Method and device for reconstructing tracking link of digital service network

Abstract

The embodiment of the invention provides a digital service network tracking link reconstruction method and device, and the method comprises the steps: obtaining the link tracking data and aggregation measurement data of a digital service network, and constructing a global directed acyclic graph according to the link tracking data; updating the global directed acyclic graph into a heterogeneous directed graph by using the aggregation metric data, and constructing a heterogeneous directed snapshot sequence according to the heterogeneous directed graph; the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence are input into a heterogeneous dynamic directed graph model to be processed, and node representations corresponding to nodes contained in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence are obtained; and link completion is performed on heterogeneous directed snapshots in the heterogeneous directed snapshot sequence according to the node features to obtain a target heterogeneous directed snapshot sequence, and a reconstructed link sequence is generated according to the target heterogeneous directed snapshot sequence.

Description

Technical Field

[0001] The embodiments of the present specification relate to the technical field of digital service networks, and more particularly to a method and device for reestablishing a trace link in a digital service network. Background Art

[0002] With the development of computer and Internet technology, microservice architecture has been widely used in digital service networks. The huge scale of service nodes and the complexity of call relationships in digital service networks have brought severe challenges to operation and maintenance management. The call chain generated by distributed tracing technology can accurately characterize the execution logic between various services in the digital service network, which has important research significance for improving the governance efficiency of the digital service network. There are two main challenges in distributed tracing in digital service networks. First, since there is no unified regulator for digital service networks, distributed tracing technology is an option for most services in digital service networks, resulting in missing links in the tracing chain. Second, distributed tracing technology focuses on the perception of call chains, and call data faces problems such as inconsistent naming and inconsistent granularity. These missing call chain data problems seriously affect the reliability of downstream analysis tasks based on observable data. Therefore, an effective solution is urgently needed to solve the above problems. Summary of the invention

[0003] In view of this, an embodiment of this specification provides a digital service network tracing link reconstruction method. One or more embodiments of this specification also relate to a digital service network tracing link reconstruction device, a computing device, a computer-readable storage medium and a computer program product to solve the technical defects existing in the prior art.

[0004] According to a first aspect of an embodiment of this specification, a digital service network tracing link reconstruction method is provided, comprising: Obtain link tracking data and aggregated metric data of the digital service network, and construct a global directed acyclic graph based on the link tracking data; Using the aggregated metric data, the global directed acyclic graph is updated to a heterogeneous directed graph, and a heterogeneous directed snapshot sequence is constructed according to the heterogeneous directed graph; Inputting the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence into a heterogeneous dynamic directed graph model for processing to obtain node representations corresponding to nodes included in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence; The heterogeneous directed snapshots in the heterogeneous directed snapshot sequence are link-completed according to the node characteristics to obtain a target heterogeneous directed snapshot sequence, and a reconstructed link sequence is generated according to the target heterogeneous directed snapshot sequence.

[0005] Optionally, constructing a global directed acyclic graph according to the link tracking data includes: Determine a plurality of link tracking sub-data in the link tracking data, and construct a directed acyclic graph corresponding to each link tracking sub-data; The directed acyclic graph corresponding to each link tracking sub-data is traversed, and the directed acyclic graph corresponding to each link tracking sub-data is merged according to the traversal result to obtain a global directed acyclic graph.

[0006] Optionally, the updating the global directed acyclic graph to a heterogeneous directed graph by using the aggregated metric data includes: Parsing the aggregated metric data, and determining a node representation vector corresponding to each global node in the global directed acyclic graph according to the parsing result; The node representation vector is configured for the global directed acyclic graph, and a heterogeneous directed graph is obtained according to the configuration result.

[0007] Optionally, determining a node representation vector corresponding to any global node in the global directed acyclic graph includes: Determine the target metric name, target timestamp, and target metric value corresponding to the target global node according to the parsing result; constructing a target semantic vector according to the target metric name, constructing a target time vector according to the target timestamp, and constructing a target vector according to the target metric value; The target semantic vector, the target time vector and the target vector are concatenated to obtain a node representation vector corresponding to the target global node.

[0008] Optionally, constructing a heterogeneous directed snapshot sequence according to the heterogeneous directed graph includes: Determine the sliding window of fixed time interval; The heterogeneous directed graph is divided based on the sliding window, and a plurality of heterogeneous directed snapshots are obtained according to the division results to form a heterogeneous directed snapshot sequence.

[0009] Optionally, determining a node representation corresponding to a node included in any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: Inputting the first heterogeneous directed snapshot into a heterogeneous dynamic directed graph model, wherein the heterogeneous dynamic directed graph model includes an intra-snapshot encoder and an inter-snapshot encoder; Using the intra-snapshot encoder to perform message passing processing and message aggregation processing on the first heterogeneous directional snapshot, and determining heterogeneous representations corresponding to nodes included in the first heterogeneous directional snapshot according to the processing results; The inter-snapshot encoder is used to perform time-dependent processing on the heterogeneous representations corresponding to the nodes included in the first heterogeneous directional snapshot to obtain the node representations corresponding to the nodes included in the first heterogeneous directional snapshot.

[0010] Optionally, link completion of any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: constructing a target directed edge according to the node representation corresponding to the node included in the second heterogeneous directed snapshot; Using a preset classifier to detect whether the second heterogeneous directed snapshot contains the target directed edge; If yes, link completion is performed on the second heterogeneous directed snapshot based on the target directed edge.

[0011] Optionally, generating a reconstruction link sequence according to the target heterogeneous directional snapshot sequence includes: Merging the target heterogeneous directed snapshots contained in the target heterogeneous directed snapshot sequence according to a preset link reconstruction algorithm to obtain a heterogeneous directed acyclic graph; Traversing the heterogeneous directed acyclic graph, and obtaining a plurality of connected subgraphs according to the traversal results; A reconstruction link sequence is constructed based on the multiple connected subgraphs.

