CN106357418B

CN106357418B - A kind of feature extracting method and device based on complex network

Info

Publication number: CN106357418B
Application number: CN201510415719.6A
Authority: CN
Inventors: 毛仁歆
Original assignee: Alibaba Group Holding Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2015-07-15
Filing date: 2015-07-15
Publication date: 2019-08-30
Anticipated expiration: 2035-07-15
Also published as: CN106357418A

Abstract

This application discloses a kind of feature extracting method and device based on complex network, this method comprises: configuration equipment is according to the corresponding directed graph of complex network, distribute all nodes included in the directed graph corresponding nodal information for each arithmetic facility, the nodal information is distributed to each arithmetic facility by the configuration equipment, so that each arithmetic facility specified node according to corresponding to the nodal information of distribution, determine the transitive relation between each specified node, the grid reference unit in the directed graph is determined according to the transitive relation, and the corresponding network feature information of the directed graph is extracted according to the grid reference unit.By this method, cyberrelationship complicated in complex network can be changed into simple network structure elements, effectively improve the accuracy for the network characteristic information extracted in complex network.

Description

A kind of feature extracting method and device based on complex network

Technical field

This application involves field of computer technology more particularly to a kind of feature extracting methods and dress based on complex network It sets.

Background technique

With the development of information technology, the communication network of carry out information exchange between from people in real life Network, social relationships net, then arrive transportation network, internet etc. complexity relational system, can be regarded as one it is huge and Complicated network, that is, complex network.Wherein, complex network refers to that (unit or individual can be referred to as a large amount of units or individual The node of complex network) between interact and the complicated network structure that is formed.

In the prior art, the feature for needing to extract complex network to the research of complex network is analyzed.For any multiple Miscellaneous network is distributed generally directed to its power law (Power Law) or gathers the features such as (Clustering) coefficient and analyzed.Its In, power-law distribution, emphasis is based on possessing the biggish node of connection number in complex network, reflects between Node Contraction in Complex Networks Inhomogeneities；And gather coefficient emphasis and be based on distribution density large area in complex network, reflect the intensive journey of complex network Degree.

But the features described above of complex network, it can not sufficiently reflect the essential attribute of complex network, that is to say, that existing There is the extracted feature in relation to complex network in technology to be inaccurate.

Summary of the invention

The embodiment of the present application provides a kind of feature extracting method and device based on complex network, to solve the prior art In the problem lower to the extracted feature accuracy of complex network.

A kind of feature extracting method based on complex network provided by the embodiments of the present application, comprising:

Equipment is configured according to the corresponding directed graph of complex network, distributes institute in the directed graph for each arithmetic facility The corresponding nodal information of all nodes for including；

The nodal information is distributed to each arithmetic facility by the configuration equipment, so that each arithmetic facility is according to distribution Nodal information corresponding to specified node, determine the transitive relation between each specified node, it is true according to the transitive relation The grid reference unit in the directed graph is made, and the directed graph is extracted according to the grid reference unit and is corresponded to Network feature information.

A kind of feature extracting method based on complex network that the embodiment of the present application also provides, comprising:

Arithmetic facility receives the nodal information sent by configuration equipment；

According to the nodal information received, determine that the transmitting between the corresponding each specified node of the nodal information is closed System；

According to the transitive relation between each specified node, the all-network benchmark that each specified node is formed is determined Unit；

According to the grid reference unit determined, the corresponding network feature information of the directed graph is extracted.

A kind of feature extracting method based on complex network that the embodiment of the present application separately provides, arithmetic facility have for network Into figure, each specified node sends instruction information, which comprises

The specified node receives the instruction information that the arithmetic facility is sent；

According to the instruction information, the transitive relation between the specified node and other specified nodes is determined；

According to the transitive relation, the all-network reference cell formed by the specified node is determined；

The grid reference unit determined is fed back into the arithmetic facility, so that the arithmetic facility extracts institute State the corresponding network feature information of directed graph.

A kind of feature deriving means based on complex network provided by the embodiments of the present application, comprising:

Configuration module, for it is oriented to distribute the network for each arithmetic facility according to the corresponding directed graph of complex network The corresponding nodal information of all nodes included in figure；

Sending module, for the nodal information to be distributed to each arithmetic facility so that each arithmetic facility according to point Specified node corresponding to the nodal information matched determines the transitive relation between each specified node, according to the transitive relation It determines the grid reference unit in the directed graph, and the directed graph pair is extracted according to the grid reference unit The network feature information answered.

A kind of feature deriving means based on complex network that the embodiment of the present application separately provides, comprising:

Receiving module, for receiving the nodal information sent by configuration equipment；

Determining module, for determining the corresponding each specified section of the nodal information according to the nodal information received Transitive relation between point；

Grid reference unit module, for determining each described specified according to the transitive relation between each specified node The all-network reference cell that node is formed；

Network feature information module, for extracting the directed graph according to the grid reference unit determined Corresponding network feature information.

A kind of feature deriving means based on complex network that the embodiment of the present application separately provides, arithmetic facility have for network Into figure, each specified node sends instruction information, and described device includes:

Receiving module, the instruction information sent for receiving the arithmetic facility；

Determining module, for according to the instruction information, determining the transmitting between the specified node and other specified nodes Relationship；

Grid reference unit module, for determining all nets formed by the specified node according to the transitive relation Network reference cell；

Network feature information module, for the grid reference unit determined to be fed back to the arithmetic facility, with So that the arithmetic facility extracts the corresponding network feature information of the directed graph.

The embodiment of the present application provides a kind of feature extracting method and device based on complex network, passes through this method, configuration All nodes included in directed graph can be respectively allocated to a large amount of arithmetic facility, arithmetic facility then can root by equipment According to the node that configuration equipment is distributed, the transitive relation between these nodes is determined, and determined according to transitive relation step Grid reference unit included in the directed graph, and the directed graph pair is extracted according to grid reference unit The network feature information answered.Unlike the prior art, the Node distribution in complex network is paid close attention in the prior art Concentration or the degree of irregularity of the possessed connection of node cannot but embody each mutual shape of node in complex network At network structure, and the mode in the embodiment of the present application is to be divided into complex network based on the transitive relation between node Cyberrelationship complicated in complex network so can be changed into simple structural unit by basic structural unit, The accuracy analyzed network characteristic in complex network is effectively improved, also, in the embodiment of the present application, for including sea For measuring node complex network, by the way of distributed arithmetic, the efficiency of operation is effectively improved.

Detailed description of the invention

The drawings described herein are used to provide a further understanding of the present application, constitutes part of this application, this Shen Illustrative embodiments and their description please are not constituted an undue limitation on the present application for explaining the application.In the accompanying drawings:

Fig. 1 a is a kind of characteristic extraction procedure based on complex network provided by the embodiments of the present application；

Fig. 1 b is the configuration diagram of distributed system provided by the embodiments of the present application；

Fig. 2 is another characteristic extraction procedure based on complex network provided by the embodiments of the present application；

The schematic diagram of oriented transitive relation of the Fig. 3 between node provided by the embodiments of the present application；

The schematic diagram of undirected transitive relation of the Fig. 4 between node provided by the embodiments of the present application；

Fig. 5 a is all types of schematic diagrames of minimal network structure provided by the embodiments of the present application；

Fig. 5 b is the schematic diagram of nonoriented edge triangle provided by the embodiments of the present application；

Fig. 6 is the corresponding all nonoriented edges of different types of grid reference unit provided by the embodiments of the present application Value；

Fig. 7 a and 7b are the schematic diagram of the transitive relation of application example interior joint provided by the embodiments of the present application；

Fig. 8 is the structural schematic diagram in the embodiment of the present application between the configuration equipment provided and multiple arithmetic facilities；

Fig. 9 to Figure 11 is the feature deriving means structural schematic diagram provided by the embodiments of the present application based on complex network.

Specific embodiment

To keep the purposes, technical schemes and advantages of the application clearer, below in conjunction with the application specific embodiment and Technical scheme is clearly and completely described in corresponding attached drawing.Obviously, described embodiment is only the application one Section Example, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not doing Every other embodiment obtained under the premise of creative work out, shall fall in the protection scope of this application.

Fig. 1 a is the feature extracting method provided by the embodiments of the present application based on complex network, and this method specifically includes following Step:

S101 configures equipment according to the corresponding directed graph of complex network, it is oriented to distribute the network for each arithmetic facility The corresponding nodal information of all nodes included in figure.

