CN107018074B - A kind of network-in-dialing path method for fast searching based on flattening - Google Patents
A kind of network-in-dialing path method for fast searching based on flattening Download PDFInfo
- Publication number
- CN107018074B CN107018074B CN201710219469.8A CN201710219469A CN107018074B CN 107018074 B CN107018074 B CN 107018074B CN 201710219469 A CN201710219469 A CN 201710219469A CN 107018074 B CN107018074 B CN 107018074B
- Authority
- CN
- China
- Prior art keywords
- vertex
- network
- sub
- vector
- flattening
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
A kind of network-in-dialing path method for fast searching based on flattening, it is characterised in that: method includes the following steps: 1) constructing the sub-network on each vertex in network；2) by sub-network flattening obtained in step 1), flattening sub-network is obtained；And 3) utilize the approximate shortest path between flattening sub-network solution network any two points.Sub-network flattening is converted into set operation with path computing and combined by the present invention, improves inquiry computation rate, can be realized second grade response.
Description
Technical field
The present invention relates to the method for searching path in graph theory, and in particular to a kind of network-in-dialing path based on flattening is fast
Fast searching method.
Background technique
Being connected between vertex in network apart from path is a basic conception in graph theory, and is widely used in artificial intelligence
With a basic conception of Web Community.Because the distance between vertex indicates the correlation between vertex, it can be expressed as
It is intended to optimal other users or content with user's Social search.In context aware search, it can be used for distributing higher
Weight order give the closer webpage of webpage relationship with current accessed.Therefore in network shortest path analysis and calculating
Play a significant role to analysis network vertex incidence relation.
And as hardware, software technology develop rapidly, the scale of data is being skyrocketed through；Smart phone, GPS terminal and base
In the universal and development of location-based service, geographic information data is in explosive growth；The network social intercourses platform such as microblogging, wechat rises abruptly rapidly
It rises, brings the social data of magnanimity.In addition, the topological structure of data becomes more complicated；Traditional static map can only be coarse
Ground describes the problems in practical application model.With the development of science and technology and social progress, the modeling of this coarseness gradually without
Method meets the needs of application.Meanwhile the demand of real-time is also indispensable；The growth of data scale and the variation of topological structure
Cause to need to be implemented more operations when calculating path, in order to meet requirement of real-time, algorithm must be more efficiently.
Communication path based on flattening calculates, and finds path by didactic beta pruning breadth-first search, and
Vertex path indexing is constructed in conjunction with sub-network flattening.It may be implemented to obtain in grade time response second using the path indexing of building
The accurately incidence relation between inquiry target object.
Summary of the invention
In view of the above shortcomings of the prior art, the purpose of the present invention is to provide a kind of network-in-dialing road based on flattening
Diameter method for fast searching.This method utilizes the path rope of building more particularly to strategies such as set operation, graph search and search beta prunings
Drawing, which may be implemented, obtains the incidence relation between accurately inquiry target object in grade time response second.
According to an embodiment of the invention, providing a kind of network-in-dialing path method for fast searching based on flattening:
A kind of network-in-dialing path method for fast searching based on flattening, method includes the following steps:
1) each vertex v in network is constructed (with prune approach)iSub-network Gi；
2) by sub-network G obtained in step 1)iFlattening obtains flattening sub-network；With
3) approximate shortest path between network any two points is solved using flattening sub-network.
