WO2021208174A1

WO2021208174A1 - Distributed-type graph computation method, terminal, system, and storage medium

Info

Publication number: WO2021208174A1
Application number: PCT/CN2020/090238
Authority: WO
Inventors: 华井雅俊; 泽奥多洛保罗斯乔治斯
Original assignee: 南方科技大学
Priority date: 2020-04-16
Filing date: 2020-05-14
Publication date: 2021-10-21
Also published as: CN111581443B; CN111581443A

Abstract

A distributed-type graph computation method, a terminal, a system, and a storage medium. On the basis of pre-processing and graph partitioning algorithms for graph data as well as an incremental graph edge sorting algorithm, original graph data is converted into computer-readable intermediate graph data, which allows rapid execution of subsequent graph partitioning, and high quality graph partitioning is further provided, with generated high quality partitions significantly reducing communication overhead, quickening distributed graph computation and analysis speeds.

Description

Distributed graph computing method, terminal, system and storage medium

Technical field

The embodiments of the present application relate to, but are not limited to, the field of graph computing technology, and in particular, to a distributed graph computing method, terminal, system, and storage medium.

Background technique

As the demand for data analysis continues to grow, such as in-depth mining of data relationships, large-scale graph computing has received widespread attention in many fields. Graph is an abstract data structure used to represent the association relationship between objects. It is described by using vertices (Vertex) and edges (Edge). The vertices represent objects, and the edges represent the relationships between objects. Based on this, data that can be abstracted into graphs is graph data. Graph computing is a process in which graphs are used as data models to express and solve problems.

At present, as the scale of graphs continues to grow, distributed computing is used to analyze large-scale graph data. When using distributed graph computing, the large-scale graph is divided into several subgraphs, and multiple slave nodes are used for calculation, which can effectively utilize multiple computing resources. However, when performing distributed graph calculations, high-quality partitioning methods consume a lot of time during the calculation, which leads to higher energy consumption in the partitioning phase. In contrast, high-speed generation of partitions will result in low-quality partitions, causing serious performance losses.

Summary of the invention

The following is an overview of the topics detailed in this article. This summary is not intended to limit the scope of protection of the claims.

The embodiment of the application provides a distributed graph calculation method, terminal, system and storage medium, which uses graph data preprocessing, and when large-scale graph data analysis is performed, the graph data is only transmitted once, which can segment the graph with high quality and efficiency. Data, increase the speed of distributed graph calculation and reduce energy consumption.

In the first aspect, an embodiment of the present application provides a distributed graph calculation method, including:

Obtain the data of the first image;

Obtain a first intermediate preprocessing image according to the image preprocessing algorithm and the first image data;

According to the graph division algorithm, the distributed architecture and the first intermediate preprocessing graph, the first division graph is obtained;

According to the first division graph, first distributed graph analysis data is obtained.

Specifically, the distributed graph calculation method further includes:

Obtain the data of the second image;

If the first graph data is the same as the second graph data, a second division graph is obtained according to the graph division algorithm, the distributed architecture, and the first intermediate preprocessing graph;

According to the second division graph, a second distributed graph analysis data is obtained.

Specifically, the distributed graph calculation method further includes:

Obtain the data of the second image;

If the first graph data is not the same as the second graph data, obtaining difference data between the second graph data and the first graph data;

Obtaining a second intermediate preprocessing map according to the first division map, the difference data and the difference map preprocessing algorithm;

According to the graph division algorithm, the distributed architecture and the second intermediate preprocessing graph, a third division graph is obtained;

According to the third division graph, a third distributed graph analysis data is obtained.

Specifically, the difference map preprocessing algorithm includes:

If the difference data between the second graph data and the first graph data is incremental data, a second intermediate preprocessed graph is obtained according to the difference data and the incremental graph preprocessing algorithm.

Specifically, the incremental graph preprocessing algorithm further includes:

Obtain the adjacent edges between the start vertex and the end vertex according to the second graph data;

According to the incremental graph edge sorting algorithm and the adjacent edges between the start vertex and the end vertex, an incremental graph edge sort is obtained.

Specifically, the incremental graph edge sorting algorithm is applied to the main computing node, and the incremental graph edge sorting algorithm includes:

Sending the second graph data to the first subordinate computing node;

Acquiring a local solution sent by the first subordinate computing node;

According to the local solution, an optimized solution is obtained;

According to the optimized solution, a local optimized solution is obtained;

Sending the local optimization solution to the first subordinate computing node.