In the embodiment of the present application, the directed graph is made according to the information transfering relation in complex network The network graphic being made of node and directed edge out.In directed graph, node table shows that the individual in complex network is (described Individual, can be the basic unit of the complicated composition networks such as terminal, user itself, city), directed edge indicate individual between Transitive relation (direction that the direction of directed edge has meant that transmitting).The generation of directed graph in the application, can be by phase The equipment answered generates after summarizing all information in relation to the complex network.Here for the generation of the directed graph, It is not intended as the restriction to the application.

A large amount of node is generally comprised under practical application scene, in directed graph, and (quantity of usual node is million Grade is even more to 1,000,000,000 grades), in this case, if only by one or several a small number of arithmetic facilities carry out operation, Since operand is excessive, then, the time needed for calculating process will be greatly increased.

Therefore, in the embodiment of the present application, the operation extraction process to the feature of directed graph is realized, it can be by distribution System is completed.Include more arithmetic facilities and configuration equipment in the distributed system, wherein the configuration equipment, for pair Each arithmetic facility in the distributed system is managed, such as: scheduling data, issue specific instruction at configuration data.Wherein, The configuration equipment includes but is not limited to: having the network equipments such as computer, the server of global control function.

That is, configuration equipment can be by all nodes included in the directed graph, respectively in distributed system All arithmetic facility distribution nodes in the distributed system.

It should be noted that the distributed system, the service system that can be Internet service provider backstage is constituted, at this Under the scene of sample, the arithmetic facility be can be in the server cluster for providing calculation process business in the service system Server.And the configuration equipment, it can be the host for being responsible for management control in the service system.Certainly, as the application Another realization scene in embodiment, the distributed system can also be point constituted by being distributed existing great amount of terminals Distribution system, under such a scenario, arithmetic facility are each terminals, and configuring equipment can be responsible for each terminal distribution data simultaneously Receive the server of operation result.Certainly, the restriction to the application is not constituted here.

In the embodiment of the present application, configuration equipment is that the node of each arithmetic facility distribution is different, in other words, if will The directed graph regards the set S comprising all nodes as, then, the node that configuration equipment is distributed by each arithmetic facility, It then can be regarded as the different subsets of the set S, also, intersection be not present between the different subsets.

Configuration equipment will generate respectively different nodal informations for different arithmetic facilities, so as not to same arithmetic facility Handle different nodes, therefore the specified node, the node exactly distributed by the configuration equipment by arithmetic facility.When So, in practical application, the nodal information can transmit by the way of configuration file, specific limit is not made herein certainly It is fixed.

The nodal information is distributed to each arithmetic facility by S102, the configuration equipment, so that each arithmetic facility root Specified node corresponding to nodal information according to distribution determines the transitive relation between each specified node, according to the transmitting Relationship determines the grid reference unit in the directed graph, and it is oriented according to the grid reference unit to extract the network Scheme corresponding network feature information.

In the embodiment of the present application, the grid reference unit is the basic structural unit constituted in directed graph. Under a kind of preferred embodiment, the grid reference unit is the structural unit being made of three nodes.For any complex web For network, it can be regarded as being made of a large amount of grid reference unit.

After each arithmetic facility has received the nodal information sent by configuration equipment, then, each operation is set It is standby to determine each specified node that configuration equipment is distributed for the arithmetic facility according to the nodal information.Further Ground, each arithmetic facility can be determined included in complete directed graph for each specified node determined All-network reference cell, to determine the corresponding network feature information of the network digraph according to grid reference unit.

Through the above steps, all nodes included in directed graph can be respectively allocated to a large amount of by configuration equipment Arithmetic facility, arithmetic facility then can determine the transitive relation between these nodes according to the node that be distributed of configuration equipment, and Grid reference unit included in the directed graph is determined according to transitive relation step, and according to grid reference list Member extracts the corresponding network feature information of the directed graph.Unlike the prior art, emphasis closes in the prior art The concentration of the Node distribution in complex network or the degree of irregularity of the possessed connection of node are infused, cannot but be embodied multiple The network structure that each node is formed between each other in miscellaneous network, and the mode in the embodiment of the present application, are based between node Complex network is divided into basic structural unit by transitive relation, so, network complicated in complex network can be closed System, is changed into simple structural unit, effectively improves the accuracy analyzed network characteristic in complex network, also, at this Apply for including magnanimity node complex network, by the way of distributed arithmetic, effectively improving fortune in embodiment The efficiency of calculation.

It should be noted that alternatively, the nodal information can also carry in configuration information.Specifically For, due to including a large amount of node in the directed graph of complex network, if configuration equipment separates these nodes Out, it will a large amount of calculation resources are expended, so, configuration equipment can be directed to directly in directed graph and distribute to difference Different labels (namely the another of nodal information embodies) is arranged in the node of arithmetic facility, then the network after label is oriented Figure carries in configuration information respectively, is distributed to different arithmetic facilities, so, different arithmetic facilities can basis Different label in directed graph, to determine each node that itself will be responsible for processing.Certainly, only nodal information here Another optional way, do not constitute the restriction to the application.

In addition, in the embodiment of the present application, determine the process of the grid reference unit of directed graph, be based on node it Between transitive relation, and each node in complex network has transitive relation all between other nodes, then, be It can understand in subsequent process, clearly indicate out transitive relation between different nodes, it is necessary to be set for all node Corresponding mark is set, that is, in the embodiment of the present application, distributing institute included in the directed graph for each arithmetic facility Before having node, method S101~S102 as shown in Figure 1 further include: the configuration equipment is in the directed graph Globally unique identification information is arranged in each node.The identification information can uniquely determine out a node.

It should also be noted that, due in practical applications, the node data for including in the directed graph of complex network It is huge, in order to guarantee to carry out the efficiency of operation to directed graph, for distributed system employed in the embodiment of the present application, Specifically used Integral synchronous parallel computation (Bulk Synchronous Parallel, BSP) model realization.BSP model can make It obtains each arithmetic facility and calculation process is carried out to the nodal information of configuration equipment distribution simultaneously, also, is logical between each arithmetic facility It is synchronous (that is, can be in communication with each other between each arithmetic facility) to cross fence, it is ensured that the complete process of whole network digraph, Corresponding operation result, pole can be exported to the directed graph for including more than one hundred million nodes using BSP model in a very short period of time The treatment effeciency to directed graph is improved greatly.

Certainly, what needs to be explained here is that, configuration equipment and arithmetic facility for above content, in the embodiment of the present application Framework it is as shown in Figure 1 b, that is, a configuration equipment corresponds to more arithmetic facilities, form BSP between different arithmetic facilities Framework.

It in the embodiment of the present application, is that each arithmetic facility is configured in directed graph not in above-mentioned configuration equipment After same node, each arithmetic facility will execute the transitive relation between corresponding determining node according to respective configuration information Operation, onestep extraction of going forward side by side goes out the feature (namely feature of complex network) in the directed graph, specifically, for each A kind of feature extracting method based on complex network is provided for arithmetic facility, in the embodiment of the present application, as shown in Fig. 2, institute State method comprising steps of

S201, arithmetic facility receive the nodal information sent by configuration equipment.

As described in above-described embodiment, the arithmetic facility in the application can be server in service system or Terminal is not intended as the restriction to the application here.

It may include the identification information for the node that configuration equipment is arithmetic facility distribution in the nodal information.Such one Come, arithmetic facility can determine itself each specified node to be treated according to identification information.

What needs to be explained here is that including network in nodal information as one of the embodiment of the present application optional way Digraph, in other words, configuration equipment is only in directed graph, node mark that will distribute to a certain arithmetic facility and handle Note comes out, and the complete directed graph (node being marked is carried in directed graph at this time) is sent to Corresponding arithmetic facility.

S202 is determined between the corresponding each specified node of the nodal information according to the nodal information received Transitive relation.

After the arithmetic facility has received the nodal information, then it can read the nodal information, and determine itself It will be responsible for each node of processing.

If continuing to use above-mentioned optional way, arithmetic facility has received a complete directed graph, this When, which can read in the directed graph, and being configured all nodes of device flag, (labeled node is exactly this Arithmetic facility specified node to be treated).Meanwhile the transitive relation between different nodes, institute are shown in directed graph With the arithmetic facility can also determine transitive relation mutual between specified node.

S203 determines all nets that each specified node is formed according to the transitive relation between each specified node Network reference cell.

Just because of having transitive relation between different nodes, to can just form network (in general, transitive relation Take the form of the line between node), then, for arithmetic facility, according to the transitive relation between specified node, just It can determine contacting between each specified node and other specified nodes, such as: that determines each node and other nodes has Xiang Bian.Further, so that it may according to the transitive relation of each specified node, determine the smallest network knot in whole network Structure body, that is, grid reference unit described herein.

