Influence Rating System

Abstract

To provide an influence evaluation system that evaluates influence on decision-making of an entity such as a company and a person.SOLUTION: An influence evaluation system 1, in network information whose influence information is a label that indicates an entity on a node, a direction of influence that the entity has on an edge, and information that indicates influence that an entity has on another entity, includes: a label update processing unit 12 that identifies an upper node of a node to be processed using an edge, and identifies an update label for updating the label of the node to be processed from the label of the upper node using the influence information; an intermediate influence calculation processing unit 13 that identifies a propagation path using the updated label identified by the label update processing unit 12 and calculates intermediate influence at a node; and an index calculation processing unit 14 that calculates an evaluation index indicating influence of the entity indicated by the label in the node using the intermediate influence in the node.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an influence evaluation system that evaluates the influence that an entity, such as a company or a person, has on other entities. [Background technology]

[0002] There is currently consideration being given to quantifying and evaluating the influence that a company or individual has over third parties.

[0003] One indicator of influence over third parties is stock ownership. Companies and individuals can influence the decision-making of a corporation by exercising the voting rights of the stocks they own at the general shareholders' meeting. The "Shapley-Shubik power index" in Non-Patent Document 1 below is known as a representative method for evaluating the influence that an entity such as a company or individual has over another entity.

[0004] The evaluation method of Non-Patent Document 1 is used as an index showing the direct influence between entities, for example, an index showing the influence based on the shareholding relationship. This is shown in Figure 31. Figure 31 shows the shareholding relationship of companies A to D and the index value of influence based on Non-Patent Document 1. Figure 31(a) shows that companies B and C each hold 30% of the shares of company A, and company D holds 40% of the shares of company A. Ordinary decision-making at a general meeting of shareholders of a stock company is made by ordinary resolution. In Japan, ordinary resolutions are passed at a general meeting of shareholders of a stock company when shareholders who hold a majority of the voting rights of shareholders who can exercise voting rights attend the meeting and a proposal is approved by a majority of the voting rights of the shareholders in attendance, unless otherwise specified in the articles of incorporation. Even if all shareholders who can exercise voting rights attend the meeting, if one person holds a majority of the shares that can exercise voting rights, it can be said that he or she can exert substantial influence on the decision-making of the company.

[0005] In Figure 31(a), none of Company B through Company D hold more than half of the shares on their own. Furthermore, any combination that results in a majority of Company A's shares can be achieved by teaming up with another company. In this case, Companies B through D have equal influence over Company A, and each can be evaluated as having 1 / 3. This is shown diagrammatically in Figure 31(b). Note that PI is an index of influence. Figure 31(c) shows the relationship between the indexes of influence as an equation.

[0006] In this way, the evaluation method of Non-Patent Document 1 is a method for evaluating the influence of only the direct relationship between two entities. Also, Non-Patent Documents 2 to 4, like Non-Patent Document 1, only consider the index of influence of only the direct relationship.

[0007] In addition, in all of Non-Patent Documents 1 to 4, since it is generally not known in advance whether or not voting rights will be exercised at the general shareholders' meeting, the influence on the entity is calculated on the assumption that all shareholders who are able to exercise voting rights will attend and exercise their voting rights. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] JP 2011-8375 A [Non-patent literature]

[0009] [Non-Patent Document 1] Shapley, Shubik, “Shapley-Shubik power index”, American Political Science Review, 48, p.787-792, 1954 [Non-Patent Document 2] Manfred J. Holler, Hannu Nurmi, “Power, Voting, and Voting Power:30 Years After”, Springer Link, 2013 [Non-Patent Document 3] JMGallardo, N. Jimenes, A. Jimenez-Losada, “A Shapley measure of power in hierarchies”, ELSEVIER Inc., 2016 [Non-Patent Document 4] Norkhairul Hafiz Bajuri, Shanti Chakravarty, Noor Hazarina Hashim,”ANALYSIS OF CORPORATE CONTROL: CAN THE VOTING POWER INDEX OUTSHINE SHAREHOLDING SIZE?”,AAMJAF, Vol.10, No.1, p75-94,2014 Summary of the Invention [Problem to be solved by the invention]

[0010] However, in the real world, there are control relationships such as parent companies, subsidiaries, and sub-subsidiaries, and the structure in which influence is exerted on other entities such as companies is more complicated. For example, as shown in Figure 32, there may be multiple hierarchical levels of relationships between entities, including shareholding relationships between companies, or circular relationships such as cross-shareholding. In the case of Figure 32, it can be evaluated that Company A is influenced by Company B through Company D, who directly hold shares, but Company C is further influenced by Company B and Company E. If Company B holds 80% of Company C's shares and Company E holds 20% of Company C's shares, only Company B can exert influence on Company C because Company B holds the majority of Company C's shares. In this case, the majority of Company A is the 30% of Company B's shares held directly and the 30% of Company B's shares held through Company C, which Company B can exert influence over, so in reality, only Company B can exert influence on Company A.

[0011] In this way, when a hierarchical structure or a circular relationship exists, it is not possible to perform evaluation using the conventional evaluation methods described in Non-Patent Documents 1 to 4.

[0012] Furthermore, Patent Document 1 is a system that performs cause analysis of management problems, etc., and is not a system that evaluates the influence of entities on each other.

[0013] Furthermore, even when using conventional technologies such as Patent Document 1 and Non-Patent Documents 1 to 4, it was not possible to evaluate or calculate the influence of an intermediate entity when there is a hierarchical structure or a circular relationship. For example, in the case of Figure 32, Company C holds shares in Company A and therefore has influence over Company A, but at the same time, Company C also has shares held by Company B and Company E, and is subject to the influence of these companies. And it was not possible to evaluate or calculate the influence mediated by Company C. [Means for solving the problem]

[0014] In view of the above problems, the inventors have invented an influence evaluation system that can index the influence that an entity has on other entities, particularly the influence of intermediate entities, even when the relationships between entities are complex, for example involving multiple hierarchies.

[0015] The first invention is an influence evaluation system having network information in which nodes are labels indicating entities, edges are the direction of influence the entities have, and influence information is information indicating the influence the entities have on other entities, the system comprising: a label update processing unit that identifies a higher-level node of a node to be processed using the edges and identifies an updated label for updating the label of the node to be processed from the label of the higher-level node using the influence information; an intermediate influence calculation processing unit that identifies a propagation path using the identified updated label and calculates an intermediate influence at the node; and an index calculation processing unit that calculates an evaluation index indicating the influence of the entity indicated by the label at the node using the intermediate influence at the node.

[0016] By configuring the present invention as described above, it is possible to index the influence that an entity has over other entities, particularly the influence of intermediate entities.

[0017] In the above-mentioned invention, the label update processing unit can be configured as an influence evaluation system that uses a label propagation method to identify the label to be updated in the node having the higher-level node.

[0018] In the case of a network with multiple hierarchical levels, the influence of an entity can be propagated by identifying the labels to be updated using the label propagation method as in the present invention, thereby indexing the influence on other entities.

[0019] In the above-mentioned invention, the label update processing unit can be configured as an influence evaluation system that identifies, by random sampling, the label to be updated in the node having the higher-level node.

[0020] It is not always certain which label of the higher-level node will affect a certain node. Therefore, by using random sampling as in the present invention, it is possible to perform a simulation.

[0021] In the above-mentioned invention, the label update processing unit can be configured as an influence evaluation system in which, when a calculated value calculated using influence information for a node above the node to be processed satisfies a condition for a predetermined threshold value, the label of the node above is identified as the update label of the node to be processed.

[0022] In the above-mentioned invention, the label update processing unit can be configured as an influence evaluation system that performs calculations using influence information for each entity for nodes higher than the node to be processed, and if the calculated value satisfies a condition for a predetermined threshold, identifies the label of the entity node as the update label of the node to be processed.

