Conversation problem diagnosis method, device, medium and electronic device

Abstract

The embodiments of the present disclosure provide a method, device, medium and electronic device for diagnosing conversation problems. The method includes: obtaining a diagnosis request for a conversation problem, the diagnosis request including conversation information and a target diagnosis model identifier; determining a target diagnosis model and a target diagnosis parameter associated with the target diagnosis model according to the target diagnosis model identifier; obtaining indicator data according to the conversation information and the target diagnosis parameter; and generating a diagnosis result by combining the target diagnosis model with the indicator data. Through the corresponding target diagnosis model, conversation problems can be automatically diagnosed, conversation problems can be quickly located, personnel input can be reduced, and problem troubleshooting efficiency can be improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technology, and more specifically, to a conversation problem diagnosis method, device, medium, and electronic device. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims. No description herein is admitted to be prior art by inclusion in this section.

[0003] With the continuous development of technology and the widespread application of communication functions, more and more users are using real-time audio or video conversations for instant communication. WebRTC (Web Real-Time Communication) is a real-time audio and video open source framework from Google. Developers can implement audio and video call functions based on WebRTC.

[0004] Currently, the troubleshooting process of real-time audio and video systems mainly relies on technicians to diagnose common problems and then hand over difficult problems to experts for diagnosis. This method is inefficient and cannot meet the growing conversation needs.

[0005] Public Content

[0006] The present disclosure provides a conversation problem diagnosis method, device, medium and electronic device to solve the problem of low efficiency in diagnosing audio and video conversation problems manually.

[0007] In a first aspect of an embodiment of the present disclosure, a method for diagnosing a session problem is provided, comprising: obtaining a diagnosis request for a session problem, the diagnosis request comprising session information and a target diagnosis model identifier; determining a target diagnosis model and target diagnosis parameters associated with the target diagnosis model according to the target diagnosis model identifier; obtaining indicator data according to the session information and the target diagnosis parameters; and generating a diagnosis result by using the target diagnosis model in combination with the indicator data.

[0008] In one embodiment of the present disclosure, the target diagnosis model is obtained by mathematically modeling the historical session diagnosis data using a mathematical calculation model.

[0009] In another embodiment of the present disclosure, the target diagnostic model includes one or more of the following: a tree-shaped diagnostic model for analyzing and diagnosing session audio freezes and / or session video freezes; an abnormal period diagnostic model for analyzing and diagnosing session audio silence and / or session video no picture problems; and a multi-event diagnostic model for analyzing and diagnosing session connectivity problems.

[0010] In another embodiment of the present disclosure, the method also includes obtaining the tree-shaped diagnostic model; wherein, obtaining the tree-shaped diagnostic model includes: constructing a tree-shaped diagnostic graph according to the before-and-after logic of the session problem diagnostic steps in the historical session diagnostic data, the tree-shaped diagnostic graph includes a root node, an intermediate node, a leaf node, and calculation indicators and calculation algorithms for each node, wherein the root node represents the first diagnostic step, the leaf node represents the last diagnostic step, and data calculation is performed on each node according to the calculation indicators and the calculation algorithm.

[0011] In another embodiment of the present disclosure, when the target diagnostic model is a tree-shaped diagnostic model, the use of the target diagnostic model in combination with the indicator data to generate a diagnostic result includes: calculating the nodes in the tree-shaped diagnostic model according to the indicator data until the path from the root node to the leaf node in the tree-shaped diagnostic model is traversed to generate a diagnostic conclusion of the path; and generating a diagnostic result based on the diagnostic conclusion of each path.

[0012] In another embodiment of the present disclosure, the calculation of the nodes in the tree diagnostic graph according to the indicator data includes: obtaining the calculation result of the parent node corresponding to the node; obtaining the node data from the indicator data according to the calculation indicator of the node; combining the calculation result and the node data to generate the calculation data of the node; calculating the calculation data according to the calculation algorithm of the node to obtain the calculation result of the node; when the node is a leaf node, obtaining a diagnostic conclusion according to the calculation result; when the node is an intermediate node, sending the calculation result to the child node corresponding to the node.

[0013] In another embodiment of the present disclosure, the analyzing and diagnosing the problem of silent audio in a session and / or no picture in a session video includes: obtaining a session abnormal state caused by an uplink operation in historical session diagnostic data, the session abnormal state including silent audio in a session and / or no picture in a session video; determining an associated event group, the associated event group including a first event triggered by an uplink operation to cause the session abnormal state to occur, and a second event triggered to end the session abnormal state; determining a time period extraction rule, the time period extraction rule including extracting the triggering time of the first event and the second event; extracting time series data from the indicator data according to the time period extraction rule to generate a time series; determining a time period for the occurrence of the session abnormal state according to the time series; and generating a diagnosis result according to the time period for the occurrence of the session abnormal state.

[0014] In another embodiment of the present disclosure, the method also includes obtaining a multi-event diagnostic model; wherein, obtaining the multi-event diagnostic model includes: constructing a finite state machine based on user connectivity events in historical session diagnostic data; the finite state machine includes a state set, a conversion event set and a state transition matrix; wherein, the state set is constructed based on the user's connectivity state, and the state set includes a start state, an end state and multiple intermediate states; the conversion event set is constructed based on events that trigger changes in the user's connectivity state, and the conversion events of the conversion event set include event identifiers, event names, event parameters and event texts; the state transition matrix is ​​constructed based on transition paths between various states, and the transition paths are marked with events that trigger state transitions.

[0015] In another embodiment of the present disclosure, when the target diagnostic model is a multi-event diagnostic model, the use of the target diagnostic model in combination with the indicator data to generate a diagnostic result includes: inputting the indicator data into the finite state machine to generate state transition data; and generating a diagnostic result based on the state transition data.

[0016] In another embodiment of the present disclosure, the obtaining of a diagnosis request for a session problem includes: obtaining session data to be diagnosed; detecting whether the session data has an abnormality according to a preset detection rule; and generating a diagnosis request for the session problem when an abnormality is detected in the session data.

[0017] In a second aspect of the embodiments of the present disclosure, a session problem diagnosis device is provided, comprising: a diagnosis request acquisition module, used to obtain a diagnosis request for a session problem, the diagnosis request including session information and a target diagnosis model identifier; a model determination module, used to determine a target diagnosis model and target diagnosis parameters associated with the target diagnosis model according to the target diagnosis model identifier; an indicator data acquisition module, used to obtain indicator data according to the session information and the target diagnosis parameters; and a diagnosis module, used to generate a diagnosis result by using the target diagnosis model in combination with the indicator data.