In the present invention, in the step 1) sub-network construction method specifically:
1.1) vertex v is traversediAll of its neighbor vertex, if adjacent vertex vk(wherein vk ∈ Vi, set ViFor vertex viNeighbour
Connect vertex set) and vertex viBetween side eikWeight wikGreater than threshold alpha, then set V is addedi1, set V ' is otherwise addedi1；
Wherein, side right value is greater than the adjacent vertex set V of threshold alphai1As vertex viSub-network GiFirst layer, and remember set Vi1Top
Point vj(vj∈Vi1) about vertex viPath be pij=(i:j), and set V 'i1Vertex beta pruning fall, i.e., without following
The step of；
1.2) opposite vertexes set Vi1In any vertex vk, traverse vkAll of its neighbor vertex, if adjacent vertex vm(wherein vm
∈Vk, set VkFor vertex vkAdjacent vertex set) and vertex vkBetween side ekmWeight wkmGreater than threshold alpha and vertex vm
It is not present in sub-network GiIn (i.e.), then set V is addedi2, set V ' is otherwise addedi2；Wherein, adjacent vertex set
Vi2As vertex viSub-network GiThe second layer, and remember set Vi2Vertex vl(vl∈Vi2) about initial vertax viPath be
pil=pik+ (k:l)=(i:k:l), and set V 'i2Vertex beta pruning fall, i.e., without subsequent operating procedure；
1.3) extended mode in step 1.2) is repeated, by vertex viSub-network GiExpand to the maximum number of plies.
In the present invention, by the method for sub-network flattening in the step 2) specifically:
2.1) whole vertex in sorting network, obtain vertex sequence S；
2.2) by vertex v obtained in step 1)iSub-network GiEach layer vertex set Vi1, Vi2, Vi3..., VinPoint
It is not converted into vector representation, it is as follows that collection is bonded to vector conversion process: setting vertex set ViIn vertex be (v1, v2,
v3..., vs), and query set ViSerial number (I of the middle vertex in sequence S1, I2, I3..., Is)；If dimension is | | S | | (| | S |
| be sequence length, i.e., all vertex numbers) vectorVector I is tieed up into (I ∈ { I respectively1, I2, I3..., Is) numerical value
It is set as 1, other dimension numerical value are set as 0, and the obtained vector is set ViVector indicate；Similarly convert the institute of sub-network
There is set, after conversion, sub-network GiIt can be expressed as by one group of vector
2.3) to each layer vertex set Vi1, Vi2, Vi3..., VinUnion is sought, to realize sub-network GiFlattening, i.e., will
Gained vector in step 2.2)Each dimension step-by-step ask or operation, the son after obtaining flattening
Network
In the present invention, the method that the approximate shortest path between network any two points is solved in the step 3) is specific
Are as follows:
3.1) assume that two vertex are respectively vi、vj, their flattening sub-network is respectivelyAll connectivity points between two o'clock are the intersection on two sub-network vertex, i.e., by flat beggar
Each dimension step-by-step and operation of network vector, obtain
3.2) path between vertex: vertex v is acquired from the connectivity points between vertexiSub-network each layer vertex set
Intersection operation, sub-network are done with the conjunction of connected vertex set obtained in step 3.1) respectively
Each layer vertex vectorBe connected to vertex vectorEach dimension step-by-step ask and operation, obtainIf vectorEach dimension is not all zero, then in vertex viThere is connection vertex v in the kth layer of sub-networkiWith vj's
Connectivity points, if vectorM dimension be not zero, then vertex vmIt is vertex viWith vjA connectivity points；It can must be pushed up by step 1)
Point viTo the vertex vmPath be pim, vertex vjTo the vertex vmPath be pjm, thus vertex viTo vertex vjYi Tiaolu
Diameter is pij=pim+pjm。
Preferably, vertex v in the step 1.3)iSub-network GiThe largest extension number of plies be 2-8 layers, preferably 3-6
Layer.
Preferably, co-occurrence frequency of the side right value between vertex indicates the degree of correlation between vertex.
Preferably, the threshold alpha is the 80% of maximum side right value.
Preferably, vector I dimension value is that " 1 " indicates vertex set V in the step 2.2)iMiddle i-th vertex exists
Exist in this layer of vertex set, numerical value is that " 0 " indicates to be not present.
Preferably, described to be ordered as being arranged on vertex whole in network by lexcographical order ascending order according to vertex value.
Preferably, the conversion for being converted to collection and being bonded to vector；Serial number I of the vertex in sequence S in set V,
Then the I dimension of vector is 1, other dimensions are 0.