[0023] The propagation of influence to other entities can be achieved by implementing processes such as those of the present invention.

[0024] In the above-mentioned invention, the label update processing unit can be configured as an influence evaluation system that sorts the nodes above the node to be processed according to predetermined conditions for each entity, performs calculations using influence information for each entity in the sorted order, identifies an entity whose calculated value satisfies a condition for a predetermined threshold, and identifies the label of the node of the identified entity as the update label of the node to be processed.

[0025] In the above-mentioned invention, the label update processing unit can be configured as an influence evaluation system in which, when there are multiple nodes of the identified entity, the nodes in the identified entity are sorted according to specified conditions, and in the sorted order, a calculation is performed using influence information for each node, to identify a node whose calculated value satisfies a condition for a specified threshold, and the identified node is identified as a node to be propagated to the node to be processed.

[0026] In the present invention, intermediate influence is calculated by identifying the propagation path. Therefore, when an entity exists across multiple nodes, the entity must be considered as a whole when calculating the influence between the entities, but when identifying the propagation path, it is necessary to identify each node. Therefore, by configuring as in these inventions, it can be realized.

[0027] In the above-mentioned invention, the intermediate influence calculation processing unit can be configured as an influence evaluation system in which the intermediate influence at the node is not stored as history, or is excluded from the intermediate influence used for processing in the index calculation processing unit, until the calculation process of the intermediate influence at the node has been executed a predetermined number of times.

[0028] When the number of executions is small, the influence of the entity does not extend from the top layer to the bottom layer of the network. If the evaluation index is calculated using the history of intermediate influence calculated using the update label during the period when the influence does not extend, the accuracy of the evaluation index may be affected. Therefore, it is preferable to configure the system so that intermediate influence up to a predetermined number of times is not used in calculating the evaluation index indicating the influence.

[0029] In the above-mentioned invention, the intermediate influence calculation processing unit can be configured as an influence evaluation system that calculates the intermediate influence in the node by using a damping coefficient when performing the calculation process of the intermediate influence in the node.

[0030] When network information is circulating, it becomes impossible to calculate the intermediate influence. Therefore, by using a damping coefficient as in the present invention, it is possible to calculate the intermediate influence.

[0031] In the above-mentioned invention, the influence evaluation system has a network information input reception processing unit that receives input of the network information, and if the influence information is information indicating stock holdings, the network information input reception processing unit can be configured as an influence evaluation system in which, when the entity holds its own company stock, correction processing is performed on the network information whose input has been received using the company's own stock.

[0032] When influence information is stock ownership, the entity may also own its own stock. If the amount of the company's own stock is not taken into account, the accuracy of the processing may be affected. Therefore, it is preferable to carry out correction processing that takes the company's own stock into account.

[0033] In the above-mentioned invention, the network information input reception processing unit can be configured like an influence evaluation system in which, when the entity holds its own stock, a new node and edge are created for the network information for which the input has been received, and a correction process is performed in which a label indicating the entity is added to the new edge, and an edge from the new node to the new node is added to the new edge.

[0034] When the company's own stock influences the company's decision-making, the above-mentioned correction process can be performed by creating a new node and edge for the company's own label (node) for the company's own stock.

[0035] In the above-mentioned invention, the network information input receiving processing unit can be configured as an influence evaluation system that, when the entity holds its own company stock, performs a correction process in the network information that received the input to calculate influence information of the labels of nodes that have edges to the entity, excluding the entity's own company stock.

[0036] As for the above-mentioned correction process, if the company's own stock does not influence the company's decision-making and the initial influence information includes the company's own stock, it is advisable to correct the influence information by deducting the company's own stock.

[0037] In the above-mentioned invention, the index calculation processing unit can be configured as an influence evaluation system that performs a calculation to weight the scale of the entity when calculating the index indicating the influence.

[0038] The size of an entity ranges from extremely large to extremely small. Therefore, if an index is calculated simply based on the influence it has on other entities, the index will be higher the more relationships it has with other entities, regardless of its size. Therefore, by weighting the entity in consideration of its size, it is possible to evaluate its influence on larger entities more highly. This makes it possible to calculate an index of influence that is closer to the real world.

[0039] In the above-mentioned invention, the influence evaluation system can be configured as an influence evaluation system that calculates the difference between an index indicating the influence before a change in an entity in the network and an index indicating the influence after a change in an entity in the network.

[0040] In this way, by calculating the difference in the index that indicates the influence before and after the change of entity in the network, it is possible to evaluate the action that changes the entity, such as the decision of M&A. It is also possible to evaluate the butterfly effect on third parties.

[0041] The first invention can be realized by loading the program of the present invention into a computer and executing it. That is, it is an influence evaluation program that causes a computer to function as a label update processing unit that identifies a higher-level node of a node to be processed using the edges and identifies an updated label that updates the label of the node to be processed from the label of the higher-level node using the influence information, in network information in which a label indicating an entity is represented as a node, a direction of influence of the entity is represented as an edge, and information indicating the influence that the entity has on other entities is represented as influence information, an intermediate influence calculation processing unit that identifies a propagation path using the identified updated label and calculates an intermediate influence at the node, and an index calculation processing unit that calculates an evaluation index indicating the influence of the entity represented by the label at the node using the intermediate influence at the node. Effect of the Invention

[0042] By using the influence evaluation system of the present invention, it is possible to index the influence that an entity has over other entities, particularly the influence of intermediate entities, even when the relationships between entities are complex, for example when there are multiple hierarchical levels. [Brief description of the drawings]

[0043] [Figure 1] 1 is a block diagram showing an example of a system configuration of an influence evaluation system according to the present invention. [Diagram 2] FIG. 2 is a diagram illustrating an example of a hardware configuration of a computer that realizes the influence evaluation system of the present invention. [Diagram 3] 4 is a flowchart showing an example of the entire processing process of the influence evaluation system of the present invention. [Figure 4] 13 is a flowchart showing an example of a process for specifying an updated label in a node of the influence evaluation system of the present invention. [Diagram 5] FIG. 10 is a diagram illustrating an example of network information expressing relationships of influence between entities; [Figure 6]13 is a diagram illustrating an example of a state in which a history of intermediate influences is stored for each node in a network information storage unit; FIG. [Figure 7] FIG. 10 is a diagram illustrating an example of network information in which random numbers are assigned to the labels of each node. [Figure 8] FIG. 10 is a diagram illustrating an example of network information in a state in which a label to be updated is specified. [Figure 9] FIG. 10 is a diagram illustrating an example of network information in a state in which a propagation route is specified. [Figure 10] FIG. 10 is a diagram illustrating an example of network information in a state in which the intermediate influence of each node is specified. [Figure 11] FIG. 10 is a diagram illustrating an example of network information in a state in which the labels of nodes have been updated. [Figure 12] FIG. 10 is a diagram illustrating an example of network information in a state where a random number is assigned to the label of each node in the first embodiment. [Figure 13] FIG. 2 is a diagram illustrating an example of network information in a state in which a propagation route is specified in the embodiment. [Figure 14] FIG. 10 is a diagram illustrating an example of network information in a state in which the intermediate influence of each node is specified in the first embodiment. [Figure 15] FIG. 10 is a diagram illustrating an example of network information in a state in which a label of a node has been updated in the first embodiment. [Figure 16] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in a third iterative process, a label to be updated, and an intermediate influence of each node are specified in the first embodiment. [Figure 17] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in a fourth iterative process, a label to be updated, and an intermediate influence of each node are specified in the first embodiment. [Figure 18] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in a fifth iterative process, a label to be updated, and an intermediate influence of each node are specified in the first embodiment. [Figure 19] 1 is a diagram showing a schematic diagram of the evaluation indexes of each entity in Example 1. FIG. [Figure 20] FIG. 13 is a diagram illustrating an example of network information in a circulating state. [Figure 21] 10 is a diagram showing an example of network information in which a propagation route, a label to be updated, and an intermediate influence of each node are specified when a damping coefficient is used; FIG. [Figure 22] FIG. 11 is a diagram illustrating an example of a process for correcting network information by adding a company's stock E as a new entity in the second embodiment. [Diagram 23] FIG. 11 is a diagram illustrating an example of a process for correcting network information by deducting company stock E from network information in Example 2. [Figure 24] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in a second iterative process, a label to be updated, and an intermediate influence of each node are specified in the fourth embodiment. [Diagram 25] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in the third iterative process, a label to be updated, and an intermediate influence of each node are specified in the fourth embodiment. [Figure 26] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in a fourth iterative process, a label to be updated, and an intermediate influence of each node are identified in the fourth embodiment. [Figure 27] FIG. 13 is a diagram illustrating an example of network information in a state in which a route to be propagated in the fifth iterative process, a label to be updated, and an intermediate influence of each node are identified in the fourth embodiment. [Figure 28] FIG. 13 is a diagram showing a schematic diagram of the evaluation indexes of each entity in Example 4. [Figure 29] FIG. 13 is a diagram illustrating an example of network information in an initial state (before buying and selling) in the fifth embodiment. [Diagram 30] FIG. 13 is a diagram illustrating an example of network information after a sale and purchase in Example 5. [Diagram 31]This is a diagram showing the calculation of index values ​​based on the "Shapley-Shubik power index," a conventional method for evaluating influence. [Diagram 32] This is a diagram showing a schematic example of a structure that influences decision-making in an actual model. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0044] In the influence evaluation system 1 of the present invention, the relationships between entities are networked, and the label propagation method is used to perform the processing in order to perform numerical calculations on the network.