Specifically, the difference map preprocessing algorithm further includes:

If the difference data between the second graph data and the first graph data is decrement data, remove the decrement data;

According to the first graph data, the second graph data after removing the decremented data and the graph preprocessing algorithm obtain a second intermediate preprocessed graph.

Specifically, the distributed graph preprocessing algorithm further includes preprocessing graph edge sorting;

The preprocessing graph edge sorting includes:

Obtaining edge data of the first graph data and vertex data of the first graph data according to the first graph data;

According to the edge data of the first graph data and the vertex data of the first graph data, the first intermediate preprocessed graph is obtained.

Specifically, the obtaining the first intermediate preprocessing graph according to the edge data and the vertex data includes:

Acquiring first vertex data of the first graph data;

According to the edge sorting of the preprocessing graph, the priority queue is obtained;

Obtain the first intermediate preprocessing map according to the breadth first search and the priority queue.

Specifically, the first intermediate preprocessing diagram includes:

The starting vertex ID of the edge and the ending vertex ID of the edge are stored in binary format.

Specifically, the graph division algorithm includes:

Acquiring nodes and node configuration information of the distributed architecture, where the node configuration information includes one or more of the number of nodes, node specifications, and node performance;

Acquiring the first intermediate preprocessing map;

Obtaining the edge data of the first intermediate preprocessing graph according to the first intermediate preprocessing graph;

Obtaining the first partition graph according to the node configuration information of the distributed architecture and the edge data of the first intermediate preprocessing graph;

Send the first partition graph to the nodes of the distributed architecture.

In a second aspect, an embodiment of the present application provides a terminal, including: a first memory, a first processor, and a computer program stored on the first memory and running on the first processor, the first processor Realize when executing the program:

The distributed graph calculation method as described in the first aspect.

In the third aspect, an embodiment of the present application provides a distributed graph computing system, including a first distributed computing device and a second distributed computing device;

The first distributed computing device includes: a second memory, a second processor, and a first computer program that is stored on the second memory and can run on the second processor; the second processor executes the first computer program A computer program is implemented: the distributed graph calculation method described in the first aspect;

The second distributed computing device includes: a third memory, a third processor, and a second computer program that is stored on the third memory and can run on the third processor; the third processor executes the first The computer program implements the distributed graph calculation method described in the first aspect.

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium that stores computer-executable instructions, and the computer-executable instructions are used to:

Perform the distributed graph calculation method described in the first aspect.

The embodiment of the application converts the original graph data into computer-readable intermediate graph data based on the preprocessing of graph data, graph division algorithm and incremental graph edge sorting algorithm, which enables the subsequent graph division to be carried out quickly, and also provides High-quality graph partitioning greatly reduces communication overhead and speeds up the calculation and analysis of distributed graphs.

Other features and advantages of the present application will be described in the following description, and partly become obvious from the description, or understood by implementing the present application. The purpose and other advantages of the application can be realized and obtained through the structures specifically pointed out in the description, claims and drawings.

Description of the drawings

The accompanying drawings are used to provide a further understanding of the technical solution of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present application, and do not constitute a limitation to the technical solution of the present application.

FIG. 1 is a schematic flowchart of a distributed graph calculation method provided by an embodiment of the application;

2 is a schematic flowchart of a distributed graph calculation method provided by another embodiment of the application;

FIG. 3 is a schematic flowchart of a distributed graph computing method provided by another embodiment of this application;

FIG. 4 is a schematic structural diagram of a graph partition algorithm provided by an embodiment of this application;

FIG. 5 is a schematic structural diagram of an incremental graph edge sorting algorithm provided by an embodiment of the application.

Detailed ways

In order to make the purpose, technical solutions, and advantages of this application clearer and clearer, the following further describes the application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the application, and are not used to limit the application.

It should be noted that although functional modules are divided in the schematic diagram of the device, and the logical sequence is shown in the flowchart, in some cases, it can be performed in a different order from the module division in the device or the sequence in the flowchart. Steps shown or described. The terms "first", "second", etc. in the specification and claims and the above-mentioned drawings are used to distinguish similar objects, and not necessarily used to describe a specific sequence or sequence.