Preferably, network is pre-processed before the sub-network on each vertex in carrying out step 1) building network.
Preferably, the pretreated method is beta pruning extended method and extensions path storage method.
In the present invention, the network refers to the figure being made of the line of several given points and connection two o'clock, this
Figure represents things with point commonly used to describe certain particular kind of relationship between certain things, indicates phase with the line of connection two o'clock
Answering has this relationship between two things.Such as the present invention is used for text search, then the network refers to using word as vertex, altogether
Existing frequency is undirected weighted graph constructed by the weight on side.It is to carry out a mistake of beta pruning processing that the side right value, which is greater than threshold alpha,
Filter condition, the value of threshold alpha or specific filter condition are depending on the circumstances.In the present invention, threshold alpha is maximum side right
The 80% of value.The unnecessary searching route that the beta pruning processing will be unsatisfactory for foregoing filtration condition is given up in advance, with simplification
It calculates, reduces computation complexity.
In the present invention, the pretreated method refers in step 1) and makes when the sub-network on each vertex in building network
Beta pruning processing method and extensions path storage method.
In the present invention, vertex viSub-network each layer vertex set respectively with vertex is connected to obtained in step 3.1)
Set does intersection operation, can quickly acquire connectivity points about vertex viThe affiliated level of sub-network, and gather step 1) storage
Path can quickly obtain connectivity points to vertex viCommunication path.It is stored when due to carrying out sub-network extension in step 1)
Routing information of the vertex to each Level Expand vertex, therefore path computing only needs inquiry, and distance can by hierarchical information
.
Compared with prior art, the device have the advantages that are as follows:
1, the present invention uses heuristic beta pruning breadth-first search, can effectively reduce path indexing amount；
2, multi-level network structure is converted into the single level structure of flattening by the present invention, is simplified and is calculated, and is improved and is calculated
Efficiency；
3, path search process is converted into set operation by the present invention, reduces computation complexity；
4, sub-network flattening is converted into set operation with path computing and combines by the present invention, improve inquiry calculate it is fast
Rate can be realized second grade response.
Detailed description of the invention
Fig. 1 is the schematic diagram of undirected weighted graph in the embodiment of the present invention 1；
Fig. 2 is the sub-network first layer schematic diagram constructed in the embodiment of the present invention 1；
Fig. 3 is the sub-network second layer schematic diagram constructed in the embodiment of the present invention 1；
Fig. 4 is two o'clock approximate shortest path schematic diagram in the embodiment of the present invention 1.
Specific embodiment
According to an embodiment of the invention, providing a kind of network-in-dialing path method for fast searching based on flattening:
A kind of network-in-dialing path method for fast searching based on flattening, method includes the following steps:
1) with each vertex v in prune approach building networkiSub-network Gi；
2) by sub-network G obtained in step 1)iFlattening obtains flattening sub-network；With
3) approximate shortest path between network any two points is solved using flattening sub-network.
In the present invention, in the step 1) sub-network construction method specifically:
1.1) vertex v is traversediAll of its neighbor vertex, if adjacent vertex vk(wherein vk∈Vi, set ViFor vertex viNeighbour
Connect vertex set) and vertex viBetween side eikWeight wikGreater than threshold alpha, then set V is addedi1, set V ' is otherwise addedi1；
Wherein, side right value is greater than the adjacent vertex set V of threshold alphai1As vertex viSub-network GiFirst layer, and remember set Vi1Top
Point vj(vj∈Vi1) about vertex viPath be pij=(i:j), and set V 'i1Vertex beta pruning fall, i.e., without following
The step of；
1.2) opposite vertexes set Vi1In any vertex vk, traverse vkAll of its neighbor vertex, if adjacent vertex vm(wherein vm
∈Vk, set VkFor vertex vkAdjacent vertex set) and vertex vkBetween side ekmWeight wkmGreater than threshold alpha and vertex vm
It is not present in sub-network GiIn (i.e.), then set V is addedi2, set V ' is otherwise addedi2；Wherein, adjacent vertex set
Vi2As vertex viSub-network GiThe second layer, and remember set Vi2Vertex vl(vl∈Vi2) about initial vertax viPath be
pil=pik+ (k:l)=(i:k:l), and set V 'i2Vertex beta pruning fall, i.e., without subsequent operating procedure；
1.3) extended mode in step 1.2) is repeated, by vertex viSub-network GiExpand to the maximum number of plies.
