System and method for roughening picture
System and method for roughening picture
 CN 101,324,937 B
 Filed: 06/15/2007
 Issued: 05/20/2015
 Est. Priority Date: 06/15/2007
 Status: Active Grant
First Claim
1. one kind for carrying out alligatoring to find the network analysis method of community in a network to figure, wherein said network is computer network, described network is modeled as figure, described figure comprises multiple summit, wherein, one is comprised to the computer network of multiple webpage, each webpage can be counted as a summit, hyperlink between each webpage can be counted as limit, and described method comprises:
 Roughening step, comprising;
A) the merging modularity yield value between described current vertex and its each adjacent vertex is calculated to a current vertex;
B) similarity between described current vertex and its each adjacent vertex is calculated;
C) based on the merging modularity yield value calculated and similarity, determine whether described current vertex can be adjacent vertex merge, and determining to merge when can merge,Refinement step, comprising;
To the figure refinement of the alligatoring that abovementioned roughening step obtains, to find the community in described network,Wherein described in modularity value formulae discovery, merge modularity yield value and similarity,Wherein step a) also comprises;
Calculate the merging modularity yield value Δ
Q of described current vertex and one of them adjacent vertex described in the following way _{c};
the modularity value Q calculating the figure be made up of described current vertex and one of them adjacent vertex described _{c}, the modularity value Q of figure to be made up of described current vertex _{a}and the modularity value Q of the figure to be made up of one of them adjacent vertex described _{b}, and calculate Δ
Q _{c}=Q _{c}Q _{a}Q _{b}, and After calculating described merging modularity yield value to each adjacent vertex of described current vertex, determine to merge the maximum summit of modularity yield value described in each adjacent vertex,Wherein step c) also comprise;
merging the maximum adjacent vertex of modularity yield value and the maximum adjacent vertex of similarity by obtaining, determining whether described current vertex can be adjacent vertex merge,Wherein step c) also comprise;
Determine whether the adjacent vertex that described merging modularity yield value is maximum and the maximum adjacent vertex of described similarity are same adjacent vertex;
If so, then determine that described current vertex can be maximum adjacent vertex with described merging modularity yield value and similarity and merge.
Chinese PRB Reexamination
Abstract
The invention discloses a system and a method for coarsening a graph. The graph comprises a plurality of vertexes. The method comprises the following steps: (a) calculating merge modularity gain value between a vertex and each adjacent vertex thereof, aiming at the current vertex; (b) calculating similarity value between the current vertex and each adjacent vertex; and (c) determining whether the current vertex is merged into the adjacent vertexes according to the calculated merge modularity gain value and the similarity value; and merging when merging can be conducted. The method can quickly and accurately coarsen the graph.
13 Claims

1. one kind for carrying out alligatoring to find the network analysis method of community in a network to figure, wherein said network is computer network, described network is modeled as figure, described figure comprises multiple summit, wherein, one is comprised to the computer network of multiple webpage, each webpage can be counted as a summit, hyperlink between each webpage can be counted as limit, and described method comprises:

Roughening step, comprising; A) the merging modularity yield value between described current vertex and its each adjacent vertex is calculated to a current vertex; B) similarity between described current vertex and its each adjacent vertex is calculated; C) based on the merging modularity yield value calculated and similarity, determine whether described current vertex can be adjacent vertex merge, and determining to merge when can merge, Refinement step, comprising; To the figure refinement of the alligatoring that abovementioned roughening step obtains, to find the community in described network, Wherein described in modularity value formulae discovery, merge modularity yield value and similarity, Wherein step a) also comprises; Calculate the merging modularity yield value Δ
Q of described current vertex and one of them adjacent vertex described in the following way _{c};
the modularity value Q calculating the figure be made up of described current vertex and one of them adjacent vertex described _{c}, the modularity value Q of figure to be made up of described current vertex _{a}and the modularity value Q of the figure to be made up of one of them adjacent vertex described _{b}, and calculate Δ
Q _{c}=Q _{c}Q _{a}Q _{b}, andAfter calculating described merging modularity yield value to each adjacent vertex of described current vertex, determine to merge the maximum summit of modularity yield value described in each adjacent vertex, Wherein step c) also comprise;
merging the maximum adjacent vertex of modularity yield value and the maximum adjacent vertex of similarity by obtaining, determining whether described current vertex can be adjacent vertex merge,Wherein step c) also comprise; Determine whether the adjacent vertex that described merging modularity yield value is maximum and the maximum adjacent vertex of described similarity are same adjacent vertex; If so, then determine that described current vertex can be maximum adjacent vertex with described merging modularity yield value and similarity and merge.