In the distributed graph computing technology of known technology, graph is a basic and ubiquitous abstract concept, which is widely used in modeling various problems in the real world. For example, in an online social network service, the vertices in the graph represent users, and the edges represent friendship relationships between users; in e-commerce services, the vertices represent users and products, and the edges represent purchase history. In the real world, graph data has been growing naturally. For example, one of the world's largest online social network services already contains about one trillion friendships. For such a large-scale graph, it is an important method to use multiple computing resources (such as high-performance computing platforms, cloud computing, etc.) to analyze the graph and to have a deep understanding of its characteristics (ie, distributed graph computing). Analyzing large-scale graph data through distributed graph computing is usually a time-consuming and expensive task.

Based on this, the embodiments of the present application provide a distributed graph computing method, terminal, system, and storage medium, which can convert original graph data into computer-readable intermediate graph data, which enables the subsequent graph division to be carried out quickly, while also providing The high-quality graph partition is created, and the generated high-quality partition greatly reduces the communication overhead, and accelerates the calculation and analysis speed of the distributed graph.

It should be noted that, in the following various embodiments, the terminal may be a mobile terminal device or a non-mobile terminal device. Mobile terminal devices can be mobile phones, tablet computers, laptops, palmtop computers, vehicle-mounted terminal devices, wearable devices, ultra-mobile personal computers, netbooks, digital cameras, video cameras or personal digital assistants, etc.; non-mobile terminal devices can be personal computers, Workstations, servers, televisions, teller machines, self-service machines, surveillance cameras or box cameras, etc.

The embodiments of the present application will be further described below in conjunction with the accompanying drawings.

An embodiment of the present application discloses a distributed graph calculation method.

Fig. 1 is a flowchart of a distributed graph calculation method. The calculation method shown in Fig. 1 at least includes the following steps:

Step S100: Obtain the data of the first image;

Step S101: graph preprocessing algorithm;

Step S102: Obtain a first intermediate preprocessing map;

Step S103: Obtain distributed architecture information;

Step S104: graph partition algorithm;

Step S105: the first division map;

Step S106: Distributed graph analysis.

In one embodiment, after obtaining the first graph data, the graph preprocessing algorithm is used to form the first intermediate preprocessing graph. The graph preprocessing algorithm uses graph edge sorting and converts the first intermediate preprocessing graph into a computer-readable binary graph format. The graph partition algorithm is applied to skip the redundant data scrambling of the graph elements to use the first intermediate preprocessing graph, and combine the distributed architecture information to generate the first partition graph. Performing distributed graph analysis according to the first division graph can obtain high-quality and high-efficiency distributed graph calculations.

In an embodiment, the first intermediate preprocessing image is in a computer-readable binary image format. Store the results continuously by using a continuous binary format. Each box unit represents a 32-bit or 64-bit integer. Every two boxes store the starting vertex ID and ending vertex ID of the edge. The graph data can be read from the machine without any communication overhead.

In one embodiment, the graph preprocessing algorithm converts the first graph data into a first intermediate preprocess graph. First, convert the first image data into the first intermediate preprocessing image. The first intermediate preprocessing graph is a computer-readable binary graph format and is expressed as an edge sequence. The expression of the conversion algorithm is:

{E ^φ [0],E ^φ [1],…E ^φ [|E|-1]},

Where E ^φ is an edge sequence E sorted by the sorting function φ:E→N.

In an embodiment, the expression of the graph preprocessing algorithm is:

Where V(E) is a set of vertices of edge E.

Fig. 2 is another flowchart of the distributed graph calculation method. The calculation method shown in Fig. 2 includes at least the following steps:

Step S200: Obtain the second image data;

Step S201: Compare the data of the second image with the data of the first image;

Step S202: the second image data is the same as the first image data;

Step S203: Obtain distributed architecture information;

Step S204: the first intermediate preprocessing map;

Step S205: graph partition algorithm;

Step S206: the second division map;

Step S207: Distributed graph analysis.

In one embodiment, after obtaining the data of the second graph, it is compared with the data of the first graph. When the data in the second graph is the same as the data in the first graph, the first intermediate preprocessing graph is used for analysis. The first intermediate preprocessing graph may be in a computer-readable binary graph format. The graph partition algorithm is applied to skip the redundant data scrambling of the graph elements to use the first intermediate preprocessing graph and combine the distributed architecture information to generate the second partition graph. Performing distributed graph analysis according to the second division graph can obtain high-quality and high-efficiency distributed graph calculations. In the calculation process, there is no need to repeat the data preprocessing process, which improves the efficiency of the calculation.