It should be noted that the grid reference unit, can be regarded as the smallest network structure elements of whole network, In the preferred embodiment of the application, since the grid reference unit is made of three nodes, then, the grid reference unit is just It can be regarded as the structural unit of triangle, which is exactly the node in directed graph, and the triangle While being exactly the directed edge for reflecting between node and node relationship of transmitting.

S204 extracts the corresponding network characterization letter of the directed graph according to the grid reference unit determined Breath.

For the complex network with heterogeneous networks structure, mutually isostructural grid reference list can be divided Member, then, entire complex network can be regarded as to be composed of a large amount of grid reference unit.So, no matter multiple Which kind of network structure is miscellaneous network structure have, and can determine its most basic grid reference unit, that is, " change is whole to be Zero ", the network feature information extracted in this way just reflects the most basic network structure of entire complex network Information, even therefore the complex network of completely not identical (such as: magnitude, the network structure of the node possessed be not identical), It can carry out going deep into comparing or analyzing by network base units.

Through the above steps, the cyberrelationship between specified node that arithmetic facility distributes configuration equipment, is divided into Grid reference unit just because of such mode can be by the complex network with heterogeneous networks structure, and being divided into has The basic network topology of same characteristics, further extracts network feature information, so as to realize the depth to complex network Degree analysis also improves the accuracy analyzed to complex network feature and comprehensive.

It should be noted that in practical applications, directed graph reflects the transitive relation of each node in complex network, When to analyze different complex networks, if using the corresponding directed graph of each complex network, it is necessary to it is analyzing During using the transitive relation between each node, still, in practical applications, for different complex networks, The difference of transitive relation between its interior joint may be very big, this will make the accuracy of subsequent analysis network characterization At influence.Therefore, in the embodiment of the present application, alternatively, directed graph can be converted to the shape of non-directed graph Formula, specifically, can determine whether the transitive relation between specified node first for each arithmetic facility, it may be assumed that

For the transitive relation in above-mentioned steps S202, determined between the corresponding each specified node of the nodal information, tool Body are as follows: be directed to each specified node, indicate that the specified node sends a notification message to immediate neighbor node, receive all have been received by The report information of the immediate neighbor node feedback of the notification message determines each specified section according to the report information All immediate neighbor nodes of point.

Wherein, the immediate neighbor node refers to and the node for specifying node to be connected directly.

What needs to be explained here is that arithmetic facility is responsible for the quantity of the specified node of processing, usually at 100,000 grades to ten million Grade determines the transitive relation between specified node if budgeting device itself traverses each specified node, then, it will also expend Longer time, and bear biggish operation pressure.But in the state of initialization.And the node in the embodiment of the present application, It can be a kind of example of data structure, that is, different nodes can communicate with each other, arithmetic facility can be to each specified Node is controlled, so that each node is in communication with each other, gives the arithmetic facility, so, arithmetic facility reporting To be effectively saved the time consumed by calculating process.But under init state, for different specified nodes, It can only know that the transitive relation issued from itself (in directed graph, shows as the node and itself is directed toward having for another node To arrow, that is, side out), but can not know that itself receiving transitive relation (in directed graph, shows as another section Point is directed toward the oriented arrow of the node itself, that is, entering side), so, arithmetic facility is in order to accurately determine all specified nodes Between the transitive relation of " transmission " and " reception " each specified node will be indicated by arithmetic facility in the embodiment of the present application, lead to The connection for itself being directed toward other nodes is crossed, is sent a notification message.

To, the node with " entering side " has also been received by the notification message of other nodes transmission, then, the node Also just having known has node itself to be connected with the node.For each specified node, after having received notification message, all can Report information is generated, arithmetic facility is sent to, in turn, arithmetic facility is also it is exactly known that biography between all specified nodes Pass relationship.

Such as: as shown in figure 3, in fig. 3 it is shown that transitive relation between four node A~D, wherein node A refers to To node C and B, node C is directed toward node D.

It is directed toward itself transitive relation since any node cannot be known, therefore at this point, for node B, and do not know With the transitive relation between other nodes.For node C, node C can determine the transitive relation with node D, but simultaneously The uncertain transitive relation with node A and B.For node D, the transitive relation between other nodes is not known.

Assuming that certain arithmetic facility indicates that all nodes in Fig. 3 send a notification message to respectively pointed node respectively, So, at this point, node A can send a notification message to node B and C, node C can send a notification message to node D.To node B And C just would know that itself there is transitive relation (being directed toward node B and node C by node A), similarly, node D between node A Can also know itself has transitive relation (being directed toward node D by node C) between node C.

It can be seen from the above example that, after all nodes that different specified nodes is directed toward to itself send a notification message, then, For each specified node, itself transitive relation between other specified nodes can be also known.

For other the specified nodes for allowing each specified node clearly to know and itself there is transitive relation, so, Under one of the embodiment of the present application optional way, indicate that the specified node sends a notification message to immediate neighbor node, specifically Are as follows: it indicates that the specified node carries the identification information of itself in notification information, and the notification message is sent to this and is referred to Determine all immediate neighbor nodes of node.

So, it for arithmetic facility, on the basis of the above, according to the report information, determines each described specified All immediate neighbor nodes of node, specifically: determine identification information carry in report information, that node itself is specified except this Other described nodes are determined as the immediate neighbor section of the specified node by other nodes corresponding to other identifier information in addition Point.

By above-mentioned steps, arithmetic facility has just known the transitive relation between each specified node (including " transmission " The transitive relation of transitive relation and " reception "), that is to say, that for each specified node, it can determine and node is specified by this The directed edge (that is, the directed edge for being directed toward the specified node) that the directed edge and the specified node that itself is issued receive.

In view of for any directed graph, wherein all directed edges are changed into two-way side in practical application When, then, which is also just changed into non-directed graph.To after having known the transitive relation between each node, just The nonoriented edge between each node can be constructed, then by the nonoriented edge of building, determines all grid reference units, That is, according to the transitive relation between each specified node, it is all to determine that each specified node is formed for above-mentioned steps Grid reference unit, specifically: according to the transitive relation between each specified node, construct each specified node and it is all directly Nonoriented edge between neighbor node, and determined between each specified node for each nonoriented edge assignment according to each nonoriented edge after assignment The all-network reference cell of formation.

The nonoriented edge is exactly the two-way side between specified node, then, nonoriented edge is converted by the directed edge, Process exactly by unidirectional when being converted into two-way.

The example now continued to use in Fig. 3 is illustrated.To will be unidirectional when being converted into two-way between 4 nodes in Fig. 3, Specifically: it is directed toward the unidirectional side of node B for node A, now increases the unidirectional side for being directed toward node A by node B, to be formed Two-way side between node A and node B, i.e., as shown in Figure 4.Similarly, for node A and node C and node C and section Unidirectional side between point D can be converted into two-way side using the same manner, that is, as shown in Figure 4.

But when being converted to nonoriented edge by unidirectional side, not indicate that actual transitive relation becomes two-way between node Transitive relation, and simply facilitate and directed graph be converted to non-directed graph.So, in the embodiment of the present application, by node Between directed edge be converted to nonoriented edge after, for actual transitive relation between clearly each node, it is necessary to what is newly constructed The corresponding mark of nonoriented edge setting (being namely the nonoriented edge assignment after building), to indicate the nonoriented edge actually corresponding transmitting Relationship.

Specifically, be each nonoriented edge assignment, specifically: determine the identification information of all specified nodes, it is suitable according to specifying All identification informations that ordered pair is determined are ranked up, and the use of building rule are each nonoriented edge according to the identification information after sequence Assignment.

In above-mentioned assignment procedure, the sequence in view of different node identification informations is needed, this is because: to the nothing of building Assignment is carried out to side, is based on actual transitive relation between node, still, if the great deal of nodes in directed graph is respectively Identification information without rule, then, will lead to and cause confusion to the assignment of the nonoriented edge between each node.

Such as: assuming that the identification information of two nodes is respectively as follows: A, aa, the actual transfer relationship between the two nodes Are as follows: node A is directed toward node aa.If constructing the nonoriented edge between the two nodes, assignment is being carried out to the nonoriented edge When, the value of the nonoriented edge both can be 1, or 0, it is clear that the nonoriented edge after assignment can not accurately reflect two Actual transitive relation between a node.

So in the embodiment of the present application, arithmetic facility will be ranked up the identification information of each specified node.And it is directed to Each specified node after sequence is nonoriented edge assignment using corresponding building rule.

In the embodiment of the present application, the building rule includes: to sort different nodes for two identification informations, when two Transitive relation between a node is the having of the higher lower node of node point identification information sorting of being sorted by identification information When to side, then the nonoriented edge between two nodes after building is assigned a value of the first numerical value；

When the transitive relation between two nodes be by identification information sort lower node point identification information sorting compared with When the directed edge of high node, then the nonoriented edge between two nodes after building is assigned a value of second value；

When the transitive relation between two nodes is two-way side, then the nonoriented edge between two nodes after building is assigned Value is third value.

