the corresponding multimedia file.

[0227] In one possible embodiment, the distribution information includes the average number of plays and the standard deviation of plays for each multimedia file; the processing module 802 is specifically used for:

[0228] The minimum value is the difference between the product of the preset weight and the standard deviation of the playback volume and the average playback volume.

[0229] The maximum value is the sum of the product of the preset weight and the standard deviation of the playback volume, and the average playback volume.

[0230] Based on the range between the obtained minimum and maximum values, the confidence interval for play count is determined.

[0231] Please refer to Figure 9The aforementioned device for detecting abnormal playback factors can run on a computer device 900. The current and historical versions of the data storage program, as well as the application software corresponding to the data storage program, can be installed on the computer device 900, which includes a processor 980 and a memory 920. In some embodiments, the computer device 900 may include a display unit 940, which includes a display panel 941 for displaying a user-interactive interface, etc.

[0232] In one possible embodiment, the display panel 941 may be configured in the form of a liquid crystal display (LCD) or an organic light-emitting diode (OLED).

[0233] The processor 980 is used to read a computer program and then execute the methods defined by the computer program. For example, the processor 980 reads a data storage program or file, thereby running the data storage program on the computer device 900 and displaying the corresponding interface on the display unit 940. The processor 980 may include one or more general-purpose processors, and may also include one or more DSPs (Digital Signal Processors) for performing related operations to implement the technical solutions provided in the embodiments of this application.

[0234] The memory 920 generally includes main memory and secondary storage. Main memory can be random access memory (RAM), read-only memory (ROM), and cache, etc. Secondary storage can be a hard disk, optical disk, USB flash drive, floppy disk, or magnetic tape drive, etc. The memory 920 is used to store computer programs and other data. The computer programs include applications corresponding to each client, and other data may include data generated after the operating system or applications are run, including system data (e.g., operating system configuration parameters) and user data. In this embodiment, program instructions are stored in the memory 920, and the processor 980 executes the program instructions in the memory 920 to implement any of the methods described in the preceding figures.

[0235] The aforementioned display unit 940 is used to receive input digital information, character information, or contact touch operations / non-contact gestures, and to generate signal inputs related to user settings and function control of the computer device 900. Specifically, in this embodiment, the display unit 940 may include a display panel 941. The display panel 941, for example, is a touch screen, which can collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or on the display panel 941), and drive corresponding connection devices according to a pre-set program.

[0236] In one possible embodiment, the display panel 941 may include two parts: a touch detection device and a touch controller. The touch detection device detects the player's touch position and the signal generated by the touch operation, and transmits the signal to the touch controller. The touch controller receives touch information from the touch detection device, converts it into touch point coordinates, and sends it to the processor 980. It can also receive and execute commands from the processor 980.

[0237] The display panel 941 can be implemented using various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the display unit 940, in some embodiments, the computer device 900 may also include an input unit 930. The input unit 930 may include an image input device 931 and other input devices 932, wherein the other input devices may include, but are not limited to, one or more of the following: a physical keyboard, function keys (such as volume control buttons, power buttons, etc.), a trackball, a mouse, and a joystick.

[0238] In addition to the above, the computer device 900 may also include a power supply 990 for powering other modules, an audio circuit 960, a near-field communication module 970, and an RF circuit 910. The computer device 900 may also include one or more sensors 950, such as an accelerometer, a light sensor, and a pressure sensor. The audio circuit 960 specifically includes a speaker 961 and a microphone 962, for example, the computer device 900 can use the microphone 962 to collect the user's voice and perform corresponding operations.

[0239] As one embodiment, the number of processors 980 can be one or more, and the processors 980 and the memory 920 can be coupled together or relatively independent.

[0240] As one example, Figure 9 The processor 980 in the middle can be used to implement, for example Figure 8 The functions of the acquisition module 801 and the processing module 802 in the process.

[0241] As one example, Figure 9 The processor 980 in the text can be used to implement the functions of the server or terminal devices discussed above.

[0242] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0243] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of software products, for example, through a computer program product. This computer program product is stored in a storage medium and includes several instructions to cause a computer device to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0244] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

[0245] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for detecting abnormal playback factors, characterized in that, include: Obtain the target occurrence rate of each abnormal factor when playing each multimedia file within the target period; Using a target probability prediction model, based on the obtained occurrence rate of each target, the target anomaly probability of each multimedia file is predicted when playing each multimedia file within the target period, wherein the target anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file; Using a target probability prediction model and target model parameters corresponding to each of the aforementioned abnormal factors, based on the occurrence rate of each target and the target anomaly probability, the target root cause probability corresponding to each abnormal factor is determined when the abnormal playback volume exists in the playback volume of each multimedia file within the target period. The target root cause probability represents the probability that an abnormal factor leads to abnormal playback volume. The target model parameters are obtained through gradient descent iterative training based on each sample data. In each training round, the model parameters are adjusted based on the mean square error between the test anomaly probability output by the probability prediction model for the sample data and the historical anomaly probability recorded in the sample data. The target model parameters are used to describe the conditional probability of abnormal playback volume given the occurrence of one abnormal factor under the statistical regularity within the historical period. Based on the obtained root cause probabilities of each target, the abnormal factors are sorted to generate the detection results of abnormal playback factors.