Fig. 3 is another flowchart of the distributed graph calculation method. The calculation method shown in Fig. 3 includes at least the following steps:

Step S300: the data of the second picture is different from the data of the first picture;

Step S301: Obtain a first division map;

Step S302: graph preprocessing algorithm;

Step S303: the second intermediate processing diagram;

Step S304: Obtain distributed architecture information;

Step S305: Obtain the change data of the first image data;

Step S306: graph partition algorithm;

Step S307: the third division map.

In one embodiment, if the first graph data is different from the second graph data, the second intermediate preprocessing graph is obtained according to the first division graph, the change data of the first graph data and the graph preprocessing algorithm. The second intermediate preprocessing image is a computer-readable binary image format. The graph partition algorithm is applied to skip the redundant data scrambling of the graph elements to use the second intermediate preprocessing graph and combine the distributed architecture information to generate the third partition graph.

In an embodiment, taking an e-commerce recommendation system as an example, the first image data includes users, products, and purchase history. The users and products are represented by the vertices of the graph, and the purchase history is represented by the edges. The graph preprocessing algorithm converts the first graph data into the second intermediate processing graph, so that the graph division algorithm can immediately generate high-quality divisions. After that, perform distributed graph analysis. For example, discovering user preferences and predicting products that may be purchased, so as to make corresponding recommendations. Due to the purchase history, new users and new product updates, the graph data will change periodically, so repeated analysis is required.

In one embodiment, the vertex data of at least one first graph data is obtained, the priority queue is obtained according to the graph edge sorting algorithm, and the first intermediate preprocessing graph is obtained according to the breadth first search (BFS) and the priority queue . Priority queue sorting is required before breadth-first search. The expression of the priority queue sorting on the edge of the graph is:

p(v):=|E|·D[v]-M[v],

Among them, D[v] is the number of unvisited vertices of v in the breadth-first search process; M[v] is the order of the largest edge among the adjacent edges of v during BFS (if the edges are not already sorted, then M[v] is 0). Based on p, the vertices are sorted in ascending order.

In an embodiment, the graph preprocessing algorithm includes an incremental graph preprocessing algorithm. Incremental graph preprocessing algorithms include incremental graph edge sorting algorithms.

Figure 4 is a structural diagram of the graph partitioning algorithm. As shown in Figure 4, the computing node obtains the number of broadcast edges from the distributed file system through the network and obtains the node configuration from the infrastructure. According to the number of edges and node configuration, each node finds a cross pointer to determine the starting point and ending point for dividing the graph data. The pointer is transferred to the file system via the network. After that, the distributed file system divides the edge into multiple partitions and sends these partitions back to the computing node. Efficiently forward partitions by dividing data into blocks. Finally, each node obtains the partition before starting the distributed graph calculation. In the existing method, the huge entire graph data is transmitted twice via the network. However, the method of the present application only transmits the graph data once, because it can calculate the partition using only metadata (ie, the number of edges and node configuration). Therefore, communication overhead can be saved and the work efficiency of each node in distributed graph calculation can be improved.

In one embodiment, the graph partition algorithm needs to use a distributed file system, the graph partition becomes faster, node configuration information, compute nodes, calculate split pointers, and obtain partitions. The graph partition algorithm obtains the forward pointer and the forward chunk through the network broadcast edge number during calculation.

In an embodiment, the node configuration information includes, for example, the number of CPUs, CPU specifications, memory size, network performance, node reliability, and so on.

In an embodiment, by calculating split vertices, the edge sequence is divided, so that the workload of each node in the process of distributed graph calculation and analysis is balanced.

In one embodiment, the graph partitioning algorithm is executed on the cloud infrastructure. The computing node is a virtual machine, and the network is a virtual network. Distributed file systems are usually located in different clusters or data centers. Therefore, the delay and bandwidth of the network are usually limited. The algorithm obtains the node configuration of the virtual host, and the node configuration of each virtual host may be different. Each node takes into account the differences in specifications, and splits the data in such a way that in the process of distributed graph analysis, the workload among the virtual hosts becomes balanced. The efficiency of moving large graph data from the file system to the virtual host is improved.