[0018] In a third aspect of the embodiments of the present disclosure, a storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for diagnosing a conversation problem as described in any one of the above items is implemented.

[0019] In a fourth aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the session problem diagnosis method as described in any one of the above items by executing the executable instructions.

[0020] According to the conversation problem diagnosis method, device, medium and electronic device of the embodiment of the present disclosure, according to the diagnosis request of the conversation problem, a suitable target diagnosis model is determined, and the diagnosis result is generated by using the target diagnosis model and related indicator data. Different conversation problems can be analyzed and diagnosed through different target diagnosis models, which reduces the input of manpower, significantly improves the efficiency of troubleshooting, and quickly locates and solves the problems generated during the user's audio and video conversation, bringing users a better audio and video conversation experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown in an exemplary and non-limiting manner, in which:

[0022] Figure 1 The schematic diagram schematically shows a framework of an application scenario according to an embodiment of the present disclosure;

[0023] Figure 2 The flowchart of the method for diagnosing a conversation problem according to an embodiment of the present disclosure is schematically shown;

[0024] Figure 3 The flowchart of the method for diagnosing a conversation problem according to another embodiment of the present disclosure is schematically shown;

[0025] Figure 4 The diagram schematically shows a diagnostic flow chart of a tree-shaped diagnostic model in a specific application scenario;

[0026] Figure 5 The flowchart of a method for diagnosing a conversation problem according to another embodiment of the present disclosure is schematically shown;

[0027] Figure 6 The flowchart of a method for diagnosing a conversation problem according to another embodiment of the present disclosure is schematically shown;

[0028] Figure 7 A schematic diagram schematically shows a state transition matrix in a specific application scenario of the present disclosure;

[0029] Figure 8 The structure diagram of the storage medium provided according to the embodiment of the present disclosure is schematically shown;

[0030] Fig. 9 The structure diagram of the conversation problem diagnosis device provided according to the embodiment of the present disclosure is schematically shown;

[0031] Fig.10 The schematic diagram schematically shows the structure of an electronic device provided according to an embodiment of the present disclosure.

[0032] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0033] The principles and spirit of the present disclosure will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0034] Those skilled in the art will appreciate that the embodiments of the present disclosure may be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0035] According to an embodiment of the present disclosure, a conversation problem diagnosis method, device, medium and electronic device are proposed.

[0036] In this article, it is important to understand that the terms used have the following meanings:

[0037] WebRTC: The name comes from the abbreviation of Web Real-Time Communication. WebRTC is a real-time audio and video open source framework from Google. Developers can implement audio and video call functions based on WebRTC. WebRTC includes audio and video acquisition and playback, encoding and decoding, pre- and post-processing, network transmission, as well as QoS modules such as anti-packet loss and anti-jitter. It is a relatively mature client-side real-time audio and video framework.

[0038] Real-time audio and video system: The real-time audio and video system includes multiple nodes such as the uplink side, signaling service, media service, downlink side, and data center. By setting up WebRTC modules on the uplink and downlink sides, the real-time audio and video conversation service functions of the uplink and downlink sides are realized. In order to continuously improve the quality of audio and video calls and ensure user experience, the developers of the real-time audio and video conversation system will add a large number of indicator reports and event reports at each node of the system, especially in the core module of the uplink / downlink side - WebRTC. These reports will play a key role in troubleshooting, algorithm optimization, and effect improvement.

[0039] In addition, any number of elements in the drawings is for illustration and not limitation, and any naming is only for distinction and does not have any limiting meaning.

[0040] The principle and spirit of the present disclosure are explained in detail below with reference to several representative embodiments of the present disclosure.

[0041] Public Overview

[0042] The inventors have found that, at present, there are three main channels for discovering audio and video conversation problems: one is direct feedback from users; the second is development and testing to discover problems during the R&D and testing phase; and the third is big data mining to discover problems. After discovering a conversation problem, a large amount of manpower is required for analysis and processing. For problems discovered through the first discovery channel, technical support is generally first connected and investigated to eliminate some common problems or user operation problems, and then transferred to technical experts for further processing based on the complexity of the problem. For problems discovered through the other two discovery channels, they are generally directly handed over to technical experts for analysis and processing.

[0043] Manual diagnosis and analysis of conversation problems has the following drawbacks:

[0044] 1. For each problem session, technical experts need to conduct several repeated analyses. Especially when there are a large number of problem sessions, the session duration is long, the problem description is inaccurate, and the proportion of common problems increases, the benefits of this repetitive work will gradually decrease, resulting in a waste of manpower.

[0045] Second, when different technical experts analyze conversation problems, they often fill in the diagnosis results according to their personal habits, which makes it difficult to form a standardized data structure. This is not friendly to automation scenarios where there may be a need to reuse the analysis results.

[0046] 3. Manual analysis of conversation problems relies heavily on expert experience. The loss of technical experts means a decline in the problem-solving capabilities of departments and companies, and an inability to form a long-term mechanism for solving conversation problems.

[0047] Fourth, for each channel for discovering session problems, the session problems need to be located and diagnosed after the session ends. It is impossible to locate and analyze the session problems in real time when they occur, resulting in a poor user experience.

[0048] Based on this, the embodiments of the present disclosure provide a conversation problem diagnosis method, device, medium and electronic device, which can analyze and diagnose different conversation problems through different target diagnosis models, reduce manpower input, significantly improve the efficiency of problem troubleshooting, and quickly locate and solve problems arising during user audio and video conversations, bringing users a better audio and video conversation experience.

[0049] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0050] Application Scenario Overview

[0051] Figure 1 A schematic diagram schematically shows a framework of an exemplary application scenario of an embodiment of the present invention.

[0052] refer to Figure 1 , the embodiments of the present disclosure are applied to a real-time audio and video conversation system 100. The real-time audio and video conversation system 100 includes an uplink end side 110, a downlink end side 120, a service cluster 130, and a data cluster 140. The service cluster 130 includes a signaling server 131 and a media server 132. The data cluster 140 includes a data center 141, a data platform 142, a data server 143, an abnormality perception server 144, and an automatic diagnosis server 145.

[0053] The uplink side 110 and the downlink side 120 are integrated with WebRTC modules, the signaling server 131 performs communication coordination, manages sessions and connections; the media server 132 receives, stores and shares media, and the uplink side 110, the downlink side 120, the signaling server 131 and the media server 132 cooperate with each other to complete the main business functions of the real-time audio and video conversation system. The data center 141 is used to persistently store various indicators, events and other data generated by the main business of the audio and video call system; the data platform 142 is used to provide big data query display; the data server 143 is used to provide a standardized interface for data query for other services, the anomaly perception server is used to mine possible anomalies from the audio and video business big data, and the automatic diagnosis server is used to diagnose conversation problems in real time.