In the present invention, by the method for sub-network flattening in the step 2) specifically:
2.1) whole vertex in sorting network, obtain vertex sequence S；
2.2) by vertex v obtained in step 1)iSub-network GiEach layer vertex set Vi1, Vi2, Vi3..., VinPoint
It is not converted into vector representation, it is as follows that collection is bonded to vector conversion process: setting vertex set ViIn vertex be (v1, v2,
v3..., vs), and query set ViSerial number (I of the middle vertex in sequence S1, I2, I3..., Is)；If dimension is | | S | | (| | S
| | be sequence length, i.e., all vertex numbers) vectorVector I is tieed up into (I ∈ { I respectively1, I2, I3..., Is) numerical value
It is set as 1, other dimension numerical value are set as 0, and the obtained vector is set ViVector indicate；Similarly convert the institute of sub-network
There is set, after conversion, sub-network GiIt can be expressed as by one group of vector
2.3) to each layer vertex set Vi1, Vi2, Vi3..., VinUnion is sought, to realize sub-network GiFlattening, i.e., will
Gained vector in step 2.2)Each dimension step-by-step ask or operation, the son after obtaining flattening
Network
In the present invention, the method that the approximate shortest path between network any two points is solved in the step 3) is specific
Are as follows:
3.1) assume that two vertex are respectively vi、vj, their flattening sub-network is respectivelyAll connectivity points between two o'clock are the intersection on two sub-network vertex, i.e., by flat beggar
Each dimension step-by-step and operation of network vector, obtain
3.2) path between vertex: vertex v is acquired from the connectivity points between vertexiSub-network each layer vertex set
Intersection operation, sub-network are done with the conjunction of connected vertex set obtained in step 3.1) respectively
Each layer vertex vectorBe connected to vertex vectorEach dimension step-by-step ask and operation, obtainIf vectorEach dimension is not all zero, then in vertex viThere is connection vertex v in the kth layer of sub-networkiWith vj's
Connectivity points, if vectorM dimension be not zero, then vertex vmIt is vertex viWith vjA connectivity points；It can must be pushed up by step 1)
Point viTo the vertex vmPath be pim, vertex vjTo the vertex vmPath be pjm, thus vertex viTo vertex vjYi Tiaolu
Diameter is pij=pim+pjm。
Preferably, vertex v in the step 1.3)iSub-network GiThe largest extension number of plies be 2-8 layers, preferably 3-6
Layer.
Preferably, co-occurrence frequency of the side right value between vertex indicates the degree of correlation between vertex.
Preferably, the threshold alpha is the 80% of maximum side right value.
Preferably, vector I dimension value is that " 1 " indicates vertex set V in the step 2.2)iMiddle i-th vertex exists
Exist in this layer of vertex set, numerical value is that " 0 " indicates to be not present.
Preferably, described to be ordered as being arranged on vertex whole in network by lexcographical order ascending order according to vertex value.
Preferably, the conversion for being converted to collection and being bonded to vector；Serial number I of the vertex in sequence S in set V,
Then the I dimension of vector is 1, other dimensions are 0.
Preferably, network is pre-processed before the sub-network on each vertex in carrying out step 1) building network.
Preferably, the pretreated method is beta pruning extended method and extensions path storage method.