For example, it is assumed that the first numerical value is 1, and second value 0, third value 2, for the nonoriented edge of the building in Fig. 4 For, 0 is assigned a value of (because being the having of the higher lower node of node point identification information sorting of being sorted by identification information To side).

In the above process, for each node, itself contacting between immediate neighbor node is known, still, in net In network digraph, all grid reference units are determined, it is also necessary to know and the related all sections of immediate neighbor node Point.Such as: in Fig. 4, for node B, node B only knows the immediate neighbor node A of itself, still, does not know but and section The connection of point C, however, may make up grid reference unit between node A, B, C, it is clear that constitute the grid reference unit, save Point B just needs to know the relationship between node C.So, in order to determine grid reference unit, in the embodiment of the present application, According to each nonoriented edge after assignment, the all-network reference cell formed between each specified node is determined, specifically: for each Specified node indicates that this is specified the link information of node itself Yu all immediate neighbor nodes by the specified node, sends respectively To each immediate neighbor node in addition to the immediate neighbor node, and the link information that all immediate neighbor nodes are sent is received, According to all link informations that each specified node receives, the corresponding all indirect neighbor nodes of each specified node, structure are determined Build the nonoriented edge of the specified node all indirect neighbor nodes corresponding with the specified node, and for specified node and it is each indirectly Each nonoriented edge assignment between node, according to the specified node after assignment and undirected between all immediate neighbor nodes The specified node behind side and assignment and the nonoriented edge between indirect neighbor node, are determined containing the specified node All-network reference cell.

Wherein, the link information includes: the identification information of specified node and immediate neighbor node.The indirect neighbor section Point refers to: not being connected directly with the specified node, with the node for specifying the immediate neighbor node of node to be connected directly.

In general, there is no transitive relation between specified node and indirect neighbor node, still, specify node and indirect inode with And direct node may be constructed the grid reference unit of triangle.That is, arithmetic facility need to construct specified node and Connect the nonoriented edge between node.To specify each nonoriented edge assignment between node and each indirect inode, specifically: determine all fingers The identification information for determining node is ranked up all identification informations determined according to specified sequence, according to the mark after sequence Information uses each nonoriented edge assignment of the building rule between specified node and each indirect inode.

Wherein, it includes: when there is no transitive relations between two nodes that the building rule, which includes: the building rule, When, the nonoriented edge between two nodes after building is assigned a value of the 4th numerical value.

It should be noted that in practical applications, the smallest network structure (is made of 3 nodes in directed graph Triangle) 13 kinds are shared, it is specific as shown in Figure 5 a.In fig 5 a, showing between various three nodes, there are different transmitting to close System, and the different triangles constituted.In this application, it for the various triangles in Fig. 5 a, need to be converted to as shown in Figure 5 b Nonoriented edge triangle.

So the all-network reference cell formed containing the specified node is determined in above-mentioned steps, specifically: For each specified node, determine to determine each nonoriented edge three using all nonoriented edge triangles for specifying node as vertex Angular corresponding value, as the all-network reference cell for specifying node containing this determined.

Wherein, the corresponding value of a nonoriented edge triangle are as follows: the assignment of every nonoriented edge of the nonoriented edge triangle combines Made of be worth.

It should be noted that as shown in fig. 6, all kinds of triangles shown in Fig. 5 a are being converted into nonoriented edge triangle Afterwards, the corresponding value of nonoriented edge triangle.Each bit value in Fig. 6 is exactly the corresponding value of nonoriented edge, the value of three nonoriented edges It has been composed the value of nonoriented edge triangle.

In practical application, arithmetic facility can for each specified node, out using this specify node as the institute on vertex whether there is or not To side triangle, and determine the corresponding value of each nonoriented edge triangle, such mode it is possible that repeat, so, at this Apply in embodiment, in order to guarantee the value of each nonoriented edge triangle determined more accurately without duplicating, therefore, According to the grid reference unit determined, before extracting the corresponding network feature information of the directed graph, the side Method further include: the arithmetic facility removes each grid reference unit determined according to the identification information of each specified node Weight.

On the basis of the above, the corresponding network feature information of the directed graph is extracted in step S204, is had Body are as follows: the quantity for counting identical grid reference unit in the directed graph, as the corresponding net of the directed graph Network characteristic information.

In the following, by being said with an application example to the feature extracting method based on complex network described in the application It is bright:

Assuming that containing 4 nodes: node A~D in the corresponding directed graph of a certain complex network, wherein the net Network digraph is as shown in Figure 7a.It, will be all in the directed graph and after assuming that configuring equipment obtains the directed graph Node is allocated to an arithmetic facility.

At this point, for the arithmetic facility following steps will be executed:

Step 1: arithmetic facility reads in the nodal information in the directed graph of configuration equipment distribution；

Step 2: arithmetic facility indicates that each node carries self identification to the immediate neighbor node of oneself transmission respectively The notification message of information.Such as: node A sends a notification message " a " to node C, and node B sends a notification message to node A, node C " b ", node C send a notification message " c " to node B, node D.Wherein, node A is carried respectively in notification message " a "~" c " The identification information of~C.

Step 3: for each node, according to the value on the message transformation side received.Such as: node A has received message " b ", that , node A just known that itself is connected directly with node B, meanwhile, node A confirms itself, and there is no be directed toward node B to go out side, So since the sequence according to identification information is (that is, A < B, indicates the higher node point identification information that sorted by identification information Sort the directed edge of lower node), according to building rule, the directed edge of node A and node B is configured to two-way by arithmetic facility Side, and 1 is set by the Value (assignment) on the two-way side between node A and node B；Simultaneously for other message received Node uses identical processing method, between Value=0, the node B and node C on the side between final node A and node C While Value=2, node C and node D between while Value=0.

Step 4: for each node, arithmetic facility indicates each node, by each immediate neighbor node of itself Link information is sent to another immediate neighbor node.Such as: link information " A is connected to B " is passed to node C by node A, this As soon as sample, node C has known the relationship between itself immediate neighbor node A and B.

Equally, for node B, link information " B is connected to A " can be passed to node C；Node C can be by link information " C It is connected to A " and passes to node B and node D；And link information " C is connected to B " is passed to node D by node C.

Step 5: arithmetic facility determines the all-network reference cell formed containing the specified node.Such as: for section Point A can determine that the value of nonoriented edge triangle can determine nothing similarly, for node B, C, D for 120,00-1 Value to side triangle is.

Step 6: duplicate removal is carried out to the value for the nonoriented edge triangle determined, by the value of the nonoriented edge triangle after duplicate removal As grid reference unit.

Step 7: extracting all identical grid reference units, determine grid reference unit respectively correspond in Fig. 63, 5 and 11 classes, and the quantity of this 3 class is counted, the network feature information as the directed graph (namely complex network).

The executing subject of the above content and step is arithmetic facility, as the optional side of one of the embodiment of the present application Formula, can also be by node itself as executing subject, that is, also providing a kind of spy based on complex network in the embodiment of the present application Extracting method is levied, as shown in figure 8, arithmetic facility indicates information, the method packet for each specified node in directed graph It includes:

S801, the specified node receive the instruction information that the arithmetic facility is sent；

S802 determines the transitive relation between the specified node and other specified nodes according to the instruction information；

S803 determines the all-network reference cell formed by the specified node according to the transitive relation；

The grid reference unit determined is fed back to the arithmetic facility, so that the arithmetic facility by S804 Extract the corresponding network feature information of the directed graph.

For above-mentioned steps S802, specifically: the specified node specifies the straight of node according to the instruction information, to this It connects neighbor node to send a notification message, and receives the notification message of other immediate neighbor nodes transmission, according to the notification message, Determine all immediate neighbor nodes of the specified node.

The all-network reference cell formed by the specified node, tool are determined according to the transitive relation for step Body are as follows: according to the transitive relation, construct the nonoriented edge between the specified node and all immediate neighbor nodes, and be each nothing The all-network reference cell formed by the specified node is determined according to each nonoriented edge after assignment to side assignment.

For the step of for each nonoriented edge assignment, specifically: extract the sequencing information for including in instruction information；Wherein, institute Identification information of the sequencing information by the arithmetic facility according to all specified nodes is stated, according to specified sequence to all identification informations It is generated after being ranked up；It the use of building rule is each nonoriented edge assignment according to the sequencing information.