[0012] According to a second aspect of an embodiment of this specification, a digital service network tracing link reconstruction device is provided, comprising: an acquisition module configured to acquire link tracking data and aggregated metric data of the digital service network, and construct a global directed acyclic graph based on the link tracking data; A construction module is configured to update the global directed acyclic graph into a heterogeneous directed graph using the aggregated metric data, and construct a heterogeneous directed snapshot sequence according to the heterogeneous directed graph; An input module is configured to input the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence into a heterogeneous dynamic directed graph model for processing, and obtain node representations corresponding to nodes included in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence; The generation module is configured to perform link completion on the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence according to the node characteristics, obtain a target heterogeneous directed snapshot sequence, and generate a reconstructed link sequence according to the target heterogeneous directed snapshot sequence.

[0013] According to a third aspect of an embodiment of this specification, a computing device is provided, including: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the digital service network tracing link reconstruction method are implemented.

[0014] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned digital service network tracing link reconstruction method are implemented.

[0015] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, including a computer program or instructions, which implement the steps of the above-mentioned digital service network tracing link reconstruction method when executed by a processor.

[0016] The present embodiment provides a digital service network tracking link reconstruction method. In order to improve the accuracy of call link reconstruction, link tracking data and aggregated measurement data of the digital service network can be first obtained. At this time, a global directed acyclic graph can be constructed according to the link tracking data; then the global directed acyclic graph can be updated to a heterogeneous directed graph using the aggregated measurement data, and a heterogeneous directed snapshot sequence can be constructed according to the heterogeneous directed graph; the heterogeneous directed snapshot sequence can reflect the link relationship at different times, and then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be input into a heterogeneous dynamic directed graph model for processing, so as to mine the relationship between nodes through the model, and then the node representation corresponding to the node included in the heterogeneous directed snapshot in the heterogeneous directed snapshot sequence can be obtained; then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be completed according to the node characteristics, and then the target heterogeneous directed snapshot sequence can be obtained, and finally the reconstructed link sequence can be generated according to the target heterogeneous directed snapshot sequence, so as to ensure the accuracy of call link reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of a digital service network tracing link reconstruction method provided by an embodiment of this specification; Figure 2 It is a schematic diagram of snapshot construction in a digital service network tracking link reconstruction method provided by an embodiment of this specification; Figure 3 It is a schematic diagram of model processing in a digital service network tracing link reconstruction method provided by an embodiment of this specification; Figure 4 It is a schematic diagram of link reconstruction in a digital service network tracing link reconstruction method provided by an embodiment of this specification; Figure 5 It is a structural diagram of a digital service network tracing link reconstruction device provided by an embodiment of this specification; Figure 6 It is a structural block diagram of a computing device provided by an embodiment of this specification. DETAILED DESCRIPTION

[0018] Many specific details are described in the following description to facilitate a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of this specification, so this specification is not limited to the specific implementation disclosed below.

[0019] The terms used in one or more embodiments of this specification are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of this specification. The singular forms of "a", "said" and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0020] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0021] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0022] In this specification, a digital service network tracing link reconstruction method is provided. This specification also relates to a digital service network tracing link reconstruction device, a computing device, a computer-readable storage medium and a computer program product, which are described in detail one by one in the following embodiments.

[0023] In practical applications, link reconstruction depends on link completion. Link completion is an important task in the field of graph data analysis. Its main goal is to predict missing links or edges in the existing graph structure to complete the topological structure of the graph. Call link completion is an application of link completion technology in distributed tracing of digital service networks, which aims to complete the missing microservice call link completion due to various reasons. Observability in digital service networks mainly includes three dimensions: link tracing, aggregation metrics, and logs. It is the basic requirement for the reliable operation of digital service networks and can solve the problem of monitoring the operation of digital service networks in highly distributed environments. Due to the large scale of service nodes and the complexity of call relationships in digital service networks, there are incomplete call chains obtained by using distributed tracing in digital service networks. Therefore, call link completion methods based on deep learning algorithms and statistical methods based on graph neural networks have emerged. Among them, the call link completion method based on graph neural networks models the tracing link as graph data, in which service instances are regarded as nodes and the call relationships between nodes are regarded as edges. Existing graph neural network-based call link completion methods are mainly performed in isomorphic graphs, and the potential call relationships between service nodes are completed by capturing and learning the complex relationships and patterns between service nodes. However, the accuracy is low, so an effective solution is urgently needed to solve the above problems.

[0024] In view of this, in order to improve the accuracy of call link reconstruction, the link tracking data and aggregated measurement data of the digital service network can be obtained first. At this time, a global directed acyclic graph can be constructed based on the link tracking data; then the aggregated measurement data can be used to update the global directed acyclic graph to a heterogeneous directed graph, and a heterogeneous directed snapshot sequence can be constructed according to the heterogeneous directed graph; the heterogeneous directed snapshot sequence can reflect the link relationship at different times, and then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be input into the heterogeneous dynamic directed graph model for processing, so as to mine the relationship between nodes through the model, and then the node representation corresponding to the nodes contained in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be obtained; then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be completed according to the node characteristics, and then the target heterogeneous directed snapshot sequence can be obtained, and finally the reconstructed link sequence can be generated according to the target heterogeneous directed snapshot sequence, so as to ensure the accuracy of call link reconstruction.

[0025] See also Figure 1 , Figure 1 A flowchart of a digital service network tracing link reconstruction method provided according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0026] Step S102, obtaining link tracking data and aggregated metric data of the digital service network, and constructing a global directed acyclic graph based on the link tracking data.