Embodiment 1
By taking the use in certain medical associative search engine as an example, be specifically described that the present invention scans on associated diagram is answered
With:
System environments is as follows:
| Machine | Tide NF5270M3,10 |
| Operating system | Ubuntu 14.04 Server |
| Database | MySQL 5.5 |
| Development language | JAVA(Oracle JDK 8) |
| Distributed computing framework | Hadoop 2.6.0 |
Make pauses in reading unpunctuated ancient writings to the document in search collections of documents, obtain sentence set, distich subclass carries out participle and obtains word set
It closes, calculates the co-occurrence frequency in set of words between any two word, be stored in MySQL.Using word as vertex, co-occurrence frequency is side
Weight, construct undirected weighted graph, as shown in Figure 1.
A kind of network-in-dialing path method for fast searching based on flattening, method includes the following steps:
1) with the sub-network on each vertex in prune approach building network, for convenience of describing, with vertex " 1 " in Fig. 1 network
For the construction method of sub-network is described:
1.1) all of its neighbor vertex for traversing vertex " 1 " retains maximum 2 adjacent vertexes of side right value in adjacent vertex and adds
Enter set V11(i.e. vertex " 2 " and vertex " 3 ", V11={ 2,3 }), set V11As vertex " 1 " sub-network G1First layer, such as
Shown in Fig. 2, and remember path p of the first layer vertex about vertex " 1 "1j=(1:j), such as the routing table of vertex " 1 " to vertex " 2 "
It is shown as p12=(1:2), and other adjacent vertexes carry out beta pruning and fall, i.e., without following step,；
1.2) opposite vertexes set V11Any vertex, traverse its all of its neighbor vertex, retain in adjacent vertex side right value most
2 big adjacent vertexes and it is not present in sub-network G1In adjacent vertex set V12Subnet of={ 5,6,7 } as vertex " 1 "
Network G1The second layer, as shown in figure 3, and remembering second layer vertex νl(νl∈V12) about initial vertax " 1 " path be pil=pik+
(k:l)=(1:k:l), wherein vertex k is upper layer vertex, such as vertex " 5 " are p about the path of vertex " 1 "15=p12+ (2:
5)=(1:2)+(2:5)=(1:2:5), and the same beta pruning of other adjacent vertexes is fallen, i.e., without subsequent operating procedure.
2) by sub-network flattening obtained in step 1), its step are as follows:
2.1) whole vertex obtain orderly vertex set V (1 ∈ V) in sorting network；
2.2) by the sub-network G of vertex obtained in step 1) " 1 "1Each layer vertex set be converted into respectively vector expression
The vector of form, i.e. first layer vertex is expressed asSecond layer vertex
Vector is expressed asAfter conversion, sub-network G1Can from one group to
Amount indicates
2.3) union is asked to each layer vertex set, the sub-network to realize the flattening of sub-network, after flatteningI.e. by gained vector in step 2.2)Each dimension step-by-step ask or operation, obtain
3) approximate shortest path between network any two points is solved using flattening sub-network, its step are as follows:
3.1) assume that two vertex are respectively " 1 " and " 7 ", their flattening sub-network vector is respectivelyAll connectivity points between two o'clock are two sub-network tops
Each dimension step-by-step of flattening sub-network vector is asked and operation, is obtained by the intersection of point
3.2) path between vertex: each layer vertex set of the sub-network of vertex " 1 " is acquired from the connectivity points between vertex
It closes and is intersection operation, sub-network G with the conjunction of connected vertex set obtained in step 3.1) respectively1=((0,1,1,0,0,0,0), (0,
0,0,0,1,1,1) each layer vertex vector) be connected to vertex vectorEach dimension step-by-step ask with
Operation, i.e.,If vectorEach dimension is not all zero, then there is connection top in the kth layer of vertex " 1 " sub-network
The connectivity points of point " 1 " and " 7 ", if vectorM dimension be not zero, then vertex vmIt is a connectivity points of vertex " 1 " Yu " 7 ",
Thus the result of step 3.2) is combinedConnectivity points vector between obvious two o'clockWith son
Network G1First layer flattening vectorAsk with the result of operation be not the 0, the 3rd dimension for 1 to get vertex " 3 " be " 1 ",
Vertex on " 7 " two vertex communication path, and it is located at the first layer of the sub-network of vertex " 1 " and " 7 " simultaneously；It again can by step 1)
The path for obtaining vertex " 1 " to vertex " 3 " is p13, the path of vertex " 7 " to vertex " 3 " is p73；To vertex " 1 " to vertex " 7 "
Approximate shortest path be p17=p13+p73=(1:3:7), as shown in Figure 4.