In one embodiment, when the distributed graph computing method is deployed on a public cloud, the computing power therein is delivered in the pay-as-you-go model, saving computing power directly reduces the payment cost of graph analysis.

In one embodiment, when the distributed graph computing method is used on a private cluster, the graph data only needs to be transmitted twice, which will result in a private cluster. The use of this distributed graph computing method can reduce energy costs, so that the graph data can be compared. Perform a more economical analysis.

In an embodiment, the distributed graph calculation method can be used in page ranking (PageRank) calculation, because more iterations can be performed, so that a more accurate ranking can be obtained.

In an embodiment, the distributed graph calculation method can be used in top-k type algorithms (such as top-k similarity analysis or top-k graph pattern matching), and more results can be obtained (k can be increased) .

In an embodiment, the distributed graph computing method can be used in graph-based machine learning. Since the distributed graph calculation method can obtain calculation results more quickly, more time can be used in the learning phase in the process of machine learning, and the prediction task will become more accurate.

In one embodiment, the distributed graph computing method enables real-time analysis and data-driven analysis. Make graph analysis more interactive.

Figure 5 is a schematic diagram of the structure of the incremental graph edge sorting algorithm. As shown in Figure 5, the incremental graph edge sorting algorithm is implemented in a distributed computing manner. An embodiment of the incremental graph edge sorting algorithm uses a master-slave architecture, which includes a master computing node and a slave computing node. First, the changed graph data is broadcast to the subordinate computing nodes. Second, each local optimal search algorithm obtains changed graph data and sorted partitions in its nodes. The algorithm calculates the approximate solution of the optimization problem of the partitioned graph locally and in parallel. Third, the main computing node collects local solutions and calculates optimized solutions. Finally, the optimized local solution is broadcast to the slave node, so that the slave node obtains the local optimal order with the smallest increment of the objective function.

After obtaining the graph difference data in an embodiment, the master node distributes the graph difference data to the slave nodes. The slave node obtains the local optimal solution according to the partition graph preprocessed in the last iteration and the local optimal solution search algorithm, and sends the local optimal solution to the master node. After the master node collects the local optimal solution, it calculates the optimal solution, then uses the optimal solution to calculate the local optimal solution, and sends the local optimal solution to the slave computing node.

In one embodiment, if the change data of the data in the first image is data removal, the subsequent calculation process is performed after the data is removed.

In one embodiment, if the change data of the first graph data is data increase, an incremental graph preprocessing algorithm is used for calculation. The expression of the incremental graph preprocessing algorithm is:

in:

Due to the large amount of data in the graph, it takes a long time to calculate a new sort from scratch. The incremental graph preprocessing algorithm only processes a part of the entire graph, that is, only scans the adjacent edges of the starting vertex and the ending vertex of the new edge. Then, a new edge order is calculated to minimize the increment of the objective function. Using the incremental graph preprocessing algorithm can reduce the complexity of the calculation when the first graph data is updated, thereby reducing energy consumption.

In an embodiment, the present application provides a terminal for executing a distributed graph calculation method.

In an embodiment, the present application provides a distributed graph computing system for executing a distributed graph computing method.

In an embodiment, the present application provides a computer-readable medium for executing a distributed graph computing method.

The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

A person of ordinary skill in the art can understand that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit . Such software may be distributed on a computer-readable medium, and the computer-readable medium may include a computer storage medium (or non-transitory medium) and a communication medium (or transitory medium). As is well known to those of ordinary skill in the art, the term computer storage medium includes volatile and non-volatile data implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Sexual, removable and non-removable media. Computer storage media include but are not limited to RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or Any other medium used to store desired information and that can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, communication media usually contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as carrier waves or other transmission mechanisms, and may include any information delivery media. .

The above is a detailed description of the preferred implementation of the application, but the application is not limited to the above-mentioned embodiments. Those skilled in the art can also make various equivalent modifications or substitutions without departing from the spirit of the application. Equivalent modifications or replacements are all included in the scope defined by the claims of this application.