[0027] The digital service network tracking link reconstruction method provided in this embodiment is used to reconstruct the call link in the digital service network, and then provide users with stable digital service network services, such as online education, telemedicine, e-commerce, digital finance, etc. The digital service network can realize a network system that is interconnected and interwoven through digital technology and Internet platforms, providing users with a convenient, efficient and personalized service experience.

[0028] In specific implementation, the digital service network tracing link reconstruction method provided in this embodiment can be divided into three steps, namely: Step 1, context-enhanced heterogeneous directed snapshot sequence construction, which is used to obtain Trace data (link tracing data) and Metric data (aggregated metric data) from the digital service network to construct a heterogeneous directed graph with Span as nodes and Metric information as node features, and slice and divide the heterogeneous directed graph to obtain a heterogeneous directed snapshot sequence. Step 2, learning the spatiotemporal representation of Span nodes, which is used to learn the Metric representation and call relationship of Span nodes in heterogeneous directed snapshots by building a heterogeneous dynamic directed graph neural network model, and obtain the node-level representation of Span. Step 3, Trace link reconstruction, which is used to determine whether the target directed edge exists and perform link completion by learning the representation of each node, and then perform Trace reconstruction tasks on the heterogeneous snapshot sequence to generate a reconstructed link sequence.

[0029] Specifically, link tracking data refers to the trace data in the digital service network, and aggregated metric data refers to the metric data in the digital service network. Correspondingly, the global directed acyclic graph refers to the acyclic graph of the matching digital service network constructed in combination with the link tracking data, which is used for subsequent link reconstruction.

[0030] Based on this, in order to improve the accuracy of link reconstruction calls, the link tracking data and aggregated measurement data of the digital service network can be obtained first. At this time, a global directed acyclic graph can be constructed based on the link tracking data; then the aggregated measurement data can be used to update the global directed acyclic graph to a heterogeneous directed graph, and a heterogeneous directed snapshot sequence can be constructed according to the heterogeneous directed graph; the heterogeneous directed snapshot sequence can reflect the link relationship at different times, and then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be input into the heterogeneous dynamic directed graph model for processing, so as to mine the relationship between nodes through the model, and then the node representation corresponding to the nodes contained in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be obtained; then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be completed according to the node characteristics, and then the target heterogeneous directed snapshot sequence can be obtained, and finally the reconstructed link sequence can be generated according to the target heterogeneous directed snapshot sequence.

[0031] In one or more optional implementations of this embodiment, constructing a global directed acyclic graph according to the link tracking data includes: Determine multiple link tracking sub-data in the link tracking data, and construct a directed acyclic graph corresponding to each link tracking sub-data; traverse the directed acyclic graph corresponding to each link tracking sub-data, merge the directed acyclic graph corresponding to each link tracking sub-data according to the traversal result, and obtain a global directed acyclic graph.

[0032] Specifically, the link tracking sub-data specifically refers to any Trace data in the link tracking data. Based on this, when constructing a directed acyclic graph, considering that the link tracking data contains multiple link tracking sub-data, it is possible to construct directed acyclic graphs separately and then merge them to construct a global directed acyclic graph. Specifically, multiple link tracking sub-data can be determined in the link tracking data first, and a directed acyclic graph corresponding to each link tracking sub-data can be constructed; at this time, the directed acyclic graph corresponding to each link tracking sub-data can be traversed to merge the directed acyclic graph corresponding to each link tracking sub-data according to the traversal results, and then a global directed acyclic graph can be obtained for subsequent use.

[0033] In summary, by adopting the method of constructing directed acyclic graphs separately and merging them, it can be ensured that the global directed acyclic graph finally constructed is more compatible with the digital service network, thereby facilitating the subsequent link reconstruction and use.

[0034] Step S104: using the aggregated metric data to update the global directed acyclic graph into a heterogeneous directed graph, and constructing a heterogeneous directed snapshot sequence according to the heterogeneous directed graph.

[0035] Specifically, a heterogeneous directed graph refers to a directed graph obtained after configuring aggregated metric data in a global directed acyclic graph. Correspondingly, a heterogeneous directed snapshot sequence refers to a sequence of heterogeneous directed snapshots obtained after snapshot processing of a heterogeneous directed graph at a set time interval. Since the heterogeneous directed graph changes over time, each time node may correspond to a different structure. Therefore, snapshot processing can obtain multiple heterogeneous directed snapshots with different structures, thereby forming a sequence for subsequent reconstruction of the link.

[0036] Based on this, after obtaining the global directed acyclic graph, the global directed acyclic graph can be updated to a heterogeneous directed graph using aggregated measurement data, so as to configure the attribute information corresponding to the node in the heterogeneous directed graph. Then, a heterogeneous directed snapshot sequence can be constructed according to the heterogeneous directed graph, so that the link reconstruction process can be completed in the subsequent combination of link tracking data and aggregated measurement data.

[0037] In one or more optional implementations of this embodiment, the updating of the global directed acyclic graph into a heterogeneous directed graph by using the aggregated metric data includes: The aggregated metric data is parsed, and a node representation vector corresponding to each global node in the global directed acyclic graph is determined according to the parsing result; the node representation vector is configured for the global directed acyclic graph, and a heterogeneous directed graph is obtained according to the configuration result.

[0038] Specifically, the global node specifically refers to the node included in the global directed acyclic graph. Correspondingly, the node representation vector specifically refers to the representation obtained after vectorizing the data of the associated global node in the aggregated metric data. Based on this, in order to ensure that the heterogeneous directed graph fully reflects the relevant attributes of the current digital service network, the aggregated metric data can be parsed to determine the node representation vector corresponding to each global node in the global directed acyclic graph according to the parsing results; thereafter, the node representation vector can be configured for the global directed acyclic graph, so as to obtain a heterogeneous directed graph according to the configuration results for subsequent use.