It should be noted that the building rule includes: to sort different nodes for two identification informations, saved when two Transitive relation between point is the directed edge of the higher lower node of node point identification information sorting of being sorted by identification information When, then the nonoriented edge between two nodes after building is assigned a value of the first numerical value；

More specifically, according to each nonoriented edge after assignment, the all-network benchmark formed by the specified node is determined Unit, specifically: the link information with any immediate neighbor node is sent respectively to except the neighbor node by the specified node Except each immediate neighbor node, and receive the link information that all immediate neighbor nodes are sent；Wherein, the link information packet Include: the value of the nonoriented edge constructed between specified node and immediate neighbor node, the specified node identification information and this directly The identification information of neighbor node；The specified node is determined according to the link information of all immediate neighbor nodes received The corresponding all indirect neighbor nodes of the specified node；Wherein, the indirect neighbor node refers to the specified node not It is connected directly, with the node for specifying the immediate neighbor node of node to be connected directly；Construct the specified corresponding institute of node itself There is the nonoriented edge of indirect neighbor node, and is each nonoriented edge assignment；According to after assignment the specified node and all direct neighbours Occupy the nonoriented edge between node and the specified node after assignment and the nonoriented edge between indirect neighbor node, determine by The all-network reference cell that the specified node is formed.

It in the above scenario, is each nonoriented edge assignment, specifically: extract the sequencing information for including in instruction information；Wherein, The sequencing information, according to the identification information of all specified nodes, is believed all marks according to specified sequence by the arithmetic facility Breath generates after being ranked up；It the use of building rule is each nonoriented edge assignment according to the sequencing information.

At this point, the building rule includes: when transitive relation is not present between two nodes, then by two after building Nonoriented edge between node is assigned a value of the 4th numerical value.

According to described after the specified node after assignment and the nonoriented edge between all immediate neighbor nodes and assignment Nonoriented edge between specified node and indirect neighbor node, determines the all-network reference cell for specifying node to be formed by this, It specifically includes:

The specified node is determined to determine each undirected using all nonoriented edge triangles for specifying node as vertex The corresponding value of side triangle, as the all-network reference cell for specifying node containing this determined, wherein a nonoriented edge The corresponding value of triangle are as follows: the value that the assignment of every nonoriented edge of the nonoriented edge triangle is composed；It is all by what is determined The corresponding value of nonoriented edge triangle, is sent to all immediate neighbor nodes and indirect neighbor node of the specified node, and receives All immediate neighbor nodes and indirect neighbor node send by immediate neighbor node and indirect neighbor node determines whether there is or not To the set of the corresponding value of side triangle；According to the corresponding value of all nonoriented edge triangles received and the specified node The corresponding value of all nonoriented edge triangles that itself is determined carries out duplicate removal；According to the value of the nonoriented edge triangle after duplicate removal, determine It out include the all-network reference cell of the specified node.

In the following, still continue to use example shown in Fig. 7 a and Fig. 7 b, to described in the present embodiment based on complex network Feature extracting method is illustrated:

After configuration equipment obtains the directed graph, all nodes in the directed graph are allocated to one Arithmetic facility.At this point, the arithmetic facility sends instruction information for all specified nodes, so that all specified nodes execute Following steps:

Step 1: each node disappears to the notice that the immediate neighbor node of oneself transmission carries self identification information respectively Breath.Such as: node A sends a notification message " a " to node C, and node B sends a notification message " b " to node A, node C, node C is to section Point B, node D send a notification message " c ".Wherein, the identification information of node A~C is carried in notification message " a "~" c " respectively.

Step 2: for each node, according to the value on the message transformation side received.Such as: node A has received message " b ", that , node A just known that itself is connected directly with node B, meanwhile, node A confirms itself, and there is no be directed toward node B to go out side, So since the sequence according to identification information is (that is, A < B, indicates the higher node point identification information that sorted by identification information Sort the directed edge of lower node), according to building rule, the directed edge of itself and node B are configured to two-way side by node A, And 1 is set by the Value (assignment) on the two-way side between node A and node B；Simultaneously for the node of other message received Using identical processing method, between final node A and node C while Value=0, node B and node C between while The Value=0 on the side between Value=2, node C and node D.

Step 3: for each node, the link information of each immediate neighbor node of itself is sent to other one A immediate neighbor node.Such as: by link information, " C is connected to A to node C, and Value=0 " passes to node B, so, node B has just known the relationship between itself immediate neighbor node A and C.Node B has received message, and " C is connected to A, Value=0 " Afterwards, discovery node B exist and node A while and Value=1, node B exist and node C while and Value=2, simultaneously because A < B < C on lexcographical order, then arrange message " C is connected to A, Value=0 " be " A:1:B:2:C:0 " (here " A:1:B:2:C: 0 " indicates, the nonoriented edge Value=1 between node A and node B, the nonoriented edge Value=2 between node B and node C, node Nonoriented edge Value=0 between C and node A).

Identical operation carried out to other nodes for receiving message, discovery node C message can be arranged for " A:1:B:2:C: 0 ", node D can arrange message as " A:0:C:0:D:-1 ", " B:2:C:0:D:-1 ".

Step 4: node determines the all-network reference cell formed containing the specified node.Such as: for node A, It can determine that the value of nonoriented edge triangle can determine nonoriented edge similarly, for node B, C, D for 120,00-1 The value of triangle is 20-1.

Step 5: the value for the nonoriented edge triangle determined is sent to arithmetic facility and carries out duplicate removal by each node, so that Arithmetic facility is obtained using the value of the nonoriented edge triangle after duplicate removal as grid reference unit.

So far, through the foregoing embodiment provided in method S201~S204 and S801~S804, can be directed to any Complex network determines network feature information included in the complex network according to its directed graph.So, exist When practical application, so that the complex network with heterogeneous networks structure, different number of nodes magnitudes, is transformed into a large amount of and simple Structural unit, and according to the corresponding network feature information of different complex networks is determined therefrom that, on this basis, so that it may to different Complex network carries out deep analysis, such as: being formed by curve (such as: different nets according to the network feature information in complex network Number curve, significance curve of network reference cell etc.), it is formed with the network feature information in known proper network Curve be compared, judge in the complex network with the presence or absence of Network Abnormal；Alternatively, according to certain specific in complex network Network feature information (such as: the significance of each classification grid reference unit), determine the concrete kind of the complex network Type (such as: WWW, social relationships net).

Specifically for example: in practical applications, complicated Capital Flow network is formd between bank, commodity website and user (Capital Flow network at this time also just constitutes a complex network), then, it, can be with according to the method in the embodiment of the present application Determine that network feature information present in the Capital Flow network (includes: all-network reference cell in the Capital Flow network Classification and quantity), and according to the quantity of grid reference unit of all categories, judge in the Capital Flow network with the presence or absence of fund Flow abnormalities (such as: money laundering).

Certainly, a kind of practical application field of the above-mentioned example as the feature extracting method in the application based on complex network Scape does not constitute the restriction to the application.Also, the above method in the embodiment of the present application, can be used in it is any with it is complicated In the relevant analysis scene of network.

The above are the feature extracting methods provided by the embodiments of the present application based on complex network, are based on same thinking, this Application embodiment also provides a kind of feature deriving means based on complex network, as shown in Figure 9.

In Fig. 9, the feature deriving means based on complex network include: configuration module 901 and sending module 902, Wherein,

The configuration module 901, for distributing the net for each arithmetic facility according to the corresponding directed graph of complex network The corresponding nodal information of all nodes included in network digraph.

The sending module 902, for the nodal information to be distributed to each arithmetic facility, so that each arithmetic facility The specified node according to corresponding to the nodal information of distribution determines the transitive relation between each specified node, according to the biography The relationship of passing determines the grid reference unit in the directed graph, and extracts the network according to the grid reference unit and have To the corresponding network feature information of figure.

Described device further include: mark module 903, for for each node setting for including in the directed graph Globally unique identification information.

On the basis of the above, the embodiment of the present application also provides a kind of feature deriving means based on complex network, such as Figure 10 It is shown.In Figure 10, described device includes:

Receiving module 1001, for receiving the nodal information sent by configuration equipment；

Determining module 1002, for determining the corresponding each finger of the nodal information according to the nodal information received Determine the transitive relation between node；

Grid reference unit module 1003, for determining each described according to the transitive relation between each specified node The all-network reference cell that specified node is formed；

Network feature information module 1004, for according to the grid reference unit determined, extracting the network to have To the corresponding network feature information of figure.

In the embodiment of the present application, the determining module 1002, is specifically used for: being directed to each specified node, indicates that this refers to Determine node to send a notification message to immediate neighbor node, receives all immediate neighbor node feedbacks for having been received by the notification message Report information all immediate neighbor nodes of each specified node are determined according to the report information.