[0054] The various modules of the real-time audio and video conversation system 100 are introduced as follows:

[0055] The uplink side 110 and the downlink side 120 log in to the signaling server 131 of the audio and video session system. After verifying the client, the signaling server 131 assigns a session room number and a session user number to it. The user shares the conference room number with other session participants through other channels. Other participants log in to the service and use the session room number to log in to the room.

[0056] The client side and the media server 132 use the SDP information of WebRTC to establish an audio and video media link with each other.

[0057] The user side, media service, and signaling service maintain time synchronization with the data center 141 during the session, and send the generated events and indicators to the data center 141 for persistent storage through a reporting process.

[0058] The data platform 142 requests the required session, user event, indicator and other data from the data center 141, displays them visually, and provides them to backend operations, development and other personnel.

[0059] The data server 143 may provide an HTTP Restful interface to provide other services with services for pulling required data from the data center 141 .

[0060] The abnormality perception server 144 obtains data such as events or indicators from the data server 143, performs big data statistics on the indicators and events of the audio and video conversations, and discovers abnormal audio and video conversations and users according to certain rules. For example, the abnormality perception server 144 can calculate the audio and video freeze rate in a large number of conversation rooms, and when the freeze rate of the conversation room is higher than the average freeze rate, the conversation room is determined to be an abnormal conversation room. Alternatively, in an abnormal conversation room, the audio and video freeze rate of each user in the conversation room is obtained, and when the audio and video freeze rate of some users is higher than that of other users, the user is determined to be an abnormal user.

[0061] The automatic diagnosis server 145 generates a diagnosis request including session information and a target diagnosis model identifier based on the session room number and user session number determined as abnormal by the anomaly perception server 144. The target diagnosis model and the target diagnosis parameters associated with the target diagnosis model are determined based on the target diagnosis model identifier in the diagnosis request. The corresponding indicator data is obtained from the data center based on the session information and the target diagnosis parameters. The target diagnosis model is combined with the indicator data to generate a diagnosis result to realize the automated analysis of certain specific problems in the session.

[0062] Understandably, Figure 1 The real-time audio and video conversation system shown is only an example of an application scenario in which the embodiments of the present disclosure can be implemented. The scope of application of the embodiments of the present disclosure is not limited in any aspect of the above framework. For example, the implementation method of the present disclosure can also be applied to a separate diagnostic system, and the diagnostic system can receive the conversation data through the data interface, perform offline diagnosis on the conversation data, etc.

[0063] Exemplary Methods

[0064] Combine the following Figure 1 For application scenarios, refer to Figure 2 To describe the conversation problem diagnosis method according to the exemplary embodiment of the present disclosure. It should be noted that the above application scenarios are only shown to facilitate understanding of the spirit and principle of the present disclosure, and the embodiments of the present disclosure are not limited in this respect. On the contrary, the embodiments of the present disclosure can be applied to any applicable scenario.

[0065] First, a conversation problem diagnosis method is introduced through a specific embodiment.

[0066] Figure 2 This is a flow chart of a method for diagnosing a conversation problem provided by an embodiment of the present disclosure. Figure 2 The conversation problem diagnosis method provided in this embodiment may include:

[0067] Step S210, obtaining a diagnosis request for a session problem, wherein the diagnosis request includes session information and a target diagnosis model identifier;

[0068] Step S220, determining a target diagnostic model and target diagnostic parameters associated with the target diagnostic model according to the target diagnostic model identifier;

[0069] Step S230, acquiring indicator data according to the session information and the target diagnostic parameter;

[0070] Step S240: Generate a diagnosis result by combining the target diagnosis model with the indicator data.

[0071] The conversation problem diagnosis method of the disclosed embodiment determines a suitable target diagnosis model according to the diagnosis request of the conversation problem, and generates a diagnosis result by using the target diagnosis model and related indicator data. Different conversation problems can be analyzed and diagnosed through different target diagnosis models, which reduces the input of manpower, significantly improves the efficiency of troubleshooting, and quickly locates and solves the problems generated during the user's audio and video conversation, bringing users a better audio and video conversation experience.

[0072] In an example of the present disclosure, the session problem diagnosis request may be manually triggered, for example, a session user reports a problem during a session, triggering a session problem diagnosis request, or a technician reports a problem during data monitoring, triggering a session problem diagnosis request.

[0073] In another example of the present disclosure, a diagnosis request for a session problem may be automatically triggered by an abnormality sensing system, etc. Specifically, step S210 obtains a diagnosis request for a session problem, which may include:

[0074] Step S211, obtaining the session data to be diagnosed. Specifically, the session data of a preset time period can be obtained regularly as the session data to be diagnosed. The session data can include various indicator data related to the target diagnosis model, such as audio data, video data, connectivity data, etc., such as the freeze rate of each session room, packet loss data, etc.

[0075] Step S212, detect whether the session data is abnormal according to the preset detection rules. Specifically, the session audio freezes and / or the session video freezes; the session audio and / or the session video is silent or has no picture; the audio and video session users are connected normally, etc. For example, detecting whether the session data is abnormal according to the preset detection rules can be: determining whether the freeze rate of the session room and the session user in the session data is higher than the preset threshold, and determining that the session is abnormal when it is higher than the preset threshold. It can also be: determining whether the video or audio has a picture or sound, and determining that the session is abnormal when there is no picture in the video or no sound in the audio.

[0076] Step S213, when an abnormality is detected in the session data, a diagnosis request for the session problem is generated. The diagnosis request includes session information and a target diagnosis model identifier. Specifically, the session information may be relevant session data saved by the data platform, such as session room information, session user information, audio information, video information, and other data.

[0077] In one example, when an abnormality is detected in the session data, the target diagnostic model identifier is determined according to the abnormality type of the session problem. The target diagnostic model identifier can be, for example, numbered 1, 2, 3, etc. Different target diagnostic model identifiers are used to trigger different target diagnostic models. For example, when the abnormality type of the session problem is session audio freeze and / or session video freeze, the target diagnostic model identifier of the diagnosis request is numbered 1. When the abnormality type of the session problem is silent session audio and / or no picture in session video, the target diagnostic model identifier of the diagnosis request is numbered 2. When the abnormality type of the session problem is a user connectivity problem, the target diagnostic model identifier of the diagnosis request is numbered 3.