[0039] In summary, by configuring node representation vectors for directed graphs, it is possible to assign attributes to directed graphs, thereby providing a basis for subsequent reconstruction of links.

[0040] In one or more optional implementations of this embodiment, determining a node representation vector corresponding to any global node in the global directed acyclic graph includes: Determine the target metric name, target timestamp and target metric value corresponding to the target global node according to the parsing result; construct a target semantic vector according to the target metric name, construct a target time vector according to the target timestamp, and construct a target vector according to the target metric value; splice the target semantic vector, the target time vector and the target vector to obtain a node representation vector corresponding to the target global node.

[0041] Specifically, the target metric name refers to the name information corresponding to the name field in the aggregated metric data, and the target timestamp and target metric value refer to the information corresponding to the first and second elements of each column element in the values ​​list in the aggregated metric data. Correspondingly, the target semantic vector refers to the vector expression generated after the semantic information of the target metric name is extracted. The target time vector refers to the vector expression obtained after the vector is constructed for the target timestamp, and the target vector refers to the vector expression directly constructed for the target metric value.

[0042] Based on this, for the node representation vector corresponding to any global node in the global directed acyclic graph, the target metric name, target timestamp and target metric value corresponding to the target global node can be determined according to the parsing results; thereafter, the target semantic vector can be constructed according to the target metric name, the target time vector can be constructed according to the target timestamp, and the target vector can be constructed according to the target metric value; on this basis, the target semantic vector, target time vector and target vector can be spliced ​​to obtain the node representation vector corresponding to the target global node, and so on, until the node representation vectors corresponding to all global nodes are constructed, and subsequent processing can be carried out.

[0043] In summary, by combining multi-dimensional information to construct a node representation vector, the attribute information corresponding to each global node can be more accurately reflected, which is convenient for subsequent use.

[0044] In one or more optional implementations of this embodiment, the step of constructing a heterogeneous directed snapshot sequence according to the heterogeneous directed graph includes: Determine a sliding window with a fixed time interval; divide the heterogeneous directed graph based on the sliding window, obtain multiple heterogeneous directed snapshots according to the division results, and form a heterogeneous directed snapshot sequence.

[0045] Specifically, the fixed time interval can be set according to actual needs, and this embodiment does not impose any limitation here, such as 1ms, 1s or 1min, etc. Based on this, after obtaining the heterogeneous directed graph, the sliding window of the fixed time interval can be determined; then the heterogeneous directed graph can be divided based on the sliding window, so that multiple heterogeneous directed snapshots can be obtained according to the division results and form a heterogeneous directed snapshot sequence.

[0046] In practical applications, such as Figure 2 As shown in the schematic diagram, the service instances in the digital service network are heterogeneous, so the call graph network generated by the Trace data should also show the characteristics of a heterogeneous graph. At the same time, the calling relationship of the service instances will change over time, making the Trace data significantly dynamic. The calling relationship between services not only reflects their association, but also reveals the interaction logic between services. When a service instance is called, its corresponding various metrics (Metric data) will change accordingly, reflecting the network operation status and containing rich contextual information. In order to effectively capture the heterogeneity and dynamics of Trace data, show the association between service instances, and make full use of Metric data to enrich the contextual information of service instances, a heterogeneous directed snapshot sequence construction module based on Metric data for context enhancement can be constructed. The snapshot sequence construction process is as follows: (1) Parse the link tracing data and generate a global directed acyclic graph. Specifically, each Span in the Trace and its corresponding operation name, IP address, call start time, duration, and call relationship can be extracted. Then, all service call relationships are parsed to form a directed acyclic graph HDAG. After obtaining the HDAG of each Trace, in order to find all the dependencies of the same nodes in different call chains, all HDAGs can be merged using an algorithm. The HDAG merging algorithm specifically refers to traversing the edges of the HDAGs to be merged, adding them to the HDAG that has been merged, and determining whether the updated HDAG has a loop within the time range of the newly added edge. If a loop occurs, it is considered that a circular call occurs within the time period and an exception is thrown; otherwise, the information of the start and end points of the edge is merged into the node corresponding to the HDAG. In this way, the merged global directed acyclic graph HDAG is obtained.

[0047] (2) Parse the aggregated metric data and complete the node representation injection enhancement. Specifically, parse the collected metric data, extract the "__name__" field in the metric data, and extract the first and second elements of each list element in the "values" value list to form a (metric name, timestamp, metric value) triple. Three methods are used to convert the three elements into vectors. The metric name can be used to extract its semantic information and generate a semantic vector using BERT. The timestamp can be used to generate a time vector using random Fourier feature transformation. The metric value can be directly used to generate a vector using its original value. After that, the three vectors are concatenated as the representation vector of the current node. Furthermore, by parsing the metric data, the corresponding relationship between the metric and the node on the global directed acyclic graph HDAG can be found through specific fields in the metric data. For example, when determining the indicator corresponding to the service in the current span, first find the node_id value in the tags data in the current span, determine the service instance corresponding to the sidecar container in the currently running pod, and then find the instance value and kubernetes_pod_name value from the metric data and match them with the node_id value. Finally, according to the startTime and duration of the span, the metric value of the corresponding timestamp is found in the metric data values ​​list, and the correspondence between a span and a metric can be completed. After determining the correspondence, the node representation vector information is injected into the global directed acyclic graph HDAG to complete the node representation mapping and injection. After all is completed, the snapshot sequence can be constructed.

[0048] (3) Heterogeneous directed snapshot sequence division. Specifically, after obtaining a heterogeneous directed graph with rich node information, a sliding window with a fixed time interval can be used to divide the heterogeneous directed graph into heterogeneous directed snapshot sequences. . This makes it easier to use later.

[0049] Step S106: input the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence into a heterogeneous dynamic directed graph model for processing, and obtain node representations corresponding to the nodes included in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence.