The determining module 1002, is specifically used for: indicating that the specified node carries the identification information of itself and believes in notice In breath, and the notification message is sent to all immediate neighbor nodes of the specified node, determine it is being carried in report information, Other nodes corresponding to the other identifier information in addition to specifying the identification information of node itself, will other described nodes it is true It is set to the immediate neighbor node of the specified node.

Specifically, the grid reference unit module 1003, is specifically used for: according to the transmitting between each specified node Relationship constructs the nonoriented edge between each specified node and all immediate neighbor nodes, and is each nonoriented edge assignment, after assignment Each nonoriented edge, determine the all-network reference cell formed between each specified node.

The grid reference unit module 1003, is specifically used for: the identification information of all specified nodes is determined, according to specified Sequence is ranked up all identification informations determined, the use of building rule is each undirected according to the identification information after sequence Side assignment.

Wherein, the building rule includes: to sort different nodes for two identification informations, when between two nodes Transitive relation be sorted by identification information the higher lower node of node point identification information sorting directed edge when, then by structure The nonoriented edge between two nodes after building is assigned a value of the first numerical value；

The grid reference unit module 1003, is specifically used for: being directed to each specified node, indicates that the specified node should The link information of specified node itself and all immediate neighbor nodes is sent respectively to each straight in addition to the immediate neighbor node Neighbor node is connect, and receives the link information that all immediate neighbor nodes are sent, all companies received according to each specified node Information is connect, determines the corresponding all indirect neighbor nodes of each specified node, constructs the specified node and the specified node pair The nonoriented edge for all indirect neighbor nodes answered, and each nonoriented edge assignment between specified node and each indirect inode, according to The specified node after the specified node after assignment and the nonoriented edge between all immediate neighbor nodes and assignment with Nonoriented edge between indirect neighbor node determines the all-network reference cell containing the specified node.

Wherein, the link information includes: the identification information of specified node and immediate neighbor node.

Wherein, the indirect neighbor node refers to is not connected directly with the specified node, specifies the direct of node with this The node that neighbor node is connected directly.

The grid reference unit module 1003, is specifically used for: the identification information of all specified nodes is determined, according to specified Sequence is ranked up all identification informations determined, the use of building rule is specified section according to the identification information after sequence Each nonoriented edge assignment between point and each indirect inode.

At this point, the building rule includes: when transitive relation is not present between two nodes, by two sections after building Nonoriented edge between point is assigned a value of the 4th numerical value.

The network feature information module 1004, is specifically used for: being directed to each specified node, determines to specify node with this For all nonoriented edge triangles on vertex, the corresponding value of each nonoriented edge triangle is determined, it is specified containing this as what is determined The all-network reference cell of node.

Described device further include: deduplication module 1005, it is each to what is determined for the identification information according to each specified node Grid reference unit carries out duplicate removal.

On the basis of above content, the network feature information module 1004 is specifically used for counting the directed graph In identical grid reference unit quantity, as the corresponding network feature information of the directed graph.

On the basis of the above, the embodiment of the present application also provides a kind of feature deriving means based on complex network, such as Figure 11 It is shown.Arithmetic facility sends instruction information for specified node each in directed graph, and described device includes:

Receiving module 1101, the instruction information sent for receiving the arithmetic facility；

Determining module 1102, for according to the instruction information, determining between the specified node and other specified nodes Transitive relation；

Grid reference unit module 1103, for determining the institute formed by the specified node according to the transitive relation There is grid reference unit；

Network feature information module 1104 is set for the grid reference unit determined to be fed back to the operation It is standby, so that the arithmetic facility extracts the corresponding network feature information of the directed graph.

The determining module 1102 is specifically used for specifying the immediate neighbor node of node to this according to the instruction information It sends a notification message, and the notification message for receiving the transmission of other immediate neighbor nodes determines the finger according to the notification message Determine all immediate neighbor nodes of node.

The grid reference unit module 1103, is specifically used for according to the transitive relation, construct the specified node with Nonoriented edge between all immediate neighbor nodes, and be each nonoriented edge assignment, according to each nonoriented edge after assignment, determine by described The all-network reference cell that specified node is formed.

The grid reference unit module 1103, specifically for extracting the sequencing information for including in instruction information；Wherein, institute Identification information of the sequencing information by the arithmetic facility according to all specified nodes is stated, according to specified sequence to all identification informations It is generated after being ranked up；It the use of building rule is each nonoriented edge assignment according to the sequencing information.

Under the scene, the building rule includes: to sort different nodes for two identification informations, when two nodes Between transitive relation be sorted by identification information the higher lower node of node point identification information sorting directed edge when, The nonoriented edge between two nodes after building is then assigned a value of the first numerical value；

The grid reference unit module 1003, specifically for distinguishing the link information with any immediate neighbor node It is sent to each immediate neighbor node in addition to the neighbor node, and receives the link information that all immediate neighbor nodes are sent, The specified node determines the corresponding institute of the specified node according to the link information of all immediate neighbor nodes received There is indirect neighbor node, construct the nonoriented edge of the specified corresponding all indirect neighbor nodes of node itself, and is each nonoriented edge Assignment, according to the specified node after assignment and the nonoriented edge between all immediate neighbor nodes and the finger after assignment Determine the nonoriented edge between node and indirect neighbor node, determines the all-network reference cell for specifying node to be formed by this.

Wherein, the link information includes: the value of the nonoriented edge constructed between specified node and immediate neighbor node, this refers to Determine the identification information of node and the identification information of the immediate neighbor node.

The grid reference unit module 1003, specifically for extracting the sequencing information for including in instruction information；Wherein, institute Identification information of the sequencing information by the arithmetic facility according to all specified nodes is stated, according to specified sequence to all identification informations It is generated after being ranked up；It the use of building rule is each nonoriented edge assignment according to the sequencing information；

In this scenario, the building rule includes: when transitive relation is not present between two nodes, then after constructing Two nodes between nonoriented edge be assigned a value of the 4th numerical value.

The grid reference unit module 1003, specifically for determining using all nonoriented edges for specifying node as vertex Triangle determines the corresponding value of each nonoriented edge triangle, as the all-network benchmark for specifying node containing this determined Unit, the corresponding value of all nonoriented edge triangles that will be determined, be sent to the specified node all immediate neighbor nodes and Indirect neighbor node, and receive that all immediate neighbor nodes and indirect neighbor node send by immediate neighbor node and adjoining The set for the corresponding value of all nonoriented edge triangles that node determines is occupied, it is corresponding according to all nonoriented edge triangles received The corresponding value of all nonoriented edge triangles that value and the specified node itself determine, carries out duplicate removal, according to undirected after duplicate removal The value of side triangle determines the all-network reference cell comprising the specified node.

In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net Network interface and memory.

Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium Example.

Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data. The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM), Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.

It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including described want There is also other identical elements in the process, method of element, commodity or equipment.

It will be understood by those skilled in the art that embodiments herein can provide as method, system or computer program product. Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application Form.It is deposited moreover, the application can be used to can be used in the computer that one or more wherein includes computer usable program code The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) Formula.

The above description is only an example of the present application, is not intended to limit this application.For those skilled in the art For, various changes and changes are possible in this application.All any modifications made within the spirit and principles of the present application are equal Replacement, improvement etc., should be included within the scope of the claims of this application.

Claims

1. a kind of feature extracting method based on complex network, which is characterized in that the described method includes:

Equipment is configured according to the corresponding directed graph of complex network, is distributed included in the directed graph for each arithmetic facility The corresponding nodal information of all nodes；

The nodal information is distributed to each arithmetic facility by the configuration equipment, so that section of each arithmetic facility according to distribution Specified node corresponding to point information, determines the transitive relation between each specified node, is determined according to the transitive relation Grid reference unit in the directed graph, and the corresponding net of the directed graph is extracted according to the grid reference unit Network characteristic information, wherein the grid reference unit is the basic structural unit for constituting the directed graph.

2. the method as described in claim 1, which is characterized in that the institute in for each arithmetic facility assignment of allocation directed graph Before all nodes for including, the method also includes:

Globally unique identification information is arranged for each node for including in the directed graph in the configuration equipment.

3. a kind of feature extracting method based on complex network, which is characterized in that the described method includes:

Arithmetic facility receives the nodal information sent by configuration equipment；The nodal information is the configuration equipment according to complex web The corresponding directed graph of network is arithmetic facility distribution；

According to the nodal information received, the transitive relation between the corresponding each specified node of the nodal information is determined；

According to the transitive relation between each specified node, the all-network benchmark list that each specified node is formed is determined Member, wherein the grid reference unit is the basic structural unit for constituting the directed graph；

4. method as claimed in claim 3, which is characterized in that determine between the corresponding each specified node of the nodal information Transitive relation specifically includes:

For each specified node, indicate that the specified node sends a notification message to immediate neighbor node；Wherein, the direct neighbour Node is occupied to refer to and the node for specifying node to be connected directly；

Receive the report information of all immediate neighbor node feedbacks for having been received by the notification message；

According to the report information, all immediate neighbor nodes of each specified node are determined.