Claims

A distributed graph calculation method, including:

Obtain the data of the first image;

Obtain a first intermediate preprocessing image according to the image preprocessing algorithm and the first image data;

According to the graph division algorithm, the distributed architecture and the first intermediate preprocessing graph, the first division graph is obtained;

According to the first division graph, first distributed graph analysis data is obtained.
The distributed graph calculation method according to claim 1, further comprising:

Obtain the data of the second image;

If the first graph data is the same as the second graph data, a second division graph is obtained according to the graph division algorithm, the distributed architecture, and the first intermediate preprocessing graph;

According to the second division graph, a second distributed graph analysis data is obtained.
The distributed graph calculation method according to claim 1, further comprising:

Obtain the data of the second image;

If the first graph data is not the same as the second graph data, obtaining difference data between the second graph data and the first graph data;

Obtaining a second intermediate preprocessing map according to the first division map, the difference data and the difference map preprocessing algorithm;

According to the graph division algorithm, the distributed architecture and the second intermediate preprocessing graph, a third division graph is obtained;

According to the third division graph, a third distributed graph analysis data is obtained.
The distributed graph calculation method according to claim 3, wherein the difference graph preprocessing algorithm comprises:

If the difference data between the second graph data and the first graph data is incremental data, a second intermediate preprocessed graph is obtained according to the difference data and the incremental graph preprocessing algorithm.
The distributed graph calculation method according to claim 4, wherein the incremental graph preprocessing algorithm further comprises:

Obtain the adjacent edges between the start vertex and the end vertex according to the second graph data;

According to the incremental graph edge sorting algorithm and the adjacent edges between the start vertex and the end vertex, an incremental graph edge sort is obtained.
The distributed graph computing method according to claim 5, wherein the incremental graph edge sorting algorithm is applied to the main computing node, and the incremental graph edge sorting algorithm comprises:

Sending the second graph data to the first subordinate computing node;

Acquiring a local solution sent by the first subordinate computing node;

According to the local solution, an optimized solution is obtained;

According to the optimized solution, a local optimized solution is obtained;

Sending the local optimization solution to the first subordinate computing node.
The distributed graph calculation method according to claim 3, wherein the difference graph preprocessing algorithm further comprises:

If the difference data between the second graph data and the first graph data is decrement data, remove the decrement data;

According to the first graph data, the second graph data after removing the decremented data and the graph preprocessing algorithm obtain a second intermediate preprocessed graph.
The distributed graph computing method according to any one of claims 1 to 7, wherein the distributed graph preprocessing algorithm further comprises preprocessing graph edge sorting;

The preprocessing graph edge sorting includes:

Obtaining edge data of the first graph data and vertex data of the first graph data according to the first graph data;

According to the edge data of the first graph data and the vertex data of the first graph data, the first intermediate preprocessed graph is obtained.
The distributed graph computing method according to claim 8, wherein the obtaining the first intermediate preprocessing graph according to the edge data and the vertex data comprises:

Acquiring first vertex data of the first graph data;

According to the edge sorting of the preprocessing graph, the priority queue is obtained;

Obtain the first intermediate preprocessing map according to the breadth first search and the priority queue.
The distributed graph calculation method according to any one of claims 1 to 7, wherein the first intermediate preprocessing graph comprises:

The starting vertex ID of the edge and the ending vertex ID of the edge are stored in binary format.
The distributed graph calculation method according to any one of claims 1 to 7, wherein the graph division algorithm comprises:

Acquiring nodes and node configuration information of the distributed architecture, where the node configuration information includes one or more of the number of nodes, node specifications, and node performance;

Acquiring the first intermediate preprocessing map;

Obtaining the edge data of the first intermediate preprocessing graph according to the first intermediate preprocessing graph;

Obtaining the first partition graph according to the node configuration information of the distributed architecture and the edge data of the first intermediate preprocessing graph;

Send the first partition graph to the nodes of the distributed architecture.
A terminal includes: a first memory, a first processor, and a computer program that is stored on the first memory and can run on the first processor, and when the first processor executes the program:

The distributed graph computing method according to any one of claims 1 to 11.
A distributed graph computing system, including a first distributed computing device and a second distributed computing device;

The first distributed computing device includes: a second memory, a second processor, and a first computer program that is stored on the second memory and can run on the second processor; the second processor executes the first computer program A computer program is implemented: the distributed graph calculation method according to any one of claims 1 to 11;

The second distributed computing device includes: a third memory, a third processor, and a second computer program that is stored on the third memory and can run on the third processor; the third processor executes the first The computer program realizes: the distributed graph computing method according to any one of claims 1 to 11.
A computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to:

The distributed graph calculation method according to any one of claims 1 to 11 is executed.