[0050] Specifically, the heterogeneous dynamic directed graph model specifically refers to a model that processes the nodes contained in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence and outputs the node representation corresponding to each node. It can complete the prediction of the node representation corresponding to any node by combining the dependency relationship between nodes in the snapshot and the dependency relationship between snapshots. Correspondingly, the node representation specifically refers to the vector expression corresponding to any node contained in any heterogeneous directed snapshot.

[0051] Based on this, after obtaining the heterogeneous directed snapshot sequence, the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be input into the heterogeneous dynamic directed graph model for processing, so as to realize the use of the model to output the node representation corresponding to the nodes contained in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence, thereby facilitating the subsequent link completion operation.

[0052] In one or more optional implementations of this embodiment, determining the node representation corresponding to the node included in any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: A first heterogeneous directed snapshot is input into a heterogeneous dynamic directed graph model, wherein the heterogeneous dynamic directed graph model includes an intra-snapshot encoder and an inter-snapshot encoder; the intra-snapshot encoder is used to perform message passing processing and message aggregation processing on the first heterogeneous directed snapshot, and the heterogeneous representations corresponding to the nodes included in the first heterogeneous directed snapshot are determined according to the processing results; the inter-snapshot encoder is used to perform time-dependent processing on the heterogeneous representations corresponding to the nodes included in the first heterogeneous directed snapshot, and node representations corresponding to the nodes included in the first heterogeneous directed snapshot are obtained.

[0053] Specifically, the intra-snapshot encoder specifically refers to an encoder that performs encoding in combination with attribute information corresponding to nodes in the heterogeneous directed snapshots. The inter-snapshot encoder specifically refers to an encoder that performs encoding in combination with time dependency information between heterogeneous directed snapshots.

[0054] Based on this, for the node representation corresponding to the node contained in any heterogeneous directed snapshot in the heterogeneous directed snapshot sequence, the first heterogeneous directed snapshot can be first input into the heterogeneous dynamic directed graph model, wherein the heterogeneous dynamic directed graph model includes an intra-snapshot encoder and an inter-snapshot encoder; thereafter, the intra-snapshot encoder can be used to perform message transmission processing and message aggregation processing on the first heterogeneous directed snapshot, so as to determine the heterogeneous representation corresponding to the node contained in the first heterogeneous directed snapshot according to the processing result; thereafter, the inter-snapshot encoder is used to perform time-dependent processing on the heterogeneous representation corresponding to the node contained in the first heterogeneous directed snapshot, so as to obtain the node representation corresponding to the node contained in the first heterogeneous directed snapshot. Similarly, after the node representations corresponding to all nodes are obtained, they can be used for subsequent link completion operations.

[0055] In practical applications, see Figure 3As shown in the schematic diagram, in the digital service network, the service instance nodes in the call chain have different attributes and functions. At the same time, the dynamic nature of the system operation makes the call chain have significant time dependencies, and the order and timing patterns of the calls are crucial to characterizing the system behavior. If the heterogeneous representations or time dependencies between nodes are ignored, it will be difficult to effectively model the contextual information of the nodes, and thus it will be impossible to accurately describe the operating logic and dynamic characteristics of the system. Therefore, a heterogeneous dynamic directed graph neural network can be used to achieve heterogeneous representation information between learning nodes and the time dependencies of the call chain and to enrich the node representation. The structure and specific process of the heterogeneous dynamic directed graph neural network are as follows: (1) Heterogeneous attention encoder within a snapshot. Specifically, for each snapshot, the heterogeneous representation of the span node is learned through two steps: message passing and message aggregation. Message passing: For the edge e = (u, v) on the kth layer, the multi-head message aggregation from node u to node v can be calculated according to the following formulas (1) and (2): (1) (2) in, For Node In the The node representation of the layer, Indicates the node type The linear mapping layer, is the number of longs.

[0056] Message aggregation: using Compute the snapshot on each source node The importance of neighbor node information is then aggregated to the target node , can be obtained by the following formula (3): (3) in, Representation Node The set of all neighbor nodes of .

[0057] (2) Transformer encoder between snapshots. Specifically, considering the temporal dependency between past and current call chains, the node embedding sequence obtained by the encoder within the snapshot is , input it into Transformer to extract time-dependent information .in, is the number of nodes in the snapshot. At this time, the node representation introduces the time dependency of the call chain. This enables the completion and reconstruction of the link based on the node representation.

[0058] Step S108 , completing the links of the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence according to the node characteristics, obtaining a target heterogeneous directed snapshot sequence, and generating a reconstructed link sequence according to the target heterogeneous directed snapshot sequence.

[0059] Specifically, link completion refers to the completion of missing edges of heterogeneous directed snapshots to ensure that the completed snapshots are more consistent with the actual situation. Accordingly, the target heterogeneous directed snapshot sequence refers to a sequence of new snapshots obtained after link completion of heterogeneous directed snapshots in the heterogeneous directed snapshot sequence. Accordingly, the reconstructed link sequence refers to a sequence of new links obtained after reconstructing the link tracking data related to the digital service network.

[0060] Based on this, when performing link reconstruction, the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be completed according to the node characteristics, so as to obtain the target heterogeneous directed snapshot sequence according to the completion result. Thereafter, the reconstructed link sequence can be generated according to the target heterogeneous directed snapshot sequence, so as to facilitate the use of downstream services.

[0061] In one or more optional implementations of this embodiment, link completion of any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: A target directed edge is constructed according to the node representation corresponding to the node included in the second heterogeneous directed snapshot; a preset classifier is used to detect whether the second heterogeneous directed snapshot includes the target directed edge; if so, link completion is performed on the second heterogeneous directed snapshot based on the target directed edge.