2. The method according to claim 1, characterized in that, The target probability prediction model includes multiple target model parameters, and different target model parameters are trained for different anomalies. The target probability prediction model, based on the occurrence rate of each target and the target anomaly probability, determines the target root cause probability corresponding to each abnormal factor when the abnormal playback volume exists in the playback volume of each multimedia file within the target period, including: For each of the aforementioned abnormal factors, perform the following operations respectively: The probability of an abnormal factor is determined by multiplying the target occurrence rate of the abnormal factor with the target model parameters corresponding to the abnormal factor. The probability of an abnormal factor represents the probability that there is an abnormal playback volume caused by the abnormal factor in the playback volume of the multimedia file. The ratio between the probability of the factor and the probability of the target anomaly is taken as the target root cause probability of the abnormal factor when the abnormal playback volume exists in the playback volume of each multimedia file within the target period.

3. The method according to claim 1 or 2, characterized in that, The target probability prediction model was trained using the following method: Obtain sample data corresponding to multiple historical periods, wherein the sample data includes the historical anomaly probability of each multimedia file in the historical period, and the historical occurrence rate of each anomaly factor in the corresponding historical period, and the historical anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file in the historical period. Multiple sample data points from each of the obtained sample data points are used as training data to establish a training dataset, and at least one sample data point from each of the sample data points is used as training data to establish a test dataset. Based on the training dataset and the test dataset, the probability prediction model is trained iteratively in multiple rounds to output the trained target probability prediction model.

4. The method according to claim 3, characterized in that, The step of performing multiple rounds of iterative training on the probabilistic prediction model based on the training dataset and the test dataset, and outputting the trained target probabilistic prediction model, includes: In each round of training iterations, perform the following operations: Each historical occurrence rate in the training data is used as the input to the probability prediction model, and the historical anomaly probability in the training data is used as the output of the probability prediction model to determine the training model parameters of the probability prediction model corresponding to each anomaly factor. Using the probability prediction model that includes the parameters of each of the obtained training models, the test anomaly probability of each multimedia file is predicted based on the historical occurrence rate in the test data, wherein the test anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file within the historical period. The probability prediction model is trained based on the error between the obtained test anomaly probability and the historical anomaly probability in the test data.

5. The method according to claim 3, characterized in that, The acquisition of sample data corresponding to each of the multiple historical periods includes: For each of the aforementioned historical periods, perform the following operations respectively: The playback count of each multimedia file and the historical occurrence rate of each abnormal factor are obtained within the historical period. Based on the playback volume of each multimedia file, the historical anomaly probability of each multimedia file within the historical period is determined. The obtained historical occurrence rates and historical anomaly probabilities are used as sample data corresponding to the historical periods.

6. The method according to claim 5, characterized in that, The determination of the historical anomaly probability of each multimedia file within the historical period based on the playback volume of each multimedia file includes: Statistical analysis of the playback volume distribution information for each of the aforementioned multimedia files; Based on statistical distribution information, the confidence interval of the playback volume of each multimedia file is determined, wherein the confidence interval of the playback volume represents the range of normal playback volume. The historical anomaly probability of each multimedia file within the historical period is determined by the ratio of the weighted sum of abnormal playback counts that are not within the confidence interval of the playback count in each multimedia file to the total number of files in each multimedia file.

7. The method according to claim 5, characterized in that, The playback count of each multimedia file within the acquisition history period includes: Obtain the number of times each multimedia file has been played within the historical period; The sum of the playback counts for each file is taken as the total playback count; The ratio between the number of times each file is played and the total number of times it is played is taken as the playback count of the corresponding multimedia file.

8. The method according to claim 6, characterized in that, The distribution information includes the average number of plays and the standard deviation of plays for each multimedia file. The determination of the confidence interval for the playback volume of each multimedia file based on statistical distribution information includes: The minimum value is the difference between the product of the preset weight and the standard deviation of the playback volume and the average playback volume. The maximum value is the sum of the product of the preset weight and the standard deviation of the playback volume, and the average playback volume. The confidence interval for the number of plays is determined based on the range between the obtained minimum and maximum values.

9. A device for detecting abnormal playback factors, characterized in that, include: Acquisition module: used to acquire the target occurrence rate of each abnormal factor when playing each multimedia file within the target period; Processing module: Used to use a target probability prediction model to predict the target anomaly probability of each multimedia file when playing each multimedia file within the target period, based on the obtained occurrence rate of each target, wherein the target anomaly probability represents the probability that there is an abnormal playback volume in the playback volume of the multimedia file; The processing module is further configured to: employ a target probability prediction model and target model parameters corresponding to each of the abnormal factors, and based on the occurrence rate of each target and the target anomaly probability, determine the target root cause probability corresponding to each abnormal factor when the abnormal playback volume exists in the playback volume of each multimedia file within the target period, wherein the target root cause probability characterizes the probability that the abnormal factor causes the abnormal playback volume; the target model parameters are obtained by gradient descent iterative training based on each sample data, and in each round of training, the model parameters are adjusted based on the mean square error between the test anomaly probability output by the probability prediction model for the sample data and the historical anomaly probability recorded by the sample data; the target model parameters are used to describe: under the statistical regularity within the historical period, the conditional probability of the existence of abnormal playback volume given the occurrence of an abnormal factor; The processing module is further configured to: sort the abnormal factors based on the obtained root cause probabilities of each target, and generate detection results of abnormal factors in playback.

10. A computer device, characterized in that, include: Memory, used to store program instructions; A processor is configured to invoke program instructions stored in the memory and execute the method as described in any one of claims 1 to 8 according to the obtained program instructions.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for causing a computer to perform the method as described in any one of claims 1 to 8.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method as described in any one of claims 1 to 8.

Images