[0045] In the following explanation of this specification, we will explain the case where the relationships between entities are networked as stock ownership relationships, and the influence that an entity has on other entities in that network is evaluated. Representative examples of entities in this case are companies and / or natural persons, but the entities are not limited to these, and can be any entity that can be involved in the decision-making of other entities, such as natural persons, companies, groups or organizations that are collections of people, funds, associations, etc.

[0046] The computer has a calculation device 70 such as a CPU that executes the calculation processing of a program, a storage device 71 such as a RAM or a hard disk that stores information, a display device 72 such as a display, an input device 73 for inputting information, and a communication device 74 for communicating various information such as the processing results of the calculation device 70 and information stored in the storage device 71. If the computer has a touch panel display, the display device 72 and the input device 73 may be integrally configured. Touch panel displays are often used in portable communication terminals such as mobile phones, smartphones, and tablet computers, but are not limited to these.

[0047] The touch panel display is a device that integrates the functions of the display device 72 and the input device 73 in that input can be made directly on the display using a specified input device (such as a pen for the touch panel) or a finger.

[0048] The influence assessment system 1 may be realized by one computer, but its functions may be realized by a plurality of computers. In this case, the computer may be, for example, a cloud server.

[0049] Furthermore, the processing units in the influence assessment system 1 of the present invention may only have logically distinct functions and may physically or practically form the same area.

[0050] The influence evaluation system 1 has a network information input reception processing unit 10, a network information storage unit 11, a label update processing unit 12, an intermediate influence calculation processing unit 13, and an index calculation processing unit .

[0051] The network information input reception processing unit 10 receives input of information (network information) on a network structure expressing the relationship of influence between entities whose influence is to be evaluated using the present invention. The relationship of influence between entities is expressed as the entities as nodes and the relationships as edges, as shown in FIG. 5(a). The edges are expressed as a directed graph in the direction of influence. When expressing influence due to stock ownership, a company or a natural person becomes a node, and a directed graph is attached as an edge from a shareholder node to a node of the issuing company of the stock held by the shareholder. In addition, the shareholding ratio is associated with an edge as information showing influence on other entities (influence information). The example in FIG. 5(a) shows the share ownership relationship in a network structure, with Company B and Company C each holding 30% of the shares of Company A, Company E holding 40%, and Company B and Company D holding 50% of the shares of Company C. There are various types of influence on other entities, including influence due to risk.

[0052] FIG. 5(b) shows an example of the network structure of FIG. 5(a) embodied as network information in order to realize processing on a computer by the influence evaluation system 1 of the present invention. In FIG. 5(b), node identification information (node ​​identification information) such as N1 and N2 is attached to each node, and the identification information (label) of an entity such as a company or a natural person at the node is substituted as the value of the node identification information (including alphanumeric characters, symbols, etc. in addition to numerical values). As the label, it is preferable that the identification information of the entity associated with the node in the initial state of the network information (initial label) and the identification information of the current entity at the node (current label) are substituted as the value of the node identification information. Note that only the current label is substituted as a value for the label associated with each node, and the initial label for each node may be stored separately. In the description in this specification, for ease of understanding, the case where both the initial label and the current label are substituted as the label of each node will be described.

[0053] Each edge is expressed as E(X,Y), where X is the starting node and Y is the end node. Influence information, for example, stock ownership ratio, is substituted as the value of these edges. Note that the method of expressing the network structure as network information is not limited to this, and other methods may be used.

[0054] In the case of FIG. 5(b), N1=Aa, N2=Ee, N3=Bb, N4=Cc, and N5=Dd are assigned to each node. Also, E(N2,N1)=40%, E(N3,N1)=30%, E(N4,N1)=30%, E(N3,N4)=50%, and E(N5,N4)=50% are assigned to each edge. In FIG. 5(b), the initial label is expressed in uppercase letters and the current label in lowercase letters, but this is not limited to this. In this specification, for ease of understanding, the initial label and the current label are assigned as values ​​for each node, and the initial label is expressed in uppercase letters and the current label in lowercase letters, but the same characters in uppercase and lowercase letters indicate the same entity. For example, the initial state of node N1 is N1=Aa, but if node N2 is propagated to node N1, the entity of node N2 is Ee, so N1=Ae. That is, A is assigned as the initial label and e as the current label. This means that the initial state entity of node N1 is A, and the propagated entity (the entity of the current node N1) is E. As mentioned above, it is also possible to assign the value of the current label to each node rather than assigning the value of the initial label. In this case, N1=a, which means that the entity of node N1 is A. At this time, the initial label values ​​of each node can be stored as initial values ​​as appropriate.

[0055] The network information input reception processing unit 10 receives input of network information in an initial state, and stores the initial state network information in the network information storage unit 11 described later. Note that the network information may be in a state in which the network structure shown in Fig. 5 etc. is appropriately embodied in computer processing.

[0056] The network information storage unit 11 stores the network information that is received as input by the network information input reception processing unit 10. As the network information, nodes and their values ​​(labels (initial label, current label)), edges and influence information between the nodes, such as stock ownership ratios, are stored in association with each other as network information in the initial state. In addition, values ​​based on the propagation path for each node (values ​​of intermediate influence) are stored in association with each other as history. In addition to the values ​​of intermediate influence for each node, the value of the current label may be stored as history. The above-mentioned nodes and their values ​​may be used as the value of the current label to be stored as history. FIG. 6 shows a schematic example of the history of the values ​​of intermediate influence for each node and the value of the current label in the network information storage unit 11.

[0057] The label update processing unit 12 identifies the label of the node to be propagated by the label propagation method at each node. This is to identify which entity in the node's direct upper node has influenced the entity in the node. In practice, this is performed by repeatedly determining which entity has influenced the entity, for example, by random sampling.

[0058] The label update processing unit 12 executes the following process as an example, but is not limited to the process as long as the label to be updated (propagated) (update label) can be specified.