[0062] Specifically, the second heterogeneous directed snapshot specifically refers to any heterogeneous directed snapshot in the heterogeneous directed snapshot sequence. Accordingly, the target directed edge specifically refers to a possible edge added to the snapshot. Accordingly, the classifier specifically refers to a classifier for judging the target directed edge.

[0063] Based on this, when completing the link of any heterogeneous directed snapshot in the heterogeneous directed snapshot sequence, the target directed edge can be first constructed according to the node representation corresponding to the node contained in the second heterogeneous directed snapshot; at this time, the preset classifier can be used to detect whether the second heterogeneous directed snapshot contains the target directed edge; if so, it means that the currently constructed target directed edge is what the snapshot should have, so the link completion of the second heterogeneous directed snapshot can be performed based on the target directed edge. If not, it means that the edge does not exist, and the heterogeneous directed snapshot can be directly added to the target heterogeneous directed snapshot sequence.

[0064] In summary, by detecting each edge to be added, it can be ensured that the snapshots contained in the target heterogeneous directed snapshot sequence are more in line with the application scenario, thereby facilitating the subsequent reconstruction of the link.

[0065] In one or more optional implementations of this embodiment, generating a reconstruction link sequence according to the target heterogeneous directional snapshot sequence includes: The target heterogeneous directed snapshots contained in the target heterogeneous directed snapshot sequence are merged according to a preset link reconstruction algorithm to obtain a heterogeneous directed acyclic graph; the heterogeneous directed acyclic graph is traversed to obtain multiple connected subgraphs according to the traversal results; and a reconstructed link sequence is constructed based on the multiple connected subgraphs.

[0066] Specifically, the link reconstruction algorithm specifically refers to an algorithm for reconstructing a link according to a target heterogeneous directed snapshot sequence, which is the inverse of the algorithm for constructing the sequence. Correspondingly, the heterogeneous directed acyclic graph specifically refers to a directed acyclic graph constructed by combining the target heterogeneous directed snapshots contained in the target heterogeneous directed snapshot sequence.

[0067] Based on this, when performing link reconstruction, the target heterogeneous directed snapshots contained in the target heterogeneous directed snapshot sequence can be merged according to the preset link reconstruction algorithm to obtain a heterogeneous directed acyclic graph; thereafter, the heterogeneous directed acyclic graph can be traversed to obtain multiple connected subgraphs according to the traversal results; finally, the reconstructed link sequence can be constructed based on the multiple connected subgraphs.

[0068] In practical applications, such as Figure 4 As shown in the schematic diagram, after obtaining a snapshot sequence with rich node representation information, in order to achieve the link reconstruction task, it is necessary to detect the missing links and complete the edges that are judged to be missing, and then complete the missing edges to the corresponding heterogeneous snapshots and reconstruct the snapshot sequence into a complete Trace sequence. In order to achieve the above functions, a link completion classifier and a Trace reconstruction algorithm can be used to achieve the link reconstruction task, as follows: (1) Link completion. Specifically, after obtaining the representation of each node, the starting point and the end point of the target directed edge can be spliced, and the spliced ​​features obtained can be input into the link completion classifier to determine whether the target directed edge exists. This can be achieved by the following formula (4): (4) in, , It represents the starting node and the ending node of the target directed edge.

[0069] During the training process, a global random negative sample sampling strategy can be adopted, and the loss function is the cross entropy loss function, as shown in the following formula (5): (5) in, is the number of edges, It is a sample The real label, It is the model judgment result.

[0070] (2) Reconstruct the trace call chain. Specifically, first add the lost edges back into the heterogeneous snapshot to obtain a new heterogeneous snapshot sequence. , for the new snapshot sequence, use the Trace reconstruction algorithm to reconstruct the snapshot sequence into a complete Trace sequence. The reconstruction algorithm process is summarized as follows: First, the same HDAG merging algorithm mentioned above merges the heterogeneous snapshot sequence into a heterogeneous directed acyclic graph DAG'. For DAG', traverse each node without a parent node, traverse each time period information of the node, and obtain the maximum connected subgraph DAG'T of DAG' under the time period information. Convert the nodes of DAG'T into Span, and convert the connection relationship into a call relationship to obtain The final sequence is That is the reconstructed call chain sequence.

[0071] The present embodiment provides a digital service network tracking link reconstruction method. In order to improve the accuracy of call link reconstruction, link tracking data and aggregated measurement data of the digital service network can be first obtained. At this time, a global directed acyclic graph can be constructed according to the link tracking data; then the global directed acyclic graph can be updated to a heterogeneous directed graph using the aggregated measurement data, and a heterogeneous directed snapshot sequence can be constructed according to the heterogeneous directed graph; the heterogeneous directed snapshot sequence can reflect the link relationship at different times, and then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be input into a heterogeneous dynamic directed graph model for processing, so as to mine the relationship between nodes through the model, and then the node representation corresponding to the node included in the heterogeneous directed snapshot in the heterogeneous directed snapshot sequence can be obtained; then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be completed according to the node characteristics, and then the target heterogeneous directed snapshot sequence can be obtained, and finally the reconstructed link sequence can be generated according to the target heterogeneous directed snapshot sequence, so as to ensure the accuracy of call link reconstruction.

[0072] Corresponding to the above method embodiment, this specification also provides a digital service network tracking link reconstruction device embodiment, Figure 5 FIG. 1 is a schematic diagram showing the structure of a digital service network tracking link reconstruction device provided by an embodiment of the present specification. Figure 5 As shown, the device comprises: An acquisition module 502 is configured to acquire link tracking data and aggregated metric data of the digital service network, and construct a global directed acyclic graph according to the link tracking data; A construction module 504 is configured to update the global directed acyclic graph into a heterogeneous directed graph using the aggregated metric data, and construct a heterogeneous directed snapshot sequence according to the heterogeneous directed graph; An input module 506 is configured to input the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence into a heterogeneous dynamic directed graph model for processing, and obtain node representations corresponding to nodes included in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence; The generating module 508 is configured to perform link completion on the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence according to the node characteristics, obtain a target heterogeneous directed snapshot sequence, and generate a reconstructed link sequence according to the target heterogeneous directed snapshot sequence.