5. method as claimed in claim 4, which is characterized in that indicate that the specified node sends notice to immediate neighbor node and disappears Breath, specifically includes:

It indicates that the specified node carries the identification information of itself in notification information, and the notification message is sent to this and is referred to Determine all immediate neighbor nodes of node；

According to the report information, determines all immediate neighbor nodes of each specified node, specifically includes:

Determine it is being carried in report information, corresponding to the other identifier information in addition to specifying the identification information of node itself Other described nodes are determined as the immediate neighbor node of the specified node by other nodes.

6. method as claimed in claim 4, which is characterized in that according to the transitive relation between each specified node, determine The all-network reference cell that each specified node is formed, specifically includes:

According to the transitive relation between each specified node, the nothing between each specified node and all immediate neighbor nodes is constructed Xiang Bian, and be each nonoriented edge assignment；

According to each nonoriented edge after assignment, the all-network reference cell formed between each specified node is determined.

7. method as claimed in claim 6, which is characterized in that be each nonoriented edge assignment, specifically include:

Determine the identification information of all specified nodes；

All identification informations determined are ranked up according to specified sequence；

It the use of building rule is each nonoriented edge assignment according to the identification information after sequence；

Wherein, the building rule includes:

Sort different nodes for two identification informations, when the transitive relation between two nodes be by identification information sequence compared with When the directed edge of the high lower node of node point identification information sorting, then by the nonoriented edge between two nodes after building It is assigned a value of the first numerical value；

When the transitive relation between two nodes is higher by the identification information lower node point identification information sorting that sorts When the directed edge of node, then the nonoriented edge between two nodes after building is assigned a value of second value；

When the transitive relation between two nodes is two-way side, then the nonoriented edge between two nodes after building is assigned a value of Third value.

8. method as claimed in claim 6, which is characterized in that according to each nonoriented edge after assignment, determine each specified node it Between the all-network reference cell that is formed, specifically include:

For each specified node, indicate that this is specified the connection of node itself and each immediate neighbor node to believe by the specified node Breath, is sent respectively to other immediate neighbor nodes in addition to the immediate neighbor node, and receive all immediate neighbor nodes The link information of transmission；Wherein, the link information includes: the identification information of specified node and immediate neighbor node；

According to all link informations that each specified node receives, the corresponding all indirect neighbor sections of each specified node are determined Point；Wherein, the indirect neighbor node refers to is not connected directly with the specified node, and the immediate neighbor section of node is specified with this The node that point is connected directly；

Construct the nonoriented edge of the specified node all indirect neighbor nodes corresponding with the specified node, and for specified node with Each nonoriented edge assignment between each indirect inode；

According to the specified node after assignment and the nonoriented edge between all immediate neighbor nodes and the finger after assignment Determine the nonoriented edge between node and indirect neighbor node, determines the all-network reference cell containing the specified node.

9. method according to claim 8, which is characterized in that each nonoriented edge between specified node and each indirect inode is assigned Value, specifically includes:

Determine the identification information of all specified nodes；

According to the identification information after sequence, assigned using each nonoriented edge of the building rule between specified node and each indirect inode Value；

Wherein, the building rule includes:

When transitive relation is not present between two nodes, the nonoriented edge between two nodes after building is assigned a value of the 4th number Value.

10. method as claimed in claim 9, which is characterized in that determine the all-network formed containing the specified node Reference cell specifically includes:

For each specified node, determine to determine each undirected using all nonoriented edge triangles for specifying node as vertex The corresponding value of side triangle, as the all-network reference cell for specifying node containing this determined；Wherein, a nonoriented edge The corresponding value of triangle are as follows: the value that the assignment of every nonoriented edge of the nonoriented edge triangle is composed.

11. method as claimed in claim 10, which is characterized in that according to the grid reference unit determined, extract institute Before stating the corresponding network feature information of directed graph, the method also includes:

The arithmetic facility carries out duplicate removal to each grid reference unit determined according to the identification information of each specified node.

12. method as claimed in claim 11, which is characterized in that extract the corresponding network characterization letter of the directed graph Breath, specifically includes:

The quantity for counting identical grid reference unit in the directed graph, as the corresponding network of the directed graph Characteristic information.

13. a kind of feature extracting method based on complex network, which is characterized in that arithmetic facility is for each in directed graph Specified node sends instruction information, which comprises

According to the transitive relation, the all-network reference cell formed by the specified node is determined, wherein the network base Quasi- unit is the basic structural unit for constituting the directed graph；

The grid reference unit determined is fed back into the arithmetic facility, so that the arithmetic facility extracts the net The corresponding network feature information of network digraph.

14. method as claimed in claim 13, which is characterized in that according to the instruction information, determine the specified node and its Transitive relation between his specified node, specifically includes:

The specified node specifies the immediate neighbor node of node to send a notification message according to the instruction information, to this, and connects Receive the notification message of other immediate neighbor nodes transmission；

According to the notification message, all immediate neighbor nodes of the specified node are determined.

15. method as claimed in claim 13, which is characterized in that according to the transitive relation, determine by the specified node The all-network reference cell of formation, specifically includes:

According to the transitive relation, the nonoriented edge between the specified node and all immediate neighbor nodes is constructed, and is each nothing To side assignment；

According to each nonoriented edge after assignment, the all-network reference cell formed by the specified node is determined.

16. method as claimed in claim 15, which is characterized in that be each nonoriented edge assignment, specifically include:

Extract the sequencing information for including in instruction information；Wherein, the sequencing information is by the arithmetic facility according to all specified The identification information of node generates after being ranked up according to specified sequence to all identification informations；

It the use of building rule is each nonoriented edge assignment according to the sequencing information；

Wherein, the building rule includes:

17. method as claimed in claim 15, which is characterized in that according to each nonoriented edge after assignment, determine by described specified The all-network reference cell that node is formed, specifically includes:

Link information with any immediate neighbor node is sent respectively to each in addition to the neighbor node by the specified node Immediate neighbor node, and receive the link information that all immediate neighbor nodes are sent；Wherein, the link information includes: specified The identification information and the immediate neighbor node of the value of the nonoriented edge constructed between node and immediate neighbor node, the specified node Identification information；

The specified node determines that the specified node is corresponding according to the link information of all immediate neighbor nodes received All indirect neighbor nodes；Wherein, the indirect neighbor node refers to is not connected directly with the specified node, specified with this The node that the immediate neighbor node of node is connected directly；

The nonoriented edge of the specified corresponding all indirect neighbor nodes of node itself is constructed, and is each nonoriented edge assignment；

According to described specified after the specified node after assignment and the nonoriented edge between all immediate neighbor nodes and assignment Nonoriented edge between node and indirect neighbor node determines the all-network reference cell for specifying node to be formed by this.

18. method as claimed in claim 17, which is characterized in that be each nonoriented edge assignment, specifically include:

Wherein, the building rule includes: when transitive relation is not present between two nodes, then by two nodes after building Between nonoriented edge be assigned a value of the 4th numerical value.

19. method as claimed in claim 18, which is characterized in that according to after assignment the specified node and all direct neighbours Occupy the nonoriented edge between node and the specified node after assignment and the nonoriented edge between indirect neighbor node, determine by The all-network reference cell that the specified node is formed, specifically includes:

The specified node is determined to determine each nonoriented edge three using all nonoriented edge triangles for specifying node as vertex Angular corresponding value, as the all-network reference cell for specifying node containing this determined, wherein a nonoriented edge triangle The corresponding value of shape are as follows: the value that the assignment of every nonoriented edge of the nonoriented edge triangle is composed；

The corresponding value of all nonoriented edge triangles that will be determined, be sent to the specified node all immediate neighbor nodes and Connect neighbor node, and receive that all immediate neighbor nodes and indirect neighbor node send by immediate neighbor node and indirect neighbor The set for the corresponding value of all nonoriented edge triangles that node determines；

All nonoriented edges three determined according to the corresponding value of all nonoriented edge triangles received and the specified node itself Angular corresponding value carries out duplicate removal；

According to the value of the nonoriented edge triangle after duplicate removal, the all-network reference cell comprising the specified node is determined.