Claims (15)
1. a kind of network-in-dialing path method for fast searching based on flattening, it is characterised in that: method includes the following steps:
1) with each vertex v in prune approach building networkiSub-network Gi；Specifically:
1.1) vertex v is traversediAll of its neighbor vertex, if adjacent vertex vkWith vertex viBetween side eikWeight wikGreater than threshold
Set V is then added in value αi1, set V ' is otherwise addedi1；Wherein: vk∈Vi, set ViFor vertex viAdjacent vertex set；Side right
Value is greater than the adjacent vertex set V of threshold alphai1As vertex viSub-network GiFirst layer, and remember set Vi1Vertex vjAbout
Vertex viPath Pij=(i:j), vj∈Vi1；And set V 'i1Vertex beta pruning fall, i.e., without following step；
1.2) opposite vertexes set Vi1In any vertex vk, traverse vkAll of its neighbor vertex, if adjacent vertex vmWith vertex vkBetween
Side ekmWeight wkmGreater than threshold alpha and vertex vmIt is not present in sub-network GiIn, i.e.,Set V is then addedi2, otherwise
Set V ' is addedi2；Wherein, vm∈Vk, set VkFor vertex vkAdjacent vertex set；Adjacent vertex set Vi2As vertex vi
Sub-network GiThe second layer, and remember set Vi2Vertex vlAbout initial vertax viPath be pil=Pik+ (k:l)=(i:k:
L), vl∈Vi2；And set V 'i2Vertex beta pruning fall, i.e., without subsequent operating procedure；
1.3) extended mode in step 1.2) is repeated, by vertex viSub-network GiExpand to the maximum number of plies；
2) by sub-network G obtained in step 1)iFlattening obtains flattening sub-network；Specifically:
2.1) whole vertex in sorting network, obtain vertex sequence S；
2.2) by vertex v obtained in step 1)iSub-network GiEach layer vertex set Vi1, Vi2, Vi3..., VinIt converts respectively
At vector representation, it is as follows that collection is bonded to vector conversion process: setting vertex set ViIn vertex be (v1, v2, v3..., vs), and
Query set ViSerial number (I of the middle vertex in sequence S1, I2, I3..., Is)；If dimension is | | S | | vector| | S | | be
Sequence length, i.e., all vertex numbers；The numerical value that vector I is tieed up is set as 1 respectively, other dimension numerical value are set as 0, such
The vector arrived is set ViVector indicate；Wherein: I ∈ { I1, I2, I3..., Is}；Similarly convert all set of sub-network, warp
After crossing conversion, sub-network GiIt can be expressed as by one group of vector
2.3) to each layer vertex set Vi1, Vi2, Vi3..., VinUnion is sought, to realize sub-network GiFlattening, i.e., by step
2.2) gained vector inEach dimension step-by-step ask or operation, the sub-network after obtaining flattening
3) approximate shortest path between network any two points is solved using flattening sub-network；Specifically:
3.1) assume that two vertex are respectively vi、vj, their flattening sub-network is respectively
All connectivity points between two o'clock are the intersection on two sub-network vertex, i.e., by each dimension step-by-step of flattening sub-network vector and fortune
It calculates, obtains
3.2) path between vertex: vertex v is acquired from the connectivity points between vertexiSub-network each layer vertex set difference
Intersection operation, sub-network are done with the conjunction of connected vertex set obtained in step 3.1)Each layer
Vertex vectorBe connected to vertex vectorEach dimension step-by-step ask and operation, obtainWherein: k=1,2,
3 ..., n；If vectorEach dimension is not all zero, then in vertex viThere is connection vertex v in the kth layer of sub-networkiWith vjConnection
Point, if vectorM dimension be not zero, then vertex vmIt is vertex viWith vjA connectivity points；Vertex v can be obtained by step 1)i
To the vertex vmPath be pim, vertex vjTo the vertex vmPath be pjm, thus vertex viTo vertex vjA paths be
pij=pim+pjm。
2. according to the method described in claim 1, it is characterized by: vertex v in the step 1.3)iSub-network GiMaximum
Extending the number of plies is 2-8 layers.