[0073] In an optional embodiment, the acquisition module 502 is further configured to: Determine multiple link tracking sub-data in the link tracking data, and construct a directed acyclic graph corresponding to each link tracking sub-data; traverse the directed acyclic graph corresponding to each link tracking sub-data, merge the directed acyclic graph corresponding to each link tracking sub-data according to the traversal result, and obtain a global directed acyclic graph.

[0074] In an optional embodiment, the construction module 504 is further configured to: The aggregated metric data is parsed, and a node representation vector corresponding to each global node in the global directed acyclic graph is determined according to the parsing result; the node representation vector is configured for the global directed acyclic graph, and a heterogeneous directed graph is obtained according to the configuration result.

[0075] In an optional embodiment, determining a node representation vector corresponding to any global node in the global directed acyclic graph includes: Determine the target metric name, target timestamp and target metric value corresponding to the target global node according to the parsing result; construct a target semantic vector according to the target metric name, construct a target time vector according to the target timestamp, and construct a target vector according to the target metric value; splice the target semantic vector, the target time vector and the target vector to obtain a node representation vector corresponding to the target global node.

[0076] In an optional embodiment, the construction module 504 is further configured to: Determine a sliding window with a fixed time interval; divide the heterogeneous directed graph based on the sliding window, obtain multiple heterogeneous directed snapshots according to the division results, and form a heterogeneous directed snapshot sequence.

[0077] In an optional embodiment, determining the node representation corresponding to the node included in any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: A first heterogeneous directed snapshot is input into a heterogeneous dynamic directed graph model, wherein the heterogeneous dynamic directed graph model includes an intra-snapshot encoder and an inter-snapshot encoder; the intra-snapshot encoder is used to perform message passing processing and message aggregation processing on the first heterogeneous directed snapshot, and the heterogeneous representations corresponding to the nodes included in the first heterogeneous directed snapshot are determined according to the processing results; the inter-snapshot encoder is used to perform time-dependent processing on the heterogeneous representations corresponding to the nodes included in the first heterogeneous directed snapshot, and node representations corresponding to the nodes included in the first heterogeneous directed snapshot are obtained.

[0078] In an optional embodiment, link completion of any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: A target directed edge is constructed according to the node representation corresponding to the node included in the second heterogeneous directed snapshot; a preset classifier is used to detect whether the second heterogeneous directed snapshot includes the target directed edge; if so, link completion is performed on the second heterogeneous directed snapshot based on the target directed edge.

[0079] In an optional embodiment, the generating module 508 is further configured to: The target heterogeneous directed snapshots contained in the target heterogeneous directed snapshot sequence are merged according to a preset link reconstruction algorithm to obtain a heterogeneous directed acyclic graph; the heterogeneous directed acyclic graph is traversed to obtain multiple connected subgraphs according to the traversal results; and a reconstructed link sequence is constructed based on the multiple connected subgraphs.

[0080] The present embodiment provides a digital service network tracking link reconstruction device. In order to improve the accuracy of link reconstruction, the link tracking data and aggregated measurement data of the digital service network can be obtained first. At this time, a global directed acyclic graph can be constructed according to the link tracking data; then the global directed acyclic graph can be updated to a heterogeneous directed graph using the aggregated measurement data, and a heterogeneous directed snapshot sequence can be constructed according to the heterogeneous directed graph; the heterogeneous directed snapshot sequence can reflect the link relationship at different times, and then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be input into a heterogeneous dynamic directed graph model for processing, so as to mine the relationship between nodes through the model, and then the node representation corresponding to the node included in the heterogeneous directed snapshot in the heterogeneous directed snapshot sequence can be obtained; then the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence can be completed according to the node characteristics, and then the target heterogeneous directed snapshot sequence can be obtained, and finally the reconstructed link sequence can be generated according to the target heterogeneous directed snapshot sequence, so as to ensure the accuracy of link reconstruction.

[0081] The above is a schematic scheme of a digital service network tracing link reconstruction device of this embodiment. It should be noted that the technical scheme of the digital service network tracing link reconstruction device and the technical scheme of the digital service network tracing link reconstruction method described above belong to the same concept, and the details not described in detail in the technical scheme of the digital service network tracing link reconstruction device can be referred to the description of the technical scheme of the digital service network tracing link reconstruction method described above.

[0082] Figure 6 The block diagram of a computing device 600 according to an embodiment of the present specification is shown. The components of the computing device 600 include but are not limited to a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and the database 650 is used to store data.

[0083] The computing device 600 also includes an access device 640 that enables the computing device 600 to communicate via one or more networks 660. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 640 may include one or more of any type of network interface (e.g., a network interface card (NIC)) that is wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a world-wide interoperability for microwave access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, and a near field communication (NFC).

[0084] In one embodiment of the present specification, the above components of the computing device 600 and Figure 6 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 6 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0085] The computing device 600 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). The computing device 600 may also be a mobile or stationary server.

[0086] The processor 620 is used to execute the following computer executable instructions, which, when executed by the processor, implement the steps of the above-mentioned digital service network tracing link reconstruction method.

[0087] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the digital service network tracing link reconstruction method described above belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the digital service network tracing link reconstruction method described above.

[0088] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned digital service network tracing link reconstruction method.

[0089] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the digital service network tracking link reconstruction method described above belong to the same concept, and the details not described in detail in the technical scheme of the storage medium can be referred to the description of the technical scheme of the digital service network tracking link reconstruction method described above.