[0059] A random number is assigned to a higher-level node that is directly connected to a certain node. Then, for the higher-level nodes, the nodes are sorted in ascending order for each entity based on the current label information based on the assigned random number. Then, in the order of the sorted nodes, the influence information between the nodes, for example, the shareholding ratio, is calculated (for example, added). If this calculated (for example, added) value exceeds a predetermined threshold, for example, 50%, the label in the node of the entity that exceeds the threshold is identified as the updated label. Note that the random numbers do not have to be sorted in ascending order, but may be sorted in descending order, or may be in another method, and any method may be used as long as it can randomly identify the order in which the influence information of the labels of each node, which will be described later, is calculated. In addition, the influence information between nodes may be calculated in the sorted order, or the labels that exceed the threshold may be identified by calculating the influence information by any calculation method in an order that satisfies a predetermined condition.

[0060] In addition, when the nodes are sorted by entity, if the entity whose influence information exceeds the threshold has multiple nodes, the nodes are further sorted based on the random number assigned to each node, and the influence information between the nodes is added in the order of the sorted nodes until it exceeds a predetermined threshold. The process of identifying the updated label in the label update processing unit 12 will be described in detail later.

[0061] The label update processing unit 12 executes the process of identifying updated labels for all nodes that have higher-level nodes than itself.

[0062] For example, suppose the network information is as shown in Figure 5(b). At this time, nodes N2, N3, and N4 are above node N1, and the random numbers "0.52", "0.63", and "0.17" are assigned to each of them. Furthermore, node N4 is the only node that has a node above it. Therefore, the random numbers "0.58" and "0.21" are assigned to nodes N3 and N5 above node N4, respectively. Figure 7 shows a schematic diagram of this state.

[0063] Then, for nodes N2, N3, and N4, which are upper nodes of node N1, sort the nodes in ascending order for each entity based on the current label information based on the random numbers assigned to each, and arrange them in the order of node N4, node N2, and node N3. In addition, for nodes N3 and N5, which are upper nodes of node N4, sort them in ascending order in the order of node N5, and arrange them in the order of node N3.

[0064] Adding influence information in order for nodes N2, N3, and N4, which are upper nodes of node N1, the influence information of edge E (N3, N1) from node N3 to node N1 is 30%, and the influence information of edge E (N2, N1) from node N2 to node N1 is 40%, so at this point it exceeds the threshold of 50%, so the current label e of entity E of node N2 is identified as the label (update label) that updates node N1. Adding influence information in order for nodes N3 and N5, which are upper nodes of node N4, the influence information of edge (N5, N4) from node N5 to node N4 is 50%, and the influence information of edge (N3, N4) from node N3 to node N4 is 50%, so at this point it exceeds the threshold of 50%, so the current label b of entity B is identified as the label (update label) that updates node N4. This is shown diagrammatically in FIG. 8.

[0065] The label update processing unit 12 updates the label (current label) of each node with the update label identified by the label update processing unit 12.

[0066] If the update label identified by the label update processing unit 12 is that shown in Fig. 8, the label (current label) of the node is updated as shown in Fig. 11. Then, the label update processing unit 12 stores the updated label (current label) as history in the network information storage unit 11. That is, the network information storage unit 11 stores the history of updating the current label for node N1 to e and the current label for node N4 to b.

[0067] The intermediate influence calculation processing unit 13 specifies the route of the label to be propagated as the propagation route using the updated label specified by the label update processing unit 12, and calculates the intermediate influence at each node (FIG. 9). In Fig. 9, the propagation route is indicated by a solid line, and the route that was not propagated is indicated by a dashed line. The intermediate influence is a calculated value calculated using the number of nodes at the lower level, including itself, in the identified propagation route. For example, in Fig. 9, the label propagates from node N2 to node N1, and from node N3 to node N4, so the intermediate influence calculation processing unit 13 specifies E(N2,N1) and E(N3,N4) as the propagation routes.

[0068] Then, the intermediate influence calculation processing unit 13 specifies the intermediate influences of the respective nodes as follows: node N1=1, node N2=2, node N3=2, node N4=1, and node N5=1. That is, since node N1 has no lower nodes below itself, its intermediate influence is 1, since node N2 has node N1 as a lower node specified as a propagation route, its intermediate influence is 2, since node N3 has node N4 as a lower node specified as a propagation route, its intermediate influence is 2, since node N4 has no lower nodes (E(N4, N1) is not specified as a propagation route), its intermediate influence is 1, and since node N5 has no lower nodes (E(N5, N4) is not specified as a propagation route), its intermediate influence is 1 (FIG. 10).

[0069] The label update processing unit 12 and the intermediate influence calculation processing unit 13 may execute processing in parallel, or the label update processing may be performed first, and then the processing of the intermediate influence calculation processing unit 13 may be executed.

[0070] Then, in this network state (the state in FIG. 9), the processing is again executed in the label update processing unit 12 and the intermediate influence calculation processing unit 13. In this manner, the processing of S110 to S150 in the flowchart in FIG. 3, which will be described later, is executed a predetermined number of times, for example 10,000 times.

[0071] The index calculation processing unit 14 calculates a performance index (NPF) as an index according to the present invention based on the history of the intermediate influence of each node stored in the network information storage unit 11. For example, the performance index (NPF) is calculated by calculating the average value of the intermediate influence of each node.

[0072] For example, after 10,000 iterative processes, the average value of the intermediate influence for each node can be calculated using the history of the intermediate influence for each node in the network information storage unit 11 shown in Figure 6, and the following evaluation indexes can be calculated for each node: NPF(N1) = 1 for node N1, NPF(N2) = 1.16 for node N2, NPF(N3) = 2.12 for node N3, NPF(N4) = 1.42 for node N4, and NPF(N5) = 1.69 for node N5.

[0073] Various calculation methods can be used to calculate the evaluation index. For example, when using the average value of the intermediate influence of each node, NPF of each node = sum of intermediate influences per node / number of iterations The calculation method is not limited to this as long as it is calculated using the intermediate influence of each node.

[0074] Then, the index calculation processing unit 14 uses the evaluation index (NPF) for each node to calculate the evaluation index (NPF) of the entity of the initial label associated with the node. For example, assuming that the evaluation index for each node is as described above, the initial label of node N1 is A, so the evaluation index NPF(A) of entity A is 1, the initial label of node N2 is E, so the evaluation index NPF(E) of entity E is 1.16, the initial label of node N3 is B, so the evaluation index NPF(B) of entity B is 2.12, the initial label of node N4 is C, so the evaluation index NPF(C) of entity C is 1.42, and the initial label of node N5 is D, so the evaluation index NPF(D) of entity D is 1.69. Note that when calculating the evaluation index from the node to the entity, any calculation may be performed instead of using the value as it is. EXAMPLES

[0075] Next, an example of the processing process of the influence evaluation system 1 of the present invention will be described with reference to the flowcharts of Figures 3 and 4. In this embodiment, the influence on decision-making due to stock ownership relationships is evaluated, and the initial network information is as shown in Figure 5. The threshold value when the influence information is added is 50%.

[0076] The operator inputs the initial state network information (FIG. 5(b)) shown in FIG. 5 to be processed, and this is accepted by the network information input acceptance processing unit 10 (S100). The network information input acceptance processing unit 10 stores the accepted initial state network information in the network information storage unit 11.

[0077] Then, the label update processing unit 12 executes the process of identifying the updated labels for all nodes that have higher-level nodes than itself (S110).

[0078] First, the label update processing unit 12 generates random numbers for the nodes N2, N3, and N4 above the node N1, respectively, and assigns, for example, the random numbers "0.52", "0.63", and "0.17", and generates random numbers for the nodes N3 and N5 above the node N4, respectively, and assigns, for example, the random numbers "0.58" and "0.21" (S200) (Figure 7).