2. method according to claim 1, described roughening step also comprises:
D) the adjacent list of having accessed summit is upgraded.

3. method according to claim 2, described roughening step also comprises:
 perform abovementioned steps a)d iteratively to each summit in figure), until have accessed all summits in figure, obtain the figure of current one deck alligatoring.

4. method according to claim 3, described roughening step also comprises:
 performing before execution step a) is the step that a stochastic ordering is distributed on each summit, and often obtain the figure of one deck alligatoring, stochastic ordering is distributed on each summit be in the figure of this layer of alligatoring.

5. method according to claim 4, described roughening step also comprises:
When number of vertices in the figure of current one deck alligatoring is less than the number of vertices in the figure of last layer alligatoring, continue to perform lower one deck alligatoring.

6. method according to claim 4, described roughening step also comprises:
When number of vertices in the figure of current one deck alligatoring equals the number of vertices in the figure of last layer alligatoring, export the atlas of the figure of each layer alligatoring.

7. method according to claim 1, wherein step c) also comprise:
If determine that the adjacent vertex that described merging modularity yield value is maximum and the maximum adjacent vertex of described similarity are not same adjacent vertexes, change the stochastic ordering of described current vertex.

8. method according to claim 1, wherein only when described current vertex has the merging modularity yield value being greater than 0, just performs the step calculating similarity.

9. one kind for carrying out alligatoring to find the network analysis system of community in a network to figure, wherein said network is computer network, described network is modeled as figure, described figure comprises multiple summit, wherein, one is comprised to the computer network of multiple webpage, each webpage can be counted as a summit, hyperlink between each webpage can be counted as limit, and described system comprises:

Alligatoring device, comprising; Initial alligatoring device, for according to stochastic ordering, calculates the modularity yield value between described current vertex and its each adjacent vertex to current vertex; Deviation adjusting device, calculates the similarity between described current vertex and its each adjacent vertex; Wherein said system, based on the modularity yield value calculated and similarity, determines whether described current vertex can be adjacent vertex merge, and determining to merge when can merge, Device for thinning, comprising; For the figure refinement of alligatoring obtained abovementioned roughening step, to find the device of the community in described network, Wherein described in modularity value formulae discovery, merge modularity yield value and similarity, Wherein said initial alligatoring device also comprises device, for calculating the merging modularity yield value Δ
Q of described current vertex and one of them adjacent vertex described in the following way _{c};
the modularity value Q calculating the figure be made up of described current vertex and one of them adjacent vertex described _{c}, the modularity value Q of figure to be made up of described current vertex _{a}and the modularity value Q of the figure to be made up of one of them adjacent vertex described _{b}, and for calculating Δ
Q _{c}=Q _{c}Q _{a}Q _{b}, andFor after calculating described merging modularity yield value to each adjacent vertex of described current vertex, determine the device merging the maximum summit of modularity yield value described in each adjacent vertex, Wherein said alligatoring device also comprises; Merging the maximum adjacent vertex of modularity yield value and the maximum adjacent vertex of similarity by obtaining, determining whether described current vertex can be adjacent the device of vertex merge, Wherein said alligatoring device also comprises; Determine to merge the device whether maximum adjacent vertex of modularity yield value and the maximum adjacent vertex of similarity be same adjacent vertex; In response to determining to merge the maximum adjacent vertex of modularity yield value and whether the maximum adjacent vertex of similarity is same adjacent vertex, determine that described current vertex can be with described merging modularity yield value and similarity the device that maximum adjacent vertex merges further.


10. system according to claim 9, described alligatoring device also comprises:
Upgrade the device of having accessed the adjacent list on summit.

11. systems according to claim 9, described initial alligatoring device also comprises:
 for the device of a stochastic ordering is distributed on each summit.

12. systems according to claim 9, when the number of vertices wherein in the figure of current one deck alligatoring is less than the number of vertices in the figure of last layer alligatoring, continue lower one deck alligatoring.

13. systems according to claim 9, when the number of vertices wherein in the figure of current one deck alligatoring equals the number of vertices in the figure of last layer alligatoring, export the atlas of the figure of each layer alligatoring.
Specification(s)