[0090] An embodiment of the present specification further provides a computer program product, including a computer program or instructions, which implements the steps of the above-mentioned digital service network tracing link reconstruction method when executed by a processor.

[0091] The above is a schematic scheme of a computer program product of this embodiment. It should be noted that the technical scheme of the computer program product and the technical scheme of the digital service network tracing link reconstruction method described above belong to the same concept, and the details not described in detail in the technical scheme of the computer program product can be referred to the description of the technical scheme of the digital service network tracing link reconstruction method described above.

[0092] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0093] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0094] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0095] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0096] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The optional embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that technicians in the relevant technical field can understand and use this specification well.

Claims

1. A digital service network tracking link reconstruction method, characterized in that: include: Obtain link tracking data and aggregated metric data of the digital service network, and construct a global directed acyclic graph based on the link tracking data; Using the aggregated metric data, the global directed acyclic graph is updated to a heterogeneous directed graph, and a heterogeneous directed snapshot sequence is constructed according to the heterogeneous directed graph; Inputting the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence into a heterogeneous dynamic directed graph model for processing to obtain node representations corresponding to nodes included in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence; The heterogeneous directed snapshots in the heterogeneous directed snapshot sequence are link-completed according to the node characteristics to obtain a target heterogeneous directed snapshot sequence, and a reconstructed link sequence is generated according to the target heterogeneous directed snapshot sequence.

2. The method according to claim 1, characterized in that The constructing a global directed acyclic graph according to the link tracking data includes: Determine a plurality of link tracking sub-data in the link tracking data, and construct a directed acyclic graph corresponding to each link tracking sub-data; The directed acyclic graph corresponding to each link tracking sub-data is traversed, and the directed acyclic graph corresponding to each link tracking sub-data is merged according to the traversal result to obtain a global directed acyclic graph.

3. The method according to claim 1, characterized in that The updating the global directed acyclic graph to a heterogeneous directed graph by using the aggregated metric data includes: Parsing the aggregated metric data, and determining a node representation vector corresponding to each global node in the global directed acyclic graph according to the parsing result; The node representation vector is configured for the global directed acyclic graph, and a heterogeneous directed graph is obtained according to the configuration result.

4. The method according to claim 3, characterized in that Determining a node representation vector corresponding to any global node in the global directed acyclic graph includes: Determine the target metric name, target timestamp, and target metric value corresponding to the target global node according to the parsing result; constructing a target semantic vector according to the target metric name, constructing a target time vector according to the target timestamp, and constructing a target vector according to the target metric value; The target semantic vector, the target time vector and the target vector are concatenated to obtain a node representation vector corresponding to the target global node.

5. The method according to claim 1, characterized in that The step of constructing a heterogeneous directed snapshot sequence according to the heterogeneous directed graph includes: Determine the sliding window of fixed time interval; The heterogeneous directed graph is divided based on the sliding window, and a plurality of heterogeneous directed snapshots are obtained according to the division results to form a heterogeneous directed snapshot sequence.

6. The method according to claim 1, characterized in that Determining the node representation corresponding to the node included in any heterogeneous directed snapshot in the heterogeneous directed snapshot sequence includes: Inputting the first heterogeneous directed snapshot into a heterogeneous dynamic directed graph model, wherein the heterogeneous dynamic directed graph model includes an intra-snapshot encoder and an inter-snapshot encoder; Using the intra-snapshot encoder to perform message passing processing and message aggregation processing on the first heterogeneous directional snapshot, and determining heterogeneous representations corresponding to nodes included in the first heterogeneous directional snapshot according to the processing results; The inter-snapshot encoder is used to perform time-dependent processing on the heterogeneous representations corresponding to the nodes included in the first heterogeneous directional snapshot to obtain the node representations corresponding to the nodes included in the first heterogeneous directional snapshot.

7. The method according to claim 1, characterized in that The link completion of any heterogeneous directional snapshot in the heterogeneous directional snapshot sequence includes: constructing a target directed edge according to the node representation corresponding to the node included in the second heterogeneous directed snapshot; Using a preset classifier to detect whether the second heterogeneous directed snapshot contains the target directed edge; If yes, link completion is performed on the second heterogeneous directed snapshot based on the target directed edge.

8. The method according to claim 1, characterized in that The generating a reconstruction link sequence according to the target heterogeneous directional snapshot sequence includes: Merging the target heterogeneous directed snapshots contained in the target heterogeneous directed snapshot sequence according to a preset link reconstruction algorithm to obtain a heterogeneous directed acyclic graph; Traversing the heterogeneous directed acyclic graph, and obtaining a plurality of connected subgraphs according to the traversal results; A reconstruction link sequence is constructed based on the multiple connected subgraphs.

9. A digital service network tracking link reconstruction device, characterized in that: include: an acquisition module configured to acquire link tracking data and aggregated metric data of the digital service network, and construct a global directed acyclic graph based on the link tracking data; A construction module is configured to update the global directed acyclic graph into a heterogeneous directed graph using the aggregated metric data, and construct a heterogeneous directed snapshot sequence according to the heterogeneous directed graph; An input module is configured to input the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence into a heterogeneous dynamic directed graph model for processing, and obtain node representations corresponding to nodes included in the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence; The generation module is configured to perform link completion on the heterogeneous directed snapshots in the heterogeneous directed snapshot sequence according to the node characteristics, obtain a target heterogeneous directed snapshot sequence, and generate a reconstructed link sequence according to the target heterogeneous directed snapshot sequence.

10. A computing device, characterized in that: include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method described in any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium, characterized in that: It stores computer executable instructions, which, when executed by a processor, implement the steps of the method described in any one of claims 1 to 8.

12. A computer program product, characterized in that The method comprises a computer program or an instruction, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 8.

Images