20. a kind of feature deriving means based on complex network, which is characterized in that described device includes:

Configuration module, for being distributed in the directed graph for each arithmetic facility according to the corresponding directed graph of complex network The corresponding nodal information of all nodes for being included；

Sending module, for the nodal information to be distributed to each arithmetic facility, so that each arithmetic facility is according to distribution Specified node corresponding to nodal information determines the transitive relation between each specified node, is determined according to the transitive relation Grid reference unit in the directed graph out, and it is corresponding according to the grid reference unit to extract the directed graph Network feature information, wherein the grid reference unit is the basic structural unit for constituting the directed graph.

21. device as claimed in claim 20, which is characterized in that described device further include: mark module, for for described Globally unique identification information is arranged in each node for including in directed graph.

22. a kind of feature deriving means based on complex network, which is characterized in that described device includes:

Receiving module, for receiving the nodal information sent by configuration equipment；The nodal information be the configuration equipment according to The corresponding directed graph distribution of complex network；

Determining module, for according to the nodal information that receives, determine the corresponding each specified node of the nodal information it Between transitive relation；

Grid reference unit module, for determining each specified node according to the transitive relation between each specified node The all-network reference cell of formation；Wherein, the grid reference unit is the basic structure list for constituting the directed graph Member；

Network feature information module, for it is corresponding to extract the directed graph according to the grid reference unit determined Network feature information.

23. device as claimed in claim 22, which is characterized in that the determining module is specifically used for:

For each specified node, indicates that the specified node sends a notification message to immediate neighbor node, receive all have been received by The report information of the immediate neighbor node feedback of the notification message determines each specified section according to the report information All immediate neighbor nodes of point；Wherein, the immediate neighbor node refers to and the node for specifying node to be connected directly.

24. device as claimed in claim 23, which is characterized in that the determining module is specifically used for: indicating the specified node The identification information of itself is carried in notification information, and the notification message is sent to all direct neighbours of the specified node Occupy node, determine it is being carried in report information, except the other identifier information institute in addition to specifying the identification information of node itself is right Other described nodes are determined as the immediate neighbor node of the specified node by other nodes answered.

25. device as claimed in claim 23, which is characterized in that the grid reference unit module is specifically used for: according to each The transitive relation between specified node, constructs the nonoriented edge between each specified node and all immediate neighbor nodes, and be Each nonoriented edge assignment determines the all-network reference cell formed between each specified node according to each nonoriented edge after assignment.

26. device as claimed in claim 24, which is characterized in that the grid reference unit module is specifically used for: determining institute There is the identification information of specified node, all identification informations determined is ranked up according to specified sequence, after sequence Identification information is each nonoriented edge assignment using building rule；

Wherein, the building rule includes: to sort different nodes for two identification informations, the transmitting between two nodes Relationship be sorted by identification information the higher lower node of node point identification information sorting directed edge when, then will building after Two nodes between nonoriented edge be assigned a value of the first numerical value；

27. device as claimed in claim 24, which is characterized in that the grid reference unit module is specifically used for: for every One specified node indicates that this is specified the link information of node itself Yu each immediate neighbor node by the specified node, sends out respectively Other immediate neighbor nodes in addition to the immediate neighbor node are given, and receive the connection that all immediate neighbor nodes are sent Information determines the corresponding all indirect neighbor sections of each specified node according to all link informations that each specified node receives Point, constructs the nonoriented edge of the specified node all indirect neighbor nodes corresponding with the specified node, and for specified node and Each nonoriented edge assignment between each indirect inode, according to the specified node after assignment and between all immediate neighbor nodes The specified node after nonoriented edge and assignment and the nonoriented edge between indirect neighbor node are determined containing described specified The all-network reference cell of node；

Wherein, the link information includes: the identification information of specified node and immediate neighbor node；

Wherein, the indirect neighbor node refers to is not connected directly with the specified node, and the immediate neighbor of node is specified with this The node that node is connected directly.

28. device as claimed in claim 27, which is characterized in that the grid reference unit module is specifically used for: determining institute There is the identification information of specified node, all identification informations determined is ranked up according to specified sequence, after sequence Identification information uses each nonoriented edge assignment of the building rule between specified node and each indirect inode；

Wherein, the building rule include: when between two nodes be not present transitive relation when, by two nodes after building it Between nonoriented edge be assigned a value of the 4th numerical value.

29. device as claimed in claim 28, which is characterized in that the grid reference unit module is specifically used for: for every One specified node is determined to determine each nonoriented edge triangle pair using all nonoriented edge triangles for specifying node as vertex The value answered, as the all-network reference cell for specifying node containing this determined；Wherein, a nonoriented edge triangle pair is answered Value are as follows: the value that the assignment of every nonoriented edge of the nonoriented edge triangle is composed.

30. device as claimed in claim 28, which is characterized in that described device further include: deduplication module, for according to each finger The identification information for determining node carries out duplicate removal to each grid reference unit determined.

31. device as claimed in claim 29, which is characterized in that the network feature information module is specifically used for statistics institute The quantity for stating identical grid reference unit in directed graph, as the corresponding network feature information of the directed graph.

32. a kind of feature deriving means based on complex network, which is characterized in that arithmetic facility is for each in directed graph Specified node sends instruction information, and described device includes:

Determining module, for according to the instruction information, determining the transitive relation between the specified node and other specified nodes；

Grid reference unit module, for determining the all-network base formed by the specified node according to the transitive relation Quasi- unit；Wherein, the grid reference unit is the basic structural unit for constituting the directed graph；

Network feature information module, for the grid reference unit determined to be fed back to the arithmetic facility, so that The arithmetic facility extracts the corresponding network feature information of the directed graph.

33. device as claimed in claim 32, which is characterized in that the determining module is specifically used for being believed according to the instruction Breath, specifies the immediate neighbor node of node to send a notification message to this, and the notice for receiving the transmission of other immediate neighbor nodes disappears Breath, according to the notification message, determines all immediate neighbor nodes of the specified node.

34. device as claimed in claim 32, which is characterized in that the grid reference unit module is specifically used for according to institute Transitive relation is stated, the nonoriented edge between the specified node and all immediate neighbor nodes is constructed, and is each nonoriented edge assignment, root According to each nonoriented edge after assignment, the all-network reference cell formed by the specified node is determined.

35. device as claimed in claim 34, which is characterized in that the grid reference unit module refers to specifically for extracting Show the sequencing information for including in information；Wherein, the sequencing information is by the arithmetic facility according to the mark of all specified nodes Information generates after being ranked up according to specified sequence to all identification informations；According to the sequencing information, it is using building rule Each nonoriented edge assignment；

36. device as claimed in claim 34, which is characterized in that the grid reference unit module, be specifically used for will with appoint The link information of one immediate neighbor node is sent respectively to each immediate neighbor node in addition to the neighbor node, and receives institute The link information for having immediate neighbor node to send, the specified node are believed according to the connection of all immediate neighbor nodes received Breath, determines the corresponding all indirect neighbor nodes of the specified node, and it is corresponding all indirect to construct the specified node itself The nonoriented edge of neighbor node, and be each nonoriented edge assignment, according to the specified node and all immediate neighbor nodes after assignment Between nonoriented edge and the specified node and indirect neighbor node after assignment between nonoriented edge, determine to be specified by this The all-network reference cell that node is formed；

Wherein, the link information includes: the value of the nonoriented edge constructed between specified node and immediate neighbor node, the specified section The identification information of point and the identification information of the immediate neighbor node；

37. device as claimed in claim 36, which is characterized in that the grid reference unit module refers to specifically for extracting Show the sequencing information for including in information；Wherein, the sequencing information is by the arithmetic facility according to the mark of all specified nodes Information generates after being ranked up according to specified sequence to all identification informations；According to the sequencing information, it is using building rule Each nonoriented edge assignment；

38. device as claimed in claim 37, which is characterized in that the grid reference unit module, specifically for determining All nonoriented edge triangles of the node as vertex are specified using this, determine the corresponding value of each nonoriented edge triangle, as determining The all-network reference cell that node is specified containing this, the corresponding value of all nonoriented edge triangles that will be determined is sent to All immediate neighbor nodes and indirect neighbor node of the specified node, and receive all immediate neighbor nodes and indirect neighbor section The set for the corresponding value of all nonoriented edge triangles determined by immediate neighbor node and indirect neighbor node that point is sent, according to All nonoriented edge triangle pairs that the corresponding value of all nonoriented edge triangles received and the specified node itself determine are answered Value, carry out duplicate removal and the all-network base comprising the specified node determined according to the value of the nonoriented edge triangle after duplicate removal Quasi- unit；

Wherein, the corresponding value of a nonoriented edge triangle are as follows: the assignment of every nonoriented edge of the nonoriented edge triangle is composed Value.