3. according to the method described in claim 2, it is characterized by: vertex v in the step 1.3)iSub-network GiMaximum
Extending the number of plies is 3-6 layers.
4. method according to any one of claim 1-3, it is characterised in that: co-occurrence of the side right value between vertex
Frequency indicates the degree of correlation between vertex；And/or
The threshold alpha is the 80% of maximum side right value.
5. method according to any one of claim 1-3, it is characterised in that: vector I dimension in the step 2.2)
Value is that " 1 " indicates vertex set ViMiddle i-th vertex exists in this layer of vertex set, and numerical value is that " 0 " indicates to be not present.
6. according to the method described in claim 4, it is characterized by: vector I dimension value is " 1 " expression in the step 2.2)
Vertex set ViMiddle i-th vertex exists in this layer of vertex set, and numerical value is that " 0 " indicates to be not present.
7. method described in any one of -3,6 according to claim 1, it is characterised in that: described to be ordered as net according to vertex value
Whole vertex are arranged by lexcographical order ascending order in network；And/or
The conversion for being converted to collection and being bonded to vector；Serial number I of the vertex in sequence S in set V, then the I of vector is tieed up
It is 1, other dimensions are 0.
8. according to the method described in claim 4, it is characterized by: described be ordered as vertex whole in network according to vertex value
It is arranged by lexcographical order ascending order；And/or
The conversion for being converted to collection and being bonded to vector；Serial number I of the vertex in sequence S in set V, then the I of vector is tieed up
It is 1, other dimensions are 0.
9. according to the method described in claim 5, it is characterized by: described be ordered as vertex whole in network according to vertex value
It is arranged by lexcographical order ascending order；And/or
The conversion for being converted to collection and being bonded to vector；Serial number I of the vertex in sequence S in set V, then the I of vector is tieed up
It is 1, other dimensions are 0.
10. according to claim 1-3,6, method described in any one of 8-9, it is characterised in that: construct net carrying out step 1)
In network before the sub-network on each vertex, network is pre-processed.
11. according to the method described in claim 4, it is characterized by: constructing the son on each vertex in network in progress step 1)
Before network, network is pre-processed.
12. according to the method described in claim 5, it is characterized by: constructing the son on each vertex in network in progress step 1)
Before network, network is pre-processed.
13. according to the method described in claim 7, it is characterized by: constructing the son on each vertex in network in progress step 1)
Before network, network is pre-processed.
14. according to the method described in claim 10, it is characterized by: the pretreated method is beta pruning extended method and expansion
Open up path storage method.