[0078] In one example, the target diagnosis model is obtained by mathematically modeling the historical conversation diagnosis data using a mathematical calculation model. Specifically, the historical conversation diagnosis data may be, for example, relevant data from experts troubleshooting and analyzing various audio and video conversation problems. Using mathematical calculation models in the computer field, the problem diagnosis methods of technical experts are converted into relevant models. For different audio and video conversation problems, due to their different diagnostic methods, corresponding target diagnosis models can be generated through different mathematical calculation models. After obtaining the model, in the subsequent conversation problem diagnosis process, the data can be directly input into the target diagnosis model, and the target diagnosis model can be automatically deduced to obtain the diagnosis results.

[0079] It is understandable that the mathematical calculation model may be, for example, a graph model (such as a tree model, etc.), a state machine, etc. Specifically, a suitable mathematical calculation model may be selected according to the diagnosis steps in the historical session data diagnosis data.

[0080] In one example, the target diagnostic model includes one or more of the following:

[0081] A tree-shaped diagnostic model is used to analyze and diagnose session audio freeze and / or session video freeze issues;

[0082] An abnormal time period diagnosis model, used to analyze and diagnose the problem of silent audio and / or no picture in the conversation video; and

[0083] Multi-event diagnosis model, used to analyze and diagnose session connectivity issues.

[0084] Different target diagnosis models can be used to analyze and diagnose problems such as session freezes, audio silence, video no picture, and session connectivity, which can solve most of the problems that occur in audio and video sessions. The construction and specific diagnosis process of the above three target diagnosis models will be described in detail below.

[0085] In step S220, the target diagnostic model and the target diagnostic parameters associated with the target diagnostic model are determined according to the target diagnostic model identifier. Specifically, the corresponding target diagnostic model is determined according to the mapping relationship between the pre-constructed target diagnostic model identifier and the target diagnostic model; and the associated target diagnostic parameters are determined according to the target diagnostic model. For example, when the target diagnostic model identifier is number 1, the target diagnostic model is determined to be a tree diagnostic model; when the target diagnostic model identifier is number 2, the target diagnostic model is determined to be an abnormal period diagnostic model; when the target diagnostic model identifier is number 3, the target diagnostic model is determined to be a multi-event diagnostic model.

[0086] In one example, the target diagnosis parameters may include parameters such as the session room number, the session user number (including the uplink user number and the downlink user number), and the session time period. It should be noted that different target diagnosis models correspond to different session problems to be diagnosed, and the algorithms used during diagnosis are different. Therefore, each target diagnosis model has its own target diagnosis parameters.

[0087] After obtaining the target diagnosis model identifier and the target diagnosis parameters, proceed to step S230: obtain indicator data according to the session information and the target diagnosis parameters. Specifically, obtain the indicator data required for each target diagnosis model from the session information according to the target diagnosis parameters. For example, the indicator data related to the target diagnosis parameters can be obtained by filtering from the session information stored in the data platform by data pulling.

[0088] In one example, the acquired indicator data may be cleaned, for example, by deleting duplicate values, supplementing missing values, normalizing the data, etc. The specific method of data cleaning may refer to the data preprocessing method in the prior art.

[0089] After obtaining the normalized index data through data cleaning, the process proceeds to step S240 to generate a diagnosis result by combining the target diagnosis model with the index data.

[0090] For the problem of session audio freeze and / or session video freeze, this type of problem usually occurs on the downlink user side. For example, the downlink user hears the voices of some uplink users and finds that the audio is intermittent. If the freeze problem is caused by the downlink, then the corresponding downlink related data can be found. If the freeze problem is caused by a certain uplink (such as: there is serious packet loss in the uplink network), which causes the downlink hearing or picture to freeze, it is necessary to find the uplink first to make an accurate diagnosis. However, in order to accurately reflect the operating status of the downlink end side, the real-time audio and video conversation system often distinguishes the connected uplink users among many downlink indicators. This results in the need to check and analyze a lot of data one by one when analyzing and diagnosing audio and video freeze problems, which is a huge workload. The present disclosure abstractly models the analysis process of the audio and video session freeze problem through a tree model. The modeled tree diagnosis model can be used for automatic analysis of audio and video freeze problems to reduce workload and improve problem diagnosis efficiency.

[0091] In one example, obtaining the tree-shaped diagnostic model may specifically include: constructing a tree-shaped diagnostic graph according to the before-and-after logic of the session problem diagnostic steps in the historical session diagnostic data, the tree-shaped diagnostic graph including a root node, an intermediate node, a leaf node, and calculation indicators and calculation algorithms for each node, wherein the root node represents the first diagnostic step, the leaf node represents the last diagnostic step, and data calculation is performed on each node according to the calculation indicators and the calculation algorithm.

[0092] Specifically, the calculation index may be, for example, an audio jam value, an uplink packet loss rate, a downlink packet loss rate, a short-term extreme value of sending RRT (Round-Trip Tim, round trip time), an uplink RTX sending bit rate (data flow used by an audio file per unit time), etc. It is understandable that the calculation index may be determined based on an index related to the generation of the audio and video session jam problem, and the present disclosure does not specifically limit this.

[0093] Based on the calculation index, relevant time series data is obtained from the audio and video conversation, and then the time series data is calculated according to the calculation algorithm to obtain the calculation result. The calculation algorithm includes but is not limited to time series sum, difference, product, quotient calculation, time series n-order difference calculation, time series threshold calculation, time series drift calculation, etc. The following is an example of several calculation algorithms.

[0094] Time series sum, difference, product and quotient calculation: calculate the sum, difference, product and quotient of the values ​​of two time series at the same time.

[0095] Time series n-order difference calculation: a time series is calculated by subtracting the k-th time value from the k+1-th time value to obtain a first-order difference sequence; for a first-order difference sequence, the k'+1-th time value minus the k'-th time value obtains a second-order difference sequence, and so on, to obtain the n-th order difference sequence.

[0096] Time series time shift calculation: The values ​​of the k+m…k+1, k, k-2…kn moments of a time series and the value of the kth moment are fused (or, and, maximum value, minimum value, etc.) in a certain way and assigned to the kth moment of the new sequence.

[0097] Short-term mean calculation of time series: the values ​​of the kth, k+1, k-2…kn moments of a time series are averaged and assigned to the kth moment of the new series.

[0098] Time series short-term average jitter calculation: The first-order difference of a time series at the kth, k+1, k-2…kn moments are averaged and assigned to the kth moment of the new sequence.

[0099] Time series short-term range calculation: the result of time drifting according to the maximum value of the absolute value sequence of the first-order difference of a time series.

[0100] Time series threshold calculation: A time series in which the value is greater than a given threshold is recorded as true, otherwise it is recorded as false, and a new time series with true / false values ​​is obtained.

[0101] Time series range calculation: A time series in which the values ​​fall within a given range are recorded as true, and vice versa, a new time series with true / false values ​​is obtained.

[0102] Time series peak (valley) detection algorithm: For a time series, for the values ​​at the k+m…k+1, k, k-2…kn moments, use a peak detection algorithm (such as 3-delta, box plot) to calculate whether the k-th moment value is a peak (valley), and obtain a new time series with true / false values.

[0103] Time series similarity calculation: For two time series, calculate the similarity of two value sequences (such as correlation coefficient, Euclidean distance). If the similarity is greater than a given threshold, they are considered similar, and a judgment on whether they are similar is obtained.

[0104] In an example, see Figure 3 In step S240, when the target diagnosis model is a tree diagnosis model, the use of the target diagnosis model in combination with the indicator data to generate a diagnosis result may specifically include:

[0105] Step 310, calculating the nodes in the tree-shaped diagnostic model according to the indicator data until the path from the root node to the leaf node in the tree-shaped diagnostic model is traversed to generate a diagnostic conclusion of the path.

[0106] Step S320: Generate a diagnosis result according to the diagnosis conclusion of each path.

[0107] In one example, in step S320, calculating the nodes in the tree diagnostic graph according to the indicator data may specifically include:

[0108] Step S321, obtaining the calculation result of the parent node corresponding to the node, that is, obtaining the calculation result of the previous node, and the calculation result can be, for example, a time series of true / false values.

[0109] Step S322, obtaining node data from the indicator data according to the calculation indicator of the node. Specifically, the node data may be time series data corresponding to the calculation indicator.

[0110] Step S323, combining the calculation result and the node data to generate the calculation data of the node.

[0111] Step S324, performing calculations on the calculation data according to the calculation algorithm of the node to obtain the calculation result of the node.

[0112] Step S325, when the node is a leaf node, a diagnosis conclusion is obtained according to the calculation result.

[0113] Step S326, when the node is an intermediate node, the calculation result is sent to the child node corresponding to the node. The child node continues to perform calculation according to steps S321-S326.

[0114] Based on each node of the tree-shaped diagnosis model, operations are performed in sequence according to the corresponding calculation indicators and calculation algorithms of the nodes until each path from the root node to the leaf node is traversed. The diagnosis result can be determined based on the calculation results of each path.

[0115] Figure 4 The following schematically shows the diagnostic flow chart of the tree-shaped diagnostic model in a specific application scenario. Figure 4 As shown, the root node of the tree-shaped diagnosis model is audio jamming, the intermediate nodes include downlink packet loss rate, uplink packet loss rate, etc., and the leaf nodes include uplink RRT short-term extreme value, downlink packet number, service downlink packet loss rate, etc. For example, when the downlink packet receiving interval of the leaf node is greater than 8, the diagnosis result is obtained: the downlink packet receiving interval is large.

[0116] It should be noted that Figure 4It is only a brief schematic diagram of the diagnosis process of the tree diagnosis model, and the calculation indicators and calculation algorithms of each node of the tree diagnosis model are not fully shown.

[0117] In another specific application scenario, when performing calculations on certain nodes, for example, the following steps may be followed:

[0118] (1) Performing timing threshold calculation on the audio freeze timing sequence, taking 0 ms as the threshold, and obtaining a freeze timing sequence with a value of true / false.

[0119] (2) Filter out the time that is the same as the time when the freeze time sequence is true from the downlink packet loss time sequence, that is, a subsequence of the downlink packet loss time sequence. Take 0ms as the threshold, perform time threshold calculation on this subsequence, and obtain a downlink packet loss time sequence with a value of true / false.

[0120] (3) Since downlink packet loss reflects full-link packet loss, the moment that is the same as the moment when the downlink packet loss timing sequence is true is selected from the uplink packet loss timing sequence, that is, a subsequence of the uplink packet loss timing sequence. With 0ms as the threshold, this subsequence is subjected to timing threshold calculation processing to obtain the uplink packet loss timing sequence. If the value of the uplink packet loss timing sequence is true, it is considered that the audio freeze is associated with the uplink packet loss. Otherwise, it is considered that there is no association, and the calculation continues to the next node.

[0121] For the problem of silent audio in conversations and / or no picture in conversation videos, the diagnosis results of such problems mainly include that the uplink audio and video is not turned on or the uplink audio and video stream is stopped. Downlink users cannot accurately predict the uplink operation, and it is difficult to think that the situation of silent audio and video or no picture may be caused by the uplink operation. Therefore, feedback from users about such problems is often received. The present disclosure can quickly, accurately and automatically locate the causes of such problems through the abnormal time period diagnosis model, quickly respond to customer questions, and reduce the investment of expert manpower. Specifically, the abnormal time period diagnosis model includes associated event groups and time period extraction rules.

[0122] See also Figure 5 In one example, when the target diagnosis model is an abnormal period diagnosis model, the target diagnosis model is combined with the indicator data to generate a diagnosis result, which may specifically include:

[0123] Step 510: Acquire a session abnormality state caused by an uplink operation in historical session diagnostic data, where the session abnormality state includes a silent session audio and / or a blank session video.

[0124] Step S520: determining an associated event group, wherein the associated event group includes a first event triggered by an uplink operation to cause the abnormal session state to occur, and a second event triggered to cause the abnormal session state to end.

[0125] Specifically, based on the expert experience in the historical session diagnosis data, we can identify logically related associated event groups. For example, if the uplink user connected to the downlink user triggers a local audio stream closing event, then until the next time the uplink user triggers a local audio stream opening event, the uplink user is in a state of not sending audio streams. In this case, the downlink must have an abnormal session state with silent audio. At this point, closing the local audio stream and opening the local audio stream are a pair of logically related associated event groups.

[0126] As shown in the following Table 1, a group of associated events in a specific application scenario of the present disclosure is schematically shown.

[0127] Table 1

[0128]

[0129]

[0130] Step 530: determine a time period extraction rule, where the time period extraction rule includes extracting the triggering time of the first event and the second event.

[0131] Specifically, in one example, the time period extraction rule is: the start time of the session is start_time, and the end time of the session is end_time; for the time period between start_time and end_time, the triggering moments that trigger the first event and the second event are extracted, and the triggering moments are arranged in a time sequence from far to near.

[0132] Step 540: extract the time series data in the indicator data according to the time period extraction rule to generate a time series. Specifically, the indicator data corresponding to the abnormal time period diagnosis model is the action time series data corresponding to the uplink user and the associated event group.

[0133] For example, in a specific application scenario, the associated event group is uplink mute on and uplink mute off. The indicator data is shown in Table 2, and the extracted uplink mute on time series data A: 1, 5, 9; uplink mute off time series data B: 2, 7, 10.

[0134] Table 2

[0135]

[0136] Step 550, determine the time period of the abnormal session state according to the time sequence. Specifically, according to the time sequence, the duration period between the first event and the second event is obtained, and the duration period is the time period of the abnormal session state. For example, in the specific application scenario shown in Table 2, the duration period between the first event and the second event includes: 1-2, 5-7, 9-10. In these time periods, the uplink mute is turned on, resulting in no audio.

[0137] Step 560, generating a diagnosis result according to the time period of the abnormal session state. Specifically, matching the obtained time period with the time point when the session problem (no sound, no video) occurs, and the cause of the session problem can be clearly identified according to the matching result.

[0138] For connectivity issues in audio and video sessions, a user may have a large number of login, logout, and re-login operations. The manual process of sorting out and analyzing and diagnosing user connectivity issues is a huge engineering effort. The present disclosure uses a multi-event diagnosis model to output the user's login abnormal status, without relying on manual sorting of events one by one, which effectively improves the diagnostic efficiency of connectivity issues.

[0139] In one example, obtaining a multi-event diagnosis model may specifically include: constructing a finite state machine according to user connectivity events in historical session diagnosis data; the finite state machine includes a state set, a conversion event set, and a state transition matrix.

[0140] Finite state machine is a mathematical model specially used to describe various state transition problems. User connectivity events are described by state sets, conversion event sets and state transition matrices. Specifically, user connectivity events include user login audio and video session events, exit audio and video session events, repeated login audio and video session events, abnormal exit audio and video session events, etc. Based on various user connectivity events appearing in historical session diagnosis data, a finite state machine is constructed as a multi-event diagnosis model to realize automatic diagnosis and analysis of user connectivity problems.

[0141] Specifically, the state set is constructed according to the user's connection state, and the state set may include a start state, a termination state, and multiple intermediate states. The start state is the initial state of the finite state machine, such as a state where the user has not joined the audio or video session. The intermediate state may be a state where the user's connection occurs normally, such as a state where the user is in a session, etc. The termination state may be a state where the user's connection occurs abnormally, such as a repeated joining state, an abnormal exit state, etc.

[0142] Specifically, the conversion event set is constructed based on the events that trigger the change of the user connectivity state. The conversion event set may include multiple events, each event corresponding to an action from one state to another. Each event may include an event identifier, an event name, event parameters, and an event text. The event identifier may be, for example, an event number, which is used to distinguish different events. The event name may be, for example, a specific action. The event parameter may be, for example, an event response code, etc. The event text is used to output text when the event occurs, so as to record the corresponding event and facilitate the diagnosis of user connectivity problems.

[0143] Specifically, the state transition matrix is ​​constructed according to the transition paths between the states, and the transition paths are marked with events that trigger state transitions. If there is no mark on the transition path, it means that the state transition is directly triggered without an event.

[0144] In an example, see Figure 6 In step S240, when the target diagnosis model is a multi-event diagnosis model, generating a diagnosis result by combining the target diagnosis model with the indicator data may include:

[0145] Step S610, input the indicator data into the finite state machine to generate state transition data. Specifically, after obtaining the finite state machine, the relevant indicator data, such as the event sequence data of the user's login action, is input into the finite state machine. The finite state machine can automatically deduce and output the user's state transition process, for example, it can output how many times the termination state (abnormal exit state, repeated joining state, etc.) has been performed, and output the triggering event or corresponding event text for each entry into the termination state.

[0146] Step S620: Generate a diagnosis result based on the state transition data. Specifically, based on the output result of the finite state machine, that is, whether the user has abnormally exited, repeatedly joined, etc., the user connectivity problem is diagnosed.

[0147] The following is a specific description using the actual application scenario of joining a room anomaly. Figure 7 The schematic diagram of the state transition matrix in this application scenario is schematically shown. In this application scenario, the state set has four elements: s represents the initial state, a represents the already in-session state, b represents the abnormal exit state, and c represents the repeated joining state. Among them, s is the starting state, a is the intermediate state, and b and c are the terminal states.

[0148] The conversion event set includes 5 events. Each conversion event is described as follows: event number: event name: {event parameter}: 'output text', as follows:

[0149] 1:JoinChannel:{code:0}:

[0150] 2:LeaveChannel:

[0151] 3:JoinChannel:{code:! 0}:'Failed to join the room'

[0152] 4:JoinChannel:{code:0}:'Duplicate join exception'

[0153] 5:JoinChannel:{code:! 0}:'Repeated join failed'

[0154] Among the above five events, the output texts of events No. 1 and No. 2 are empty, that is, no text is output when these two conversion events are triggered.

[0155] like Figure 7 As shown in the figure, the dotted arrows represent the transition paths between states and the triggering events. For example:

[0156] When the system is in the s state, if the 1:JoinChannel:{code:0}: event occurs, the system will transfer to the a state.

[0157] When the system is in state a, if event 2:LeaveChannel: occurs, the system will transfer to state s.

[0158] When the system is in state a and event 4:JoinChannel:{code:0} occurs, the system will transfer to state c and output "Duplicate join exception". Then the system will directly transfer from state c back to state a.

[0159] When the system is in state a, if event 5:JoinChannel:{code:! 0} occurs, the system will transfer to state b, output "repeated join failed", and then the system will directly transfer from state b back to state s.

[0160] When the system is in the s state, if the 3:JoinChannel:{code:! 0} event occurs, the system will transfer to the b state and output "failed to join the room", and then the system will directly return to the s state from the b state.

[0161] When a session diagnosis request is received, the event sequence data of the user's login action is input into the finite state machine, and the number and events of user repeated joining exceptions, repeated joining failures, and failed joining room can be automatically deduced to diagnose user connectivity problems.

[0162] Exemplary Media

[0163] After introducing the method of the exemplary embodiment of the present disclosure, next, refer to Figure 8 refer to Figure 8 As shown, in some possible implementations, various aspects of the present disclosure may also be implemented as a storage medium 80 on which program code is stored, and when the program code is executed by a processor of a device, it is used to implement the steps of the real-time recommendation method for social network users according to various exemplary implementations of the present application described in the above “Exemplary Method” section of this specification.

[0164] Specifically, when the processor of the device executes the program code, it is used to implement the following steps: step S210, obtaining a diagnosis request for a session problem, the diagnosis request including session information and a target diagnosis model identifier; step S220, determining a target diagnosis model and target diagnosis parameters associated with the target diagnosis model according to the target diagnosis model identifier; step S230, obtaining indicator data according to the session information and the target diagnosis parameters; step S240, generating a diagnosis result using the target diagnosis model in combination with the indicator data.

[0165] It should be noted that the above-mentioned medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0166] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, wherein readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to: electromagnetic signals, optical signals, or any suitable combination of the above. A readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0167] The program code contained in the readable medium can be transmitted with any appropriate medium, including but not limited to: wireless, wired, optical cable, RF, etc., or any suitable combination of the above. The program code for performing the operation of the present application can be written in any combination of one or more programming languages, and the programming language includes object-oriented programming languages-such as Java, C++, etc., and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user computing device, partially on the user computing device, partially on the remote computing device, or completely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network-including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0168] Exemplary Devices

[0169] After introducing the medium of the exemplary embodiment of the present disclosure, next, reference is made to Fig. 9 A conversation problem diagnosis device according to an exemplary embodiment of the present disclosure is described, which is used to implement the method in any of the above embodiments. Its implementation principle and technical effect are similar and will not be described in detail here.

[0170] refer to Fig. 9 , Fig. 9 This is a schematic diagram of the structure of a conversation problem diagnosis device provided by an embodiment of the present disclosure. Fig. 9 As shown, the conversation problem diagnosis device includes:

[0171] A diagnosis request acquisition module 910, used to acquire a diagnosis request for a session problem, wherein the diagnosis request includes session information and a target diagnosis model identifier;

[0172] A model determination module 920, configured to determine a target diagnosis model and target diagnosis parameters associated with the target diagnosis model according to the target diagnosis model identifier;

[0173] An indicator data acquisition module 930, configured to acquire indicator data according to the session information and the target diagnostic parameter;

[0174] The diagnosis module 940 is used to generate a diagnosis result by using the target diagnosis model in combination with the indicator data.

[0175] Among them, according to the diagnosis request of the conversation problem, the appropriate target diagnosis model is determined, and the diagnosis result is generated by using the target diagnosis model and related indicator data. Different conversation problems can be analyzed and diagnosed through different target diagnosis models, which reduces manpower input, significantly improves the efficiency of problem troubleshooting, and quickly locates and solves problems that arise during the user's audio and video conversation, bringing users a better audio and video conversation experience.

[0176] In one example, the diagnosis request acquisition module 910 is specifically used to: acquire session data to be diagnosed; detect whether the session data has an abnormality according to a preset detection rule; and generate a diagnosis request for the session problem when an abnormality is detected in the session data.

[0177] In one example, the target diagnostic model includes one or more of the following: a tree-shaped diagnostic model for analyzing and diagnosing session audio freezes and / or session video freezes; an abnormal period diagnostic model for analyzing and diagnosing session audio silence and / or session video no picture problems; and a multi-event diagnostic model for analyzing and diagnosing session connectivity problems.

[0178] In one example, the device also includes a module for obtaining a tree-shaped diagnostic model, which is specifically used to: construct a tree-shaped diagnostic diagram based on the before-and-after logic of the conversation problem diagnosis steps in the historical conversation diagnosis data, the tree-shaped diagnostic diagram including a root node, an intermediate node, a leaf node, and calculation indicators and calculation algorithms for each node, wherein the root node represents the first diagnostic step, the leaf node represents the last diagnostic step, and data calculation is performed on each node according to the calculation indicators and the calculation algorithm.

[0179] When the target diagnostic model is a tree-shaped diagnostic model, the diagnostic module 940 is specifically used to: calculate the nodes in the tree-shaped diagnostic model according to the indicator data until the path from the root node to the leaf node in the tree-shaped diagnostic model is traversed to generate a diagnostic conclusion of the path; and generate a diagnostic result based on the diagnostic conclusion of each path.

[0180] In one example, when the target diagnostic model is an abnormal period diagnostic model, the diagnostic module 940 is specifically used to: obtain a session abnormal state caused by an uplink operation in historical session diagnostic data, the session abnormal state including silent session audio and / or no session video; determine an associated event group, the associated event group including a first event triggered by an uplink operation to cause the session abnormal state to occur, and a second event that triggers the end of the session abnormal state; determine a period extraction rule, the period extraction rule including extracting the trigger time of the first event and the second event; extract the time series data in the indicator data according to the period extraction rule to generate a time series; determine the time period for the occurrence of the session abnormal state according to the time series; and generate a diagnostic result according to the time period for the occurrence of the session abnormal state.

[0181] In one example, the device further includes a module for obtaining a multi-event diagnostic model, and the module for obtaining a multi-event diagnostic model may specifically include: constructing a finite state machine according to user connectivity events in historical session diagnostic data; the finite state machine includes a state set, a conversion event set, and a state transition matrix. The state set is constructed according to the user's connectivity state, and the state set includes a start state, an end state, and multiple intermediate states; the conversion event set is constructed according to the event that triggers a change in the user's connectivity state, and the conversion event of the conversion event set includes an event identifier, an event name, an event parameter, and an event text; the state transition matrix is ​​constructed according to the transition path between each state, and the event that triggers the state transition is marked on the transition path.

[0182] In one example, when the target diagnosis model is a multi-event diagnosis model, the diagnosis module 940 is specifically used to: input the indicator data into the finite state machine to generate state transition data; and generate a diagnosis result according to the state transition data.

[0183] Exemplary Electronic Devices

[0184] After introducing the method, medium and apparatus of the exemplary embodiments of the present disclosure, next, reference is made to Fig.10 An electronic device according to an exemplary embodiment of the present disclosure is described.

[0185] Fig.10 The electronic device 1000 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0186] like Fig.10As shown, the electronic device 1000 is in the form of a general computing device. The components of the electronic device may include but are not limited to: at least one processing unit 1010, at least one storage unit 1020, and a bus 1030 connecting different system components (including the storage unit 1020 and the processing unit 1010).

[0187] The storage unit stores program codes, which can be executed by the processing unit 1010, so that the processing unit 1010 executes the steps described in the above “exemplary method” section of this specification according to various exemplary embodiments of the present disclosure.

[0188] The storage unit 1020 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 1021 and / or a cache memory unit 1022 , and may further include a read-only memory unit (ROM) 1023 .

[0189] The storage unit 1020 may also include a program / utility 1024 having a set (at least one) of program modules 1025, such program modules 1025 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0190] Bus 1030 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0191] The electronic device 1000 may also communicate with one or more external devices 1070 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device 1000, and / or any device that enables the electronic device 1000 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed through an input / output (I / O) interface 1050. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 1060. Fig.10 As shown, the network adapter 1060 communicates with other modules of the electronic device 1000 via the bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0192] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0193] It should be noted that although several units / modules or sub-units / modules of the conversation problem diagnosis device are mentioned in the above detailed description, such division is merely exemplary and not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided to be embodied by multiple units / modules.

[0194] In addition, although the operations of the disclosed method are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0195] Although the spirit and principle of the present disclosure have been described with reference to several specific embodiments, it should be understood that the present disclosure is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present disclosure is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims.

Claims

1. A method for diagnosing conversation problems, characterized in that: include: Obtaining a diagnosis request for a session problem, wherein the diagnosis request includes session information and a target diagnosis model identifier; Determining a target diagnostic model and target diagnostic parameters associated with the target diagnostic model according to the target diagnostic model identifier, wherein the target diagnostic model is obtained by mathematically modeling historical session diagnostic data using a mathematical calculation model, and the target diagnostic model includes a tree-shaped diagnostic model for analyzing and diagnosing session audio freezes and / or session video freezes; Acquire the tree-shaped diagnostic model, wherein acquiring the tree-shaped diagnostic model comprises: Constructing a tree-shaped diagnostic diagram according to the logic before and after the session problem diagnostic steps in the historical session diagnostic data, the tree-shaped diagnostic diagram includes a root node, an intermediate node, a leaf node, and a calculation index and a calculation algorithm for each node, wherein the root node represents the first diagnostic step, the leaf node represents the last diagnostic step, and data calculation is performed on each node according to the calculation index and the calculation algorithm; Acquiring indicator data according to the session information and the target diagnostic parameter; The target diagnosis model is combined with the indicator data to generate a diagnosis result.

2. The method for diagnosing conversation problems according to claim 1, characterized in that: The target diagnostic model also includes one or more of the following: An abnormal time period diagnosis model, used to analyze and diagnose the problem of silent audio and / or no picture in the video of the conversation; and Multi-event diagnosis model, used to analyze and diagnose session connectivity issues.

3. The method for diagnosing conversation problems according to claim 1, characterized in that: When the target diagnosis model is a tree-shaped diagnosis model, generating a diagnosis result by combining the target diagnosis model with the indicator data includes: Calculating the nodes in the tree-shaped diagnostic model according to the indicator data until the path from the root node to the leaf node in the tree-shaped diagnostic model is traversed to generate a diagnostic conclusion of the path; A diagnosis result is generated according to the diagnosis conclusion of each of the paths.

4. The method for diagnosing conversation problems according to claim 3, characterized in that: The calculation of the nodes in the tree-shaped diagnostic diagram according to the indicator data includes: obtaining the calculation result of the parent node corresponding to the node; obtaining the node data from the indicator data according to the calculation indicator of the node; combining the calculation result and the node data to generate the calculation data of the node; calculating the calculation data according to the calculation algorithm of the node to obtain the calculation result of the node; when the node is a leaf node, obtaining a diagnostic conclusion according to the calculation result; when the node is an intermediate node, sending the calculation result to the child node corresponding to the node.

5. The method for diagnosing conversation problems according to claim 1, characterized in that: The analysis and diagnosis of the problem of silent audio and / or no picture in the conversation video includes: Acquire a session abnormality state caused by an uplink operation in historical session diagnostic data, wherein the session abnormality state includes a silent session audio and / or a silent session video; Determine an associated event group, the associated event group including a first event that an uplink operation triggers the occurrence of the abnormal session state, and a second event that triggers the end of the abnormal session state; Determining a time period extraction rule, wherein the time period extraction rule includes extracting triggering times of the first event and the second event; Extracting the time series data in the indicator data according to the time period extraction rule to generate a time series; Determine the time period when the abnormal session state occurs according to the time series; A diagnosis result is generated according to the time period when the abnormal session state occurs.

6. The method for diagnosing conversation problems according to claim 2, characterized in that: The method further includes obtaining a multi-event diagnostic model; wherein obtaining the multi-event diagnostic model includes: A finite state machine is constructed based on user connectivity events in historical session diagnostic data; the finite state machine includes a state set, a conversion event set, and a state transition matrix; wherein, The state set is constructed according to the user's connectivity state, and the state set includes a start state, a stop state, and a plurality of intermediate states; The conversion event set is constructed according to the event that triggers the change of the user connection state, and the conversion event of the conversion event set includes an event identifier, an event name, an event parameter and an event text; The state transfer matrix is ​​constructed according to the transfer paths between various states, and the events triggering the state transfer are marked on the transfer paths.

7. The method for diagnosing conversation problems according to claim 6, characterized in that: When the target diagnosis model is a multi-event diagnosis model, generating a diagnosis result by using the target diagnosis model in combination with the indicator data includes: inputting the indicator data into the finite state machine to generate state transition data; and generating a diagnosis result according to the state transition data.

8. The method for diagnosing conversation problems according to claim 1, characterized in that: The request for obtaining the diagnosis of the session problem includes: Get the session data to be diagnosed; Detecting whether the session data is abnormal according to preset detection rules; When an abnormality is detected in the session data, a diagnosis request for the session problem is generated.

9. A conversation problem diagnosis device, characterized in that: include: A diagnosis request acquisition module, used to acquire a diagnosis request for a session problem, wherein the diagnosis request includes session information and a target diagnosis model identifier; A model determination module, used to determine a target diagnostic model and target diagnostic parameters associated with the target diagnostic model according to the target diagnostic model identifier, wherein the target diagnostic model is obtained by mathematically modeling historical session diagnostic data using a mathematical calculation model, and the target diagnostic model includes a tree-shaped diagnostic model, used to analyze and diagnose session audio freeze and / or session video freeze problems; An indicator data acquisition module, used to acquire indicator data according to the session information and the target diagnostic parameter; A diagnosis module, used to generate a diagnosis result by combining the target diagnosis model with the indicator data; Obtain a tree-shaped diagnostic model module, which is specifically used to: construct a tree-shaped diagnostic diagram according to the before-and-after logic of the conversation problem diagnosis steps in the historical conversation diagnosis data, the tree-shaped diagnostic diagram includes a root node, an intermediate node, a leaf node, and a calculation indicator and a calculation algorithm for each node, wherein the root node represents the first diagnostic step, the leaf node represents the last diagnostic step, and data calculation is performed on each node according to the calculation indicator and the calculation algorithm.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for diagnosing conversation problems according to any one of claims 1 to 8 is implemented.

11. An electronic device, characterized in that: include: processor; as well as A memory for storing executable instructions of the processor; wherein the processor is configured to execute the conversation problem diagnosis method according to any one of claims 1 to 8 by executing the executable instructions.

Images