[0079] Next, the label update processing unit 12 sorts the nodes N2 to N4 above node N1 in ascending order for each entity based on the current label information based on the random numbers assigned to each (S210), leaving the order as node N4 (entity C), node N2 (entity E), node N3 (entity B). Similarly, the label update processing unit 12 sorts the nodes N3 (entity B) and N5 (entity D), above node N4, in ascending order (S210), leaving the order as node N5 (entity D), node N3 (entity B).

[0080] 8, for the upper nodes N2 to N4 of node N1, the influence information is added in the sorted order (S220), and since the threshold value of 50% is exceeded when the influence information of edge E (N2, N1) from node N2 to node N1 is added, the current label e of entity E is identified as the update label of node N1 (S230). Also, for the upper nodes N3 and N5 of node N4, the influence information is added in the sorted order based on the influence information (S220), and since the threshold value of 50% is exceeded when the influence information of edge E (N3, N4) from node N3 to node N4 is added, the current label b of entity B is identified as the update label of node N4 (S230).

[0081] As described above, when the label update processing unit 12 identifies the update labels for all nodes that have higher-level nodes, it then judges whether the identified entity (current label) has multiple nodes (S240). Here, since the entity E identified in the processing of the higher-level node of node N1 is only one node (node ​​N2), it is judged that there are no multiple nodes (S240). Also, since the entity B identified in the processing of the higher-level node of node N4 is only one node (node ​​N3), it is judged that there are no multiple nodes (S240).

[0082] When the label update processing unit 12 specifies the updated label, the intermediate influence calculation processing unit 13 specifies the propagation path (S120). That is, in the case of Fig. 8, the edge E(N2, N1) from node N2 to node N1 and the edge E(N3, N4) from node N3 to node N4 are specified as the propagation path (Fig. 9).

[0083] Then, the intermediate influence calculation processing unit 13 calculates the intermediate influence for each node, and stores the intermediate influence for each node as history in the network information storage unit 11 (S130). In the network information of Fig. 9, the intermediate influence is specified as 1 for node N1, 2 for node N2 (node ​​N2 has propagated to node N1, and the number of lower nodes including node N2 is 2 for nodes N2 and N1), 2 for node N3 (node ​​N3 has propagated to node N4, and the number of lower nodes including node N3 is 2 for nodes N3 and N4), 1 for node N4, and 1 for node N5 (Fig. 10).

[0084] Furthermore, the label update processing unit 12 updates the label (current label) of the node with the updated label (FIG. 11), and stores the label (current label) of each node as history in the network information storage unit 11 (S140). That is, the network information storage unit 11 stores the current label e for node N1, the current label e for node N2, the current label b for node N3, the current label b for node N4, and the current label d for node N5 as history.

[0085] If the predetermined number of times is 10,000 repetitions, then since this is the first time the process has been executed, the process from S110 onwards is repeated again (S150). That is, in the state of the network information in FIG. 11, the process from S110 onwards is executed again.

[0086] The label update processing unit 12 executes the process of identifying updated labels for all nodes that have higher-level nodes than itself, in the same manner as described above (S110).

[0087] The label update processing unit 12 generates random numbers for the nodes N2, N3, and N4 above the node N1, respectively, and assigns them the random numbers "0.15", "0.59", and "0.38", and generates random numbers for the nodes N3 and N5 above the node N4, respectively, and assigns them the random numbers "0.98" and "0.08" (S200). The network information in this state is shown in Fig. 12(a).

[0088] Next, the label update processing unit 12 sorts the nodes in ascending order for each entity based on the current label information for the higher nodes N2 to N4 of node N1 based on the random numbers assigned to each of them (S210). Here, the current labels of nodes N3 and N4 are both b, which are the same entity B. Therefore, nodes N3 and N4 are treated as the same entity B when sorting the nodes (when arranging the nodes in order). At this time, the random number of either node N3 or node N4 may be adopted, or the average value may be adopted, or any other method may be used to determine the random number. For example, here, the smaller random number is adopted (FIG. 12(b)). Then, for nodes N3 and N4, the same random number "0.38" is assigned when they are in ascending order, and they can be processed as one entity when they are in ascending order below.

[0089] That is, the label update processing unit 12 first sorts the upper nodes N2 to N4 of the node N1 in ascending order based on the assigned random number for each entity based on the information of the current label (S210). Here, the random number of the node N2 is "0.15" and the random number of the combination of the node N3 and the node N4 is "0.38", so the nodes can be sorted in the order of node N2 (entity E) and the combination of the node N3 and the node N4 (entity B). The influence information of the edge E (N2, N1) from the node N2 to the node N1 is 40%, which does not exceed the threshold. The influence information of the next combination of the node N3 and the node N4 is 60% of the sum of the influence information of the edge E (N3, N1) from the node N3 to the node N1 and the influence information of the edge (N4, N1) from the node N4 to the node N1. Therefore, since the threshold is exceeded at this point, the current label b of the combination of the node N3 and the node N4 (entity B) can be identified as the update label of the node N1 (FIG. 12(b)).

[0090] As described above, when the label update processing unit 12 identifies the update labels for all nodes that have higher-level nodes, it then judges whether the identified entity (current label) has multiple nodes (S240). Here, since the entity B identified in the processing of the higher-level node of node N1 has two nodes (node ​​N3, node N4), it is judged that there are multiple nodes (S240).

[0091] In this case, since there are multiple nodes (combinations are used as the same entity), the propagation path cannot be specified as it is, so the label update processing unit 12 further sorts each node in the same entity, that is, node N3 and node N4 (S250). That is, in the combination of node N3 and node N4, sorting is performed based on the random numbers for each node originally assigned to node N3 and node N4. Then, since the combination of node N3 and node N4 is sorted in ascending order with "0.59" of node N3 and "0.38" of node N4, the order is node N4, node N3. Therefore, if the nodes N2 to N4 above node N1 are sorted in ascending order, the order is node N2, node N4, node N3, and if the influence information is added in order until it exceeds the threshold (S260), the influence information of the edge (E(N4, N1)) from node N4 to node N1 exceeds the threshold, so node N4 is specified as the node of the updated label to be propagated to node N1 (FIG. 12(c)) (S270).

[0092] Similarly, the label update processing unit 12 sorts the upper nodes of node N4, nodes N3 and N5, in ascending order (S210), so that the order is node N5, node N3. Then, for the upper nodes N3 and N5 of node N4, addition is performed based on the influence information in the sorted order (S220), and since the threshold value of 50% is exceeded when edge E(N3,N4) from node N3 to node N4 is added, the current label b is identified as the update label for node N4 (S230). Then, since the entity B identified in the processing of the upper nodes of node N4 is only one node (node ​​N3), it is determined that there are no multiple nodes (S240).

[0093] As described above, when the label update processing unit 12 executes the process of identifying updated labels for all nodes that have higher-level nodes, the intermediate influence calculation processing unit 13 identifies the propagation path (S120). That is, in the case of Fig. 12, the propagation path to node N1 is identified as the path that propagates edge E(N4,N1) from node N4, which is the path from the node identified in S270, to node N1, and the propagation path to node N4 is identified as edge E(N3,N4) from node N3 to node N4 (Fig. 13).

[0094] Then, the intermediate influence calculation processing unit 13 calculates the intermediate influence for each node, and stores the intermediate influence for each node as history in the network information storage unit 11 (S130). In Fig. 13, the calculation is 1 for node N1, 1 for node N2, 3 for node N3 (node ​​N3 has propagated to node N4, and node N4 has further propagated to node N1, so the number of lower nodes including node N3 is 3 for nodes N3, N4, and N1), 2 for node N4 (node ​​N4 has propagated to node N1, so the number of lower nodes including node N4 is 2 for nodes N4 and N1), and 1 for node N5 (Fig. 14).

[0095] Furthermore, the label update processing unit 12 updates the label (current label) of the node with the updated label (S140), and stores the label (current label) of each node as history in the network information storage unit 11 (S140). That is, the network information storage unit 11 stores the current label b for node N1, the current label e for node N2, the current label b for node N3, the current label b for node N4, and the current label d for node N5 as history. The network information in this state is shown in FIG.

[0096] If the predetermined number of times is 10,000 repetitions, then since this is the second time the process has been executed, the process from S110 onwards is repeated again (S140). That is, in the state of FIG. 15, the process from S110 to S130 is repeated.

[0097] For example, assume that the network information after three to five iterations of the iterative process, and the current label and intermediate influence of each node are as shown in Figures 16 to 18. Figure 16 shows the third iterative process, in which the current label b of node N3 is propagated to node N1 (propagation route is E(N3,N1)), and the current label d of node N5 is propagated to node N4 (propagation route is E(N5,N4)). The current label and intermediate influence of each node at this time are as follows: node N1's current label is b and its intermediate influence is 1, node N2's current label is e and its intermediate influence is 1, node N3's current label is b and its intermediate influence is 2, node N4's current label is d and its intermediate influence is 1, and node N5's current label is d and its intermediate influence is 2.

[0098] Figure 17 shows the fourth iteration, where the current label e of node N2 is propagated to node N1 (the propagation route is E(N2,N1)), and the current label d of node N5 is propagated to node N4 (the propagation route is E(N5,N4)). The current label and intermediate influence of each node at this time are as follows: node N1's current label is e and its intermediate influence is 1, node N2's current label is e and its intermediate influence is 2, node N3's current label is b and its intermediate influence is 1, node N4's current label is d and its intermediate influence is 1, and node N5's current label is d and its intermediate influence is 2.

[0099] Figure 18 shows the fifth iteration, where the current label d of node N4 is propagated to node N1 (the propagation route is E(N4,N1)), and the current label d of node N5 is propagated to node N4 (the propagation route is E(N5,N4)). The current label and intermediate influence of each node at this time are as follows: node N1's current label is d and its intermediate influence is 1; node N2's current label is e and its intermediate influence is 1; node N3's current label is b and its intermediate influence is 1; node N4's current label is d and its intermediate influence is 2; and node N5's current label is d and its intermediate influence is 3.

[0100] In this way, after repeating the processes from S110 to S140 a predetermined number of times (S150), the index calculation processing unit 14 calculates the evaluation index (NPF) for each node by a predetermined calculation method using the intermediate influence, such as calculating an average value of the intermediate influence for each node, based on the history of the intermediate influence of each node stored in the network information storage unit 11 (S160). In this way, the evaluation index (NPF) of each node is calculated as the evaluation index of the entity of the initial label associated with the node.

[0101] For example, after 10,000 processings, the NPF for node N1 is 1, the NPF for node N2 is 1.16, the NPF for node N3 is 2.12, the NPF for node N4 is 1.42, and the NPF for node N5 is 1.69. The NPFs for the initial label entities corresponding to each node can be calculated as follows: NPF(A)=1 for entity A (node ​​N1), NPF(B)=2.12 for entity B (node ​​N3), NPF(C)=1.42 for entity C (node ​​N4), NPF(D)=1.69 for entity D (node ​​N5), and NPF(E)=1.16 for entity E (node ​​N2). This is shown diagrammatically in Figure 19.

[0102] In the above-mentioned processes from S110 to S150, the current label in each node may be returned to the initial state based on the initial label at any timing. This makes it possible to avoid the occurrence of local minimum problems due to initial value dependency.

[0103] To return the current label of each node to its initial state, for example, a random number p may be generated before the process of S110, and if the random number p satisfies the condition related to the initialization reference value, the current label of each node may be returned to its initial label. For example, when the random number p is 0≦p≦1, if the condition for the initialization reference value for returning to the initial state is 0≦p≦0.005, when the random number p is 0≦p≦0.005, the current label of each node is returned to the initial label, and this is identified as the updated label. On the other hand, when the random number p is >0.005, the process from S200 onwards may be executed to identify the updated label.

[0104] The initialization reference value can be set arbitrarily.

[0105] In addition, in the network information, a predetermined number of trials are required for the influence of an entity to propagate from the top layer to the bottom layer of the network. Therefore, the processes of S110 to S150 are executed a predetermined number of times from the first time, for example up to about 19 times, but the calculated intermediate influence may not be stored as history. Or, even if it is stored, it may be excluded from the history of the intermediate influence used in calculating the evaluation index. The predetermined number of trials can be set sequentially based on the size of the network, such as the number of layers.

[0106] In addition, when the network information of each entity circulates as shown in FIG. 20, the intermediate influence at each node will circulate in the intermediate influence calculation processing unit 13. Therefore, when identifying the intermediate influence, an attenuation coefficient q may be used. That is, when identifying the intermediate influence, set itself to 1, its lower level to 1×q, and its lower level to 1×q×q, and sum the values multiplied by the attenuation coefficient. For example, taking a certain node of entity A in FIG. 20 as a reference and setting the attenuation coefficient q = 0.85 (rounding to the third decimal place), a certain node of entity A is 1, a certain node of the lower-level entity B is 0.85, a certain node of the lower-level entity C is 0.72, a certain node of the lower-level entity D is 0.61, and a certain node of the lower-level entity E is 0.52. Then, the intermediate influence of a certain node of entity A becomes 3.7 when the exponential calculation processing unit 14 adds these values, so the evaluation index (NPF(A)) of entity A is 3.7. Note that the calculation is performed by rounding to the third decimal place.

[0107] Also, in the case of the network information in FIG. 18, as shown in FIG. 21, the intermediate influence of each node is such that the intermediate influence of node N1 is 1, the current label of node N2 is e, the intermediate influence is 1, the intermediate influence of node N3 is 1, the intermediate influence of node N4 is 1.85 (=1 + 1×0.85), and the intermediate influence of node N5 is 2.57 (=1 + 1×0.85 + 1×0.85×0.85). These are stored in the network information storage unit 11 as the history of the intermediate influence of each node.

[0108] Of course, even when the network information does not circulate, the attenuation coefficient q may be used. Also, although 0.85 is given as an example of the attenuation coefficient q, it is not limited thereto, and 0 < q < 1 is sufficient. For example, when the attenuation coefficient q is 0.5, it will look downstream up to about 3 levels, and when the attenuation coefficient q is 0.85, it will look downstream up to about 12 levels.

Example

[0109] In the first embodiment, the shareholding ratio is used as it is as the influence information, but the company's own stock may be taken into consideration. That is, the company that issues the stocks for which voting rights can be exercised may also hold them (so-called company stocks). In Japan, the company's own stocks cannot be used for voting rights, so they do not affect the decision-making, but if the company's own stocks are included in the shareholding ratio indicating the influence information, this may be taken into consideration. For example, in the network information of FIG. 5, for entity E, if entity F holds 50% of the issued shares for which voting rights can be exercised, and the remaining 50% are the company's own stocks, the network information may be corrected so that the company's own stocks E are added as a new entity. This is shown diagrammatically in FIG. 22. That is, while the network information is originally provided with nodes, edges, and influence information as shown in FIG. 22(a), the upper node of entity E is only entity F, and the total of the influence information is not 100%. Therefore, as shown in Figure 22 (b), a node N7 indicating the company's own stock is added, and an edge E (N7, N2) is further added from node N7 to node N2 (the company's own node).The network information input reception processing unit 10 can then receive the input of the network information and execute the processing of Example 1.

[0110] In addition to the above, the company's own shares may be processed as follows: That is, the influence information may be calculated based on the ownership ratio of shares for which voting rights can be exercised among all issued shares excluding the company's own shares, with respect to the ownership ratio of shares indicating the influence information.

[0111] That is, the shareholding ratio of a certain entity may be calculated by dividing the number of shares held by the entity by the number of issued shares on which voting rights can be exercised minus the number of the entity's own shares. For example, in the case of FIG. 22(a) above, if entity F holds 50% of the issued shares on which entity E can exercise voting rights, and the remaining 50% are the entity's own shares, then entity F's shareholding ratio may be calculated as 100% (= number of shares of E held by F / (total number of issued shares - number of the entity's own shares) x 100). An example of network information processed in this way is shown in FIG. 23. Note that in FIG. 23, the correction of the shareholding ratio is calculated using the number of shares, but it may also be calculated using the shareholding ratio.

[0112] Furthermore, in the case of a company that issues classes of stock that influence the company's decision-making, such as so-called golden shares (stocks that can exercise veto rights over certain resolutions at the general shareholders' meeting), stocks with multiple voting rights, and stocks with no voting rights, the influence information is not determined solely based on the shareholding ratio. Therefore, the influence information may be corrected taking into account the classes of stock, and then the information may be input to the network information input reception processing unit 10 as the initial state of the network information. EXAMPLES

[0113] In the first and second embodiments, when identifying the label to be updated in the label update processing unit 12, the random numbers attached to the labels of each node are sorted, the influence information of each label is added in that order, and the label that exceeds a predetermined threshold is identified as the label to be updated, but this threshold may be another threshold other than 50%. For example, it may be set to 2 / 3 (67%, 66.7%, etc.) or 3 / 4 (75%).

[0114] In Japan, important company matters such as the transfer of a significant part of a company's business, changes to the articles of incorporation, and reductions in the amount of capital are decided by special resolutions or specialized resolutions rather than ordinary resolutions, so the above threshold can be set based on this as a standard for the influence over a company. EXAMPLES

[0115] When using the influence evaluation system 1 of the present invention, the influence is evaluated regardless of the size of the entity. Therefore, an entity that holds a large number of shares in a very small company may end up with a higher evaluation index than an entity that holds only a small number of shares in a very large company.

[0116] Therefore, when the index calculation processing unit 14 calculates the evaluation index, the intermediate influence for each entity may be weighted with information indicating the size of the entity, such as net assets and market capitalization. For example, the network information input reception processing unit 10 receives input of the size (node ​​size (weighting coefficient)) for each entity, and the intermediate influence calculation processing unit 13 calculates the intermediate influence for each node by the total value of the size of the nodes below it, including itself, in the identified propagation route, taking into account the information indicating the size of the entity. At this time, companies with small entity sizes may not be weighted (weighting coefficient = 1), and companies with large entity sizes may be multiplied by a predetermined coefficient as a weighting (for example, weighting coefficient = 2), and the weighted intermediate influence may be used.

[0117] For example, if entity C is larger than the other entities and therefore its influence on entity C is weighted at double, and label propagation is performed as in Figure 14, the intermediate influence of node N1 is calculated as 1, that of node N2 as 1, that of node N3 as 4 (node ​​N3 has propagated to node N4, and node N4 has further propagated to node N1, so the number of lower nodes including node N3 is node N3, node N4 (intermediate influence weighted at double), and node N1, so 1+2+1=4), that of node N4 as 3 (node ​​N4 has propagated to node N1, so the number of lower nodes including node N4 is node N4 (intermediate influence weighted at double), and node N1, so 1+2=3), and that of node N5 as 1 (Figure 24).

[0118] Also, assume that the network information after the iterative process is executed, and the current label and intermediate influence of each node are as shown in Figures 25 to 27. Figure 25 shows the third iterative process, in which the label Bb of node N3 is propagated to node N1 (propagation route is E(N3,N1)) and the label Dd of node N5 is propagated to node N4 (propagation route is E(N5,N4)). At this time, the current label and intermediate influence of each node are as follows: node N1's current label is b, its intermediate influence is 1, node N2's current label is e, its intermediate influence is 1, node N3's current label is b, its intermediate influence is 2, node N4's current label is d, its intermediate influence is 2, node N5's current label is d, its intermediate influence is 3 (the sum of the intermediate influence of node N4 of 2 and the intermediate influence of node N5 of 1 is 3).

[0119] Figure 26 shows the fourth iteration, where the label Ee of node N2 is propagated to node N1 (the propagation route is E(N2,N1)), and the label Dd of node N5 is propagated to node N4 (the propagation route is E(N5,N4)). The current label and intermediate influence of each node at this time are as follows: node N1's current label is e and its intermediate influence is 1, node N2's current label is e and its intermediate influence is 2, node N3's current label is b and its intermediate influence is 1, node N4's current label is d and its intermediate influence is 2, node N5's current label is d and its intermediate influence is 3 (the sum of the intermediate influence of node N4, 2, and the intermediate influence of node N5, 1, is 3).

[0120] Figure 27 shows the fifth iteration, where the label Cd of node N4 is propagated to node N1 (the propagation route is E(N4,N1)) and the label Dd of node N5 is propagated to node N4 (the propagation route is E(N5,N4)). The current label and intermediate influence of each node at this time are as follows: node N1's current label is d, its intermediate influence is 1; node N2's current label is e, its intermediate influence is 1; node N3's current label is b, its intermediate influence is 1; node N4's current label is d, its intermediate influence is 3 (node ​​N4 has propagated to node N1, and the number of lower nodes including node N4 is node N4 (influence is weighted twice) and node N1, so 1+2=3); node N5's current label is d, and its intermediate influence is 4 (the sum of the intermediate influence of 1 of node N1, the intermediate influence of 3 of node N4, and the intermediate influence of 1 of node N5 is 4).

[0121] In this way, after repeating the processes from S110 to S140 in the above-mentioned first embodiment a predetermined number of times, the index calculation processing unit 14 calculates the evaluation index (NPF) for each node by a predetermined calculation method, such as calculating an average value of the intermediate influence for each node, based on the history of the intermediate influence of each node stored in the network information storage unit 11. Then, the index calculation processing unit 14 calculates the evaluation index (NPF) of each node as the evaluation index of the entity of the initial label associated with the node.

[0122] For example, when the result of 10,000 processing operations is the same as that shown in FIG. 19, in this embodiment, the intermediate influence of entity C is multiplied by twice the weighting coefficient, so that the index calculation processing unit 14 calculates an evaluation index (NPF) for node N1 of 1, an evaluation index (NPF) for node N2 of 1.16, an evaluation index (NPF) for node N3 of 2.62 (=2.12+0.5) (the intermediate influence of entity C (node ​​N4) is divided by 0.5 each between entity B (node ​​N3) and entity D (node ​​N5)), an evaluation index (NPF) for node N4 of 2.42 (=1.42+1) (the weighting of entity C is doubled, so 1 is added to the intermediate influence), and an evaluation index (NPF) for node N5 of 2.19 (=1.69+0.5) (the intermediate influence of entity C (node ​​N4) is divided by 0.5 each between entity B (node ​​N3) and entity D (node ​​N5)). The index calculation processing unit 14 can then calculate the NPFs of the entities of the initial labels associated with each node as follows: 1 for entity A (node ​​N1), 2.62 for entity B (node ​​N3), 2.42 (weighted) for entity C (node ​​N4), 2.19 for entity D (node ​​N5), and 1.16 for entity E (node ​​N2). This is shown diagrammatically in Figure 28.

[0123] By weighting the entities in this way according to their size, larger entities can be given a higher evaluation index, and the evaluation index can be calculated taking into account not only the number of entities affected but also their size, resulting in an index that is more in line with the actual influence.

[0124] This weighting can be set appropriately depending on the size of the entity itself and the size of the country to which it belongs, and the weighting can also be set in multiple stages depending on the size.

[0125] Furthermore, weighting can be applied according to the purpose of calculating the evaluation index. For example, when evaluating based on ESG (Environment, Social, Governance), each entity in each of the above-mentioned examples can be weighted based on a predetermined standard on whether it takes into consideration the environment, social responsibility, and corporate governance, and an evaluation index for entities (companies) with ESG risks can be calculated by weighting entities that do not take ESG into consideration and weighting entities that do take ESG into consideration.

[0126] This makes it possible to avoid unexpected liabilities and reputational risks, for example, in M&A or invested companies.

[0127] Of course, it is also possible to evaluate the proactiveness towards ESG by giving a higher weight to entities that take ESG into consideration and a lower weight to entities that do not. EXAMPLES

[0128] Furthermore, each of the above-mentioned embodiments can be used to evaluate corporate value in M&A and other cases.

[0129] For example, suppose that the initial network information is as shown in Fig. 29. Such network information is read into the influence evaluation system 1, and the evaluation index for each entity is calculated. As a result, suppose that the evaluation index for each entity is entity A = 1.68, entity B = 1.33, and entity C = 1.33.

[0130] In this case, when entity B purchases all of entity α's shares from entity A (i.e., B purchases company α from A), the present invention can be used to evaluate the corporate value of company α.

[0131] First, as shown in Figure 30, the network information after entity B purchases entity α is read and the evaluation index for each entity is calculated. As a result, the evaluation index for each entity is assumed to be entity A=1, entity B=2, and entity C=1.

[0132] Then, the value of the entity involved in the transaction can be calculated by the difference between the evaluation index before and after the transaction. In the case of Figures 29 and 30, the value of entity α involved in the transaction can be evaluated as 0.68 (=|1.68-1|) for entity A, 0.67 (=|1.33-2|) for entity B, and 0.33 (=|0-0.33|) for entity C.

[0133] Therefore, if Entity α is traded at a price that corresponds to an evaluation index between 0.67 and 0.68, it will be a fair price for Entity A and Entity B. Furthermore, it is also possible to calculate the impact (butterfly effect) on a third party (Entity C) before and after the trade.

[0134] That is, as in this embodiment, by using the influence evaluation system 1 of the present invention, it is possible to calculate the evaluation index before and after a change in an entity, and by comparing the difference, to evaluate not only the value of the entity resulting from the change in the entity, but also the impact on third parties (butterfly effect). EXAMPLES

[0135] The same as in each of the above-mentioned embodiments can also be applied to the calculation of stock ownership, i.e., influence on decision-making in various fields other than influence on company decision-making. For example, it can be applied to influence on national decision-making. In this case, when some legislation is enacted in the Diet, the number of seats held by each political party in the Diet indicates the influence information of the political party (entity) in the Diet. And, for example, the party's support group (entity) has influence on that political party, and the influence can be set as influence information by quantifying the amount of political donations, the number of members of the group, the number of Diet members from that group, etc. Furthermore, the influence on the support group can be set as influence information by evaluating the number of people constituting the faction (entity) within the support group, and the influence can be evaluated.

[0136] In this way, the connections between entities in society, such as people and companies, can be represented as a network structure, and the influence of each can be quantified using specified indicators and set as influence information, making it possible to evaluate the influence that an entity has on other entities. [Industrial Applicability]

[0137] By using the influence evaluation system 1 of the present invention, even when the relationships between entities are complex, for example when there are multiple hierarchical levels, it is possible to index the influence that an entity has over other entities, particularly the influence of intermediate entities. [Explanation of symbols]

[0138] 1: Influence rating system 10: Network information input reception processing unit 11: Network information storage unit 12: Label update processing unit 13: Intermediate influence calculation processing unit 14: Index calculation processing unit 70: Arithmetic device 71:Storage device 72:Display device 73: Input device 74:Communication equipment

Claims

1. a label update processing unit for identifying an upper node of a node to be processed using the edges, in network information in which a label indicating an entity is represented as a node, a direction of influence of the entity is represented as an edge, and information indicating influence that the entity has on other entities is represented as influence information, and for identifying an update label for updating the label of the node to be processed from the label of the upper node using the influence information; an intermediate influence calculation processing unit that calculates an intermediate influence at a node by specifying a propagation path using the specified update label; an index calculation processing unit that calculates an evaluation index indicating the influence of an entity indicated by a label at the node, using the intermediate influence at the node; An influence evaluation system comprising:

2. The label update processing unit includes: Identifying a label to be updated in a node having the higher-level node using a label propagation method; 2. The influence evaluation system according to claim 1 .

3. The label update processing unit: Identifying labels to be updated in the nodes having the higher-level nodes using random sampling; 3. The influence evaluation system according to claim 1 or 2.

4. The label update processing unit: When a calculated value calculated using the influence information for a node above the node to be processed satisfies a condition for a predetermined threshold, the label of the node above is specified as an update label of the node to be processed.

4. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

5. The label update processing unit includes: Calculating the influence of each entity on the nodes above the node being processed using the influence information; If the calculated value satisfies a condition for a predetermined threshold, the label of the node of the calculated value is specified as an update label of the node to be processed.

5. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

6. The label update processing unit: Sort the nodes above the node to be processed according to a predetermined condition for each entity. performing a calculation using the influence information for each entity in the sorted order, and identifying an entity whose calculated value satisfies a condition with respect to a predetermined threshold value; Identifying the label of the node of the identified entity as an update label of the node to be processed; 6. The influence evaluation system according to claim 1,

7. The label update processing unit: If there are multiple nodes of the identified entity, further sorting the nodes in the identified entity according to a predetermined condition; A calculation is performed using the influence information for each node in the sorted order, and a node whose calculated value satisfies a condition for a predetermined threshold is identified; Identifying the identified node as a node to be propagated to the processing target node; 7. The influence evaluation system according to claim 6.

8. The intermediate impact calculation processing unit: not storing the intermediate influence in the node as a history or excluding it from the intermediate influence used in the processing in the index calculation processing unit until the calculation processing of the intermediate influence in the node is executed a predetermined number of times; 8. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

9. The intermediate impact calculation processing unit: When performing the calculation process of the intermediate influence in the node, the intermediate influence in the node is calculated using a damping coefficient.

9. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

10. The influence evaluation system includes: a network information input reception processing unit that receives the input of the network information, If the influence information is information indicating stock ownership, The network information input reception processing unit includes: When the entity holds its own stock, the network information for which the input was received is corrected based on the company's own stock.

10. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

11. The network information input reception processing unit includes: When the entity holds its own stock, the node and the edge are newly created for the network information for which the input was received; performing a correction process in which a label indicating the entity is added to the new edge, and an edge from the new node to the new node is added to the new edge; The influence evaluation system according to claim 10 .

12. The network information input reception processing unit includes: When the entity holds its own company stock, a correction process is performed to calculate influence information of the label of a node having an edge to the entity in the network information received as the input, excluding the entity's own company stock. The influence evaluation system according to claim 10 .

13. The index calculation processing unit includes: When calculating the index indicating the influence, a calculation is performed to weight the size of the entity, 13. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

14. The influence evaluation system includes: Calculating a difference between an index indicating the influence before a change in an entity in the network and an index indicating the influence after a change in an entity in the network; 14. The influence evaluation system according to claim 1, wherein the influence evaluation system is a system for evaluating an influence of a plurality of objects.

15. Computer, a label update processing unit for identifying an upper node of a node to be processed using the edges, in network information in which a label indicating an entity is represented as a node, a direction of influence of the entity is represented as an edge, and information indicating influence that the entity has on other entities is represented as influence information, and for identifying an update label for updating the label of the node to be processed from the label of the upper node using the influence information; an intermediate influence calculation processing unit that specifies a propagation path using the specified update label and calculates an intermediate influence at a node; an index calculation processing unit that calculates an evaluation index indicating the influence of an entity indicated by a label at the node, using the intermediate influence at the node; An impact assessment program characterized by functioning as:

Images