15. method described in any one of 1-13 according to claim 1, it is characterised in that: the pretreated method is beta pruning
Extended method and extensions path storage method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710219469.8A CN107018074B (en) | 2017-04-06 | 2017-04-06 | A kind of network-in-dialing path method for fast searching based on flattening |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710219469.8A CN107018074B (en) | 2017-04-06 | 2017-04-06 | A kind of network-in-dialing path method for fast searching based on flattening |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107018074A CN107018074A (en) | 2017-08-04 |
| CN107018074B true CN107018074B (en) | 2019-11-05 |
Family
ID=59445252
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710219469.8A Active CN107018074B (en) | 2017-04-06 | 2017-04-06 | A kind of network-in-dialing path method for fast searching based on flattening |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107018074B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110704434B (en) | 2019-09-24 | 2022-09-13 | 北京百度网讯科技有限公司 | Method and device for inquiring shortest path of map, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102662974A (en) * | 2012-03-12 | 2012-09-12 | 浙江大学 | A network graph index method based on adjacent node trees |
| CN105745882A (en) * | 2014-12-16 | 2016-07-06 | 北京大学深圳研究生院 | Realization method for flat networks routing based on network |
| CN106503789A (en) * | 2016-11-08 | 2017-03-15 | 西安电子科技大学宁波信息技术研究院 | Loop-free shortest path searching method based on Di Jiesitela and minimax ant colony |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9258222B2 (en) * | 2012-09-26 | 2016-02-09 | Avaya, Inc. | Implementing a shortest path bridging (SPB) network using a virtual routing and forwarding (VRF) identifier |
-
2017
- 2017-04-06 CN CN201710219469.8A patent/CN107018074B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102662974A (en) * | 2012-03-12 | 2012-09-12 | 浙江大学 | A network graph index method based on adjacent node trees |
| CN105745882A (en) * | 2014-12-16 | 2016-07-06 | 北京大学深圳研究生院 | Realization method for flat networks routing based on network |
| CN106503789A (en) * | 2016-11-08 | 2017-03-15 | 西安电子科技大学宁波信息技术研究院 | Loop-free shortest path searching method based on Di Jiesitela and minimax ant colony |
Non-Patent Citations (2)
| Title |
|---|
| Using Multi-objective Domain Optimization for Routing in Hierarchical Networks;Kyriakos Manousakis;《2005 International Conference on Wireless Networks, Communications and Mobile Computing》;20050616;第1-6页 * |
| 基于网络拓扑分析的分层规划算法研究;李梦杨;《中国优秀硕士学位论文全文数据库》;20140401;第1-50页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107018074A (en) | 2017-08-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dickerson et al. | Two-site Voronoi diagrams in geographic networks | |
| US20160232179A1 (en) | Recommending points of interests in a region | |
| CN109145169A (en) | A kind of address matching method based on statistics participle | |
| CN110837602A (en) | User recommendation method based on representation learning and multi-mode convolutional neural network | |
| CN106462624A (en) | Tile-based geocoder | |
| CN108446312B (en) | Optical remote sensing image retrieval method based on deep convolution semantic net | |
| CN109948066B (en) | Interest point recommendation method based on heterogeneous information network | |
| CN110442603A (en) | Address matching method, apparatus, computer equipment and storage medium | |
| CN104346444B (en) | A kind of the best site selection method based on the anti-spatial key inquiry of road network | |
| CN106326923B (en) | A kind of position data clustering method of registering taking position into account and repeating with density peaks point | |
| CN107766433A (en) | A kind of range query method and device based on Geo BTree | |
| CN106446162A (en) | Orient field self body intelligence library article search method | |
| CN109189959A (en) | A kind of method and device constructing image data base | |
| CN105260410A (en) | Microblog interestingness circle mining method based on intimacy degree and influence power and microblog interestingness circle mining device based on intimacy degree and influence power | |
| CN104657361A (en) | Data processing method and data processing device | |
| CN103377237B (en) | The neighbor search method of high dimensional data and fast approximate image searching method | |
| CN111815468B (en) | Multi-source social network construction method based on user identity association | |
| CN109344263A (en) | A kind of address matching method | |
| CN106203494A (en) | A kind of parallelization clustering method calculated based on internal memory | |
| CN105117494B (en) | Spatial entities mapping method in fuzzy context | |
| CN108446357A (en) | A kind of mass data spatial dimension querying method based on two-dimentional geographical location | |
| KR20220160534A (en) | Encoding method of geographical location zone, method and apparatus for constructing encoding model | |
| CN112069416B (en) | Cross-social network user identity recognition method based on community discovery | |
| CN105893486A (en) | Large-scale graph shortest distance indexing method based on cluster | |
| CN107016566A (en) | User model construction method based on body |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |