JP6917972B2 - Generation device, generation method, and generation program - Google Patents

Description

The present invention relates to a generator, a generator, and a generator.

Conventionally, a technique for performing clustering based on the presence or absence of edges (connections) between nodes in a graph has been provided. For example, a technique for extracting a cluster structure based on a node in a graph and a set of nodes connected to the node is provided. For example, a technique for locally clustering is provided based on the number of nodes common to a set of nodes connected to a certain node and a set of nodes connected to another node in the graph.

Japanese Unexamined Patent Publication No. 2015-156163

T. Uno et al. Micro-clustering, “Finding small clusters in large diversity.” CoRR, 2015.

However, in the above-mentioned prior art, it is not always possible to appropriately generate a graph that connects subjects (for example, users) on a network based on a desired relationship. For example, when clustering is performed based on the number of nodes common to a set of nodes connected to a certain node and a set of nodes connected to another node, it is determined whether to perform clustering based on the number of common nodes. It becomes. In such a case, it is difficult to properly generate a graph that connects each node based on a desired relationship.

The present application has been made in view of the above, and an object of the present application is to provide a generation device, a generation method, and a generation program for appropriately generating a graph that connects subjects on a network based on a desired relationship. do.

The generator according to the present application is a first graph including a plurality of nodes corresponding to each of the subjects on the network and an edge connecting the nodes having a predetermined correspondence relationship, and the attribute information corresponds to each node. The edge is changed based on the attribute information of the acquisition unit that acquires the attached first graph and the node belonging to a predetermined cluster that classifies the plurality of nodes included in the first graph acquired by the acquisition unit. It is characterized in that it is provided with a generation unit for generating the second graph.

According to one aspect of the embodiment, there is an effect that a graph connecting the subjects on the network based on a desired relationship can be appropriately generated.

FIG. 1 is a diagram showing an example of a generation process according to an embodiment. FIG. 2 is a diagram showing an example of the generation process according to the embodiment. FIG. 3 is a diagram showing a configuration example of the generator according to the embodiment. FIG. 4 is a diagram showing an example of the transaction information storage unit according to the embodiment. FIG. 5 is a diagram showing an example of the first graph information storage unit according to the embodiment. FIG. 6 is a diagram showing an example of the category information storage unit according to the embodiment. FIG. 7 is a diagram showing an example of the attribute information storage unit according to the embodiment. FIG. 8 is a diagram showing an example of the score information storage unit according to the embodiment. FIG. 9 is a diagram showing an example of the second graph information storage unit according to the embodiment. FIG. 10 is a flowchart showing a generation processing procedure according to the embodiment. FIG. 11 is a diagram showing a configuration example of a generator according to a modified example. FIG. 12 is a diagram showing an example of the attribute information storage unit according to the modified example. FIG. 13 is a diagram showing an example of category integration. FIG. 14 is a diagram showing an example of division of categories. FIG. 15 is a diagram showing an example of generation of a second graph in which a plurality of second graphs are integrated. FIG. 16 is a hardware configuration diagram showing an example of a computer that realizes the function of the generator.

Hereinafter, a generator, a generation method, and a mode for carrying out the generation program (hereinafter, referred to as “the embodiment”) according to the present application will be described in detail with reference to the drawings. It should be noted that this embodiment does not limit the generator, the generation method, and the generation program according to the present application. Further, in each of the following embodiments, the same parts are designated by the same reference numerals, and duplicate description is omitted.

(Embodiment)
[1-1. Generation process (1st graph)]
1 and 2 show a case where a buyer (user) in an electronic commerce (for example, Internet shopping) is a node and a graph in which buyers (nodes) are connected by an edge is generated. In FIG. 1, the generation device 100 (see FIG. 3) generates a graph (hereinafter, also referred to as “first graph”) in which buyers are connected by edges based on the commonality of sellers who have purchased goods. Show the case. Further, FIG. 2 shows a case where the generation device 100 generates a graph (hereinafter, also referred to as “second graph”) in which the edges between the nodes included in the first graph are changed.

First, an example of the generation process according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram showing an example of a generation process according to an embodiment. Specifically, FIG. 1 is a diagram showing an example of the generation process of the first graph.

In the example shown in FIG. 1, the generation device 100 generates the first graph G11 based on the correspondence between the buyer and the seller (user) in the electronic commerce as shown in the transaction list TL11 (step S1). Each user may be included in one of the buyer and the seller, or may be included in both. For example, the transaction list TL11 is derived based on the transaction information stored in the transaction information storage unit 121 shown in FIG. As shown in the transaction list TL11, the buyer (node) includes a buyer (node) identified by a buyer ID (node ID) "N11" to "N14", "N21" to "N24", or the like. In the following, the node identified by the buyer ID "N11" will be referred to as node N11. Nodes identified by other buyer IDs are also described. For example, the node identified by the buyer ID "N21" is described as node N21. In the example shown in FIG. 1, only the eight nodes N11 to N14 and N21 to N24 in the transaction list TL11 are shown for explanation, but the transaction list TL11 shows the number of buyers in the electronic commerce. For example, one million nodes may be included.

Further, as shown in the transaction list TL11, the seller includes a seller identified by the seller IDs "SR11" to "SR14", "SR21" to "SR24", and the like. In the following, the seller identified by the seller ID "SR11" will be referred to as the seller SR11. The seller identified by another seller ID is also described in the same manner. For example, the seller identified by the seller ID "SR21" is described as the seller SR21.

In the transaction list TL11, the numerical values of the squares where the nodes N11 to N14, N21 to N24, etc. shown in the vertical row and the sellers SR11 to SR14, SR21 to SR24 shown in the horizontal column intersect are the corresponding buyers. Indicates the number of times the seller has made a commercial transaction. For example, the numerical value "5" in the square (upper left square) where the node N11 and the seller SR11 intersect indicates that the buyer corresponding to the node N11 and the seller corresponding to the seller SR11 have made five transactions. show. Further, for example, the numerical value "10" in the square where the node N22 and the seller SR22 intersect indicates that the buyer corresponding to the node N22 and the seller corresponding to the seller SR22 have made transactions 10 times.

Here, in the generation device 100, among the nodes corresponding to the buyer who is the first entity, the similarity between the set of sellers who are the second entity who have made transactions a predetermined number of times or more is equal to or more than a predetermined threshold value. Generates a first graph in which the nodes connected by edges are connected. In FIG. 1, the generator 100 is a first graph in which among the nodes corresponding to the buyer, the nodes having a Jaccard index of 0.2 or more between the sets of sellers who have made five or more transactions are connected by an edge. Generate G11. That is, FIG. 1 shows a case where the predetermined threshold value is set to 5 times and the Jaccard coefficient is used as an index value indicating the degree of similarity between the seller's sets.

In the first graph G11 shown in FIG. 1, eight nodes N11 to N14 and N21 to N24 are illustrated. Each node N11 to N14 and N21 to N24 correspond to the buyer included in the transaction list TL11. For example, the node N11 in the first graph G11 corresponds to the node (buyer) identified by the buyer ID "N11" in the transaction list TL11. In the example shown in FIG. 1, only eight nodes N11 to N14 and N21 to N24 in the first graph G11 are shown for explanation, but the first graph G11 includes the transaction list TL11. It may include a number of buyers, eg one million nodes.

In the example of FIG. 1, as shown in the transaction list TL11, the buyer corresponding to the node N11 has purchased the product from the seller corresponding to the seller SR11 (hereinafter, also referred to as “the number of transactions”) five times. .. That is, the buyer corresponding to the node N11 is a buyer whose number of purchases (number of transactions) of goods from the seller corresponding to the seller SR11 is equal to or greater than a predetermined threshold value. Further, the buyer corresponding to the node N11 has purchased the product from the seller corresponding to the seller SR13 seven times. That is, the buyer corresponding to the node N11 is a buyer whose number of purchases of goods from the seller corresponding to the seller SR13 is equal to or greater than a predetermined threshold value. That is, with respect to the buyer corresponding to the node N11, the set of sellers whose number of transactions is equal to or greater than a predetermined threshold value (hereinafter, also referred to as "seller set of node N11") includes the seller SR11, the seller SR13, and the like. Further, as shown in the transaction list TL11, the buyer corresponding to the node N13 has purchased the product from the seller corresponding to the seller SR11 six times. That is, the buyer corresponding to the node N13 is a buyer whose number of purchases of goods from the seller corresponding to the seller SR11 is equal to or greater than a predetermined threshold value. Further, the buyer corresponding to the node N13 has purchased the product from the seller corresponding to the seller SR14 seven times. That is, the buyer corresponding to the node N13 is a buyer whose number of purchases of goods from the seller corresponding to the seller SR14 is equal to or greater than a predetermined threshold value. That is, with respect to the buyer corresponding to the node N13, the set of sellers whose number of transactions is equal to or greater than a predetermined threshold value (hereinafter, also referred to as “seller set of node N13”) includes the seller SR11, the seller SR14, and the like. Here, the generation device 100 connects the node N11 and the node N13 with an edge when the Jaccard coefficient between the seller set of the node N11 and the seller set of the node N13 is 0.2 or more. Here, in the above example, the case where the seller included in the seller set of the node N11 is the seller SR11 and the seller SR13 and the seller included in the seller set of the node N13 is the seller SR11 and the seller SR14 will be described as an example. In this case, the Jaccard index (sim) between the seller set of the node N11 and the node N13 is "sim = (the intersection of the seller set of the node N11 and the seller set of the node N13) / (the seller set of the node N11). It is calculated by "sum set with seller set of node N13)". That is, the Jaccard index (sim) between the seller set between the node N11 and the node N13 is "sim = 1/3 (0.33 ...)". Specifically, the denominator of the Jaccard index (sim) between the seller sets of nodes N11 and N13 is the number of elements of the union of the seller set of node N11 and the seller set of node N13, that is, the sellers SR11 and SR13. , SR14 "3". Further, the molecule of the Jaccard coefficient (sim) between the seller set of the node N11 and the node N13 is the number of elements of the intersection of the seller set of the node N11 and the seller set of the node N13, that is, "1" of the seller SR11. Become. As described above, the Jaccard coefficient (sim) between the seller set of the node N11 and the seller set of the node N13 is "0.33 ...", which is equal to or higher than the threshold value "0.2". 100 generates the first graph G11 in which the node N11 and the node N13 are connected by an edge. In order to illustrate the calculation of the Jaccard index (sim), the case where the seller set of the node N11 and the seller set of the node N13 include only two elements is illustrated, but the seller set of the node N11 and the node The seller set of N13 may contain a large number of elements.

Further, in the example of FIG. 1, as shown in the transaction list TL11, the buyer corresponding to the node N21 has purchased the product from the seller corresponding to the seller SR22 six times. Therefore, with respect to the buyer corresponding to the node N21, the set of sellers whose number of transactions is equal to or greater than a predetermined threshold value (hereinafter, also referred to as “seller set of node N21”) includes the seller SR22 and the like. Further, as shown in the transaction list TL11, the buyer corresponding to the node N22 has purchased the product from the seller corresponding to the seller SR22 10 times. Therefore, with respect to the buyer corresponding to the node N21, the set of sellers whose number of transactions is equal to or greater than a predetermined threshold value (hereinafter, also referred to as “seller set of node N22”) includes the seller SR22 and the like. Further, as shown in the transaction list TL11, the buyer corresponding to the node N24 has purchased the product from the seller corresponding to the seller SR22 eight times. Therefore, with respect to the buyer corresponding to the node N21, the set of sellers whose number of transactions is equal to or greater than a predetermined threshold value (hereinafter, also referred to as “seller set of node N24”) includes the seller SR22 and the like. Further, in the example of FIG. 1, it is assumed that the Jaccard coefficient (sim) between the seller set of the node N21 and the seller set of the node N22 is 0.2 or more. Further, it is assumed that the Jaccard coefficient (sim) between the seller set of the node N21 and the seller set of the node N24 is 0.2 or more. Further, it is assumed that the Jaccard coefficient (sim) between the seller set of the node N22 and the seller set of the node N24 is 0.2 or more. As a result, the generation device 100 generates the first graph G11 in which each of the node N21, the node N22, and the node N24 is connected by an edge. In this way, the generation device 100 generates the first graph G11 as shown in FIG. The generation device 100 may generate the first graph based on various indexes, not limited to the Jaccard coefficient. For example, in the generation device 100, among the nodes corresponding to the buyer, the number of sellers included in the intersection of the set of sellers who have made transactions a predetermined number of times or more is equal to or more than a predetermined threshold value (for example, "2"). A first graph in which nodes are connected by edges may be generated.

[1-2. Generation process (second graph)]
Next, an example of the generation process according to the embodiment will be described with reference to FIG. FIG. 2 is a diagram showing an example of the generation process according to the embodiment. Specifically, FIG. 2 is a diagram showing an example of the generation process of the second graph. In FIG. 2, a case where the generation device 100 generates the first graph by the generation process of the first graph shown in FIG. 1 or the first graph has already been acquired from the external device will be described as an example. In the example shown in FIG. 2, the case where the category information of the product purchased by the buyer (user) corresponding to each node is used as the attribute information of the node is shown.

Further, in the example shown in FIG. 2, a case where the second layer of the categories shown in the category information storage unit 123 shown in FIG. 6 is used as the category information of the product is shown. That is, in the example shown in FIG. 2, the generation device 100 uses the category information of the products purchased by the nodes N11 to N14 and N21 to N24 as the attribute information as shown in the attribute information storage unit 124. Specifically, the category C11 in the horizontal column shown in the attribute information storage unit 124 is in the second layer "literary arts" (see FIG. 6) below the category C1 of the first layer "books and magazines". handle. That is, in the example shown in FIG. 2, it is shown that the buyer corresponding to the node N11 has purchased the product belonging to the category C11 of the “literary art” five times. Specifically, the category C21 in the horizontal column shown in the attribute information storage unit 124 is set to the second layer "ladies" (see FIG. 6), which is a lower layer of the category C2 of the "fashion" which is the first layer. handle. That is, in the example shown in FIG. 2, it is shown that the buyer corresponding to the node N22 has purchased the product belonging to the category C21 of the “ladies” seven times. For example, in the category information of the products purchased by the nodes N11 to N14 and N21 to N24, the horizontal columns are represented by k columns, that is, k-dimensional vectors. In this case, the attribute information of each node N11 to N14 and N21 to N24 is represented by a horizontal column of k columns, that is, a k-dimensional vector.

In FIG. 2, the generation device 100 generates the second graph G12 from the first graph G11 (step S2). Here, the generator 100 connects one node (hereinafter, also referred to as “first node”) and another node (hereinafter, also referred to as “second node”) in the first graph G11 with an edge. Calculate the standard score. Hereinafter, a node other than the first node and the second node for which the score is to be calculated may be referred to as the third node. For example, the generator 100 has a similarity between the third node and the first node connected to the first node in the first graph G11, and the third node and the second node connected to the second node in the first graph G11. Calculate the score based on the similarity with the node. For example, the generation device 100 calculates the score in the combination of the first node and the second node by using the following equation (1).

The f (u, v) shown on the left side in the above equation (1) indicates the score in the combination of the node u and the node v. Further, J (U, na) shown on the right side in the above equation (1) indicates the degree of similarity of the attribute information between the node u and the third node (for example, node n). That is, "U" in the above equation (1) indicates a vector of attribute information of the node u (for example, a k-dimensional vector). Further, "na" in the above equation (1) indicates a vector of attribute information of the node n (for example, a k-dimensional vector). Further, J (V, na) shown on the right side in the above equation (1) indicates the degree of similarity of the attribute information between the node v and the third node (for example, node n). That is, "V" in the above equation (1) indicates a vector of attribute information of the node v (for example, a k-dimensional vector). Here, the function J corresponds to, for example, Jackard similarity. For example, the value of the function J is calculated by the following equation (2).

In the above equation (2), for the sake of explanation, J (U, V) is used, and the similarity of the attribute information between the node u and the node v is shown. That is, "U" in the above equation (2) indicates a vector of the attribute information of the node u (for example, a k-dimensional vector), and "V" is a vector of the attribute information of the node v (for example, a k-dimensional vector). Vector) is shown. For example, the formula (2) are shown in the right-hand side u _{i (hereinafter,} also referred to as "first index") indicates a value corresponding to the i-dimensional elements of the attribute information of the node u. For example, u _i, during the information indicating the attribute information storage unit 124, the number of purchases of goods corresponding to the i-dimensional elements of the attribute information of the node u is "1" when it is more than the predetermined reference value, When it is less than a predetermined reference value, it becomes a value such that it is "0". Further, for example, the formula (2) are shown in the right-hand side v _{i (hereinafter,} also referred to as a "second index") indicates a value corresponding to the i-dimensional elements of the attribute information of the node v.

For example, a node u is the node N11, when i is 1, _{u 1} shows one-dimensional element of the attribute information of the node N11, for example, a value corresponding to the category C11. For example, when node u is node N11, i is 1, and a predetermined reference value is 3, u ₁ is one-dimensional of the attribute information of node u in the information shown in the attribute information storage unit 124. Since the number of purchases of an element, for example, a product corresponding to category C11 is 5, which is equal to or higher than a predetermined reference value, it is set to "1". The denominator in the right side of the formula (2) in the respective dimension of the attribute information (e.g., 1 to k dimensions) represents the sum of the first exponent u _i and the second index v _i Tonouchi maximum at. Moreover, molecules in the right-hand side in the formula (2), each dimension of the attribute information (e.g., 1 to k dimensions) represents the sum of the first exponent u _i and the second index v _i Tonouchi minimum at. For example, J (U, V) in the above equation (2) has a value of 0 to 1. Note that the first exponent u _i and the second index v _i, the value of the number of purchases of goods in the attribute information of the corresponding node u and node v may be used.

From here, it will be described by returning to the above equation (1). N (u) in the above equation (1) represents a set of third nodes connected to the node u. The third node connected to the node u here is not limited to the node directly connected to the node u by the edge, but is indirectly connected to the node u via another third node. Is also included. The range of nodes to be included in N (u) may be appropriately set according to the purpose and the like. For example, N (u) in the above equation (1) may include up to a node connected to a node connected to the node u by an edge. N (v) in the above equation (1) indicates a set of third nodes connected to the node v. The range of nodes to be included in N (v) may be appropriately set according to the purpose and the like as in the case of N (u) described above.

The denominator on the right side in the above equation (1) is the attribute similarity between each third node and node u in the union of the third node connected to node u and the third node connected to node v, and each. The sum of the values of the attribute similarity having the maximum value among the attribute similarity between the third node and the node v is shown. The molecule on the right side in the above equation (1) is the attribute similarity between each third node and node u in the intersection of the third node connected to node u and the third node connected to node v, and each. The sum of the values of the attribute similarity with the smallest value among the attribute similarity between the third node and the node v is shown. For example, f (u, v) in the above equation (1) has a value of 0 to 1.

The generation device 100 generates the second graph G12 in which the edge between the first node and the second node is changed based on the score calculated by the above formula (1) and a predetermined threshold value. For example, the generation device 100 releases the edge connecting the nodes in the first graph G11 based on the score calculated by the above equation (1) and the predetermined threshold value, or uses a new edge between the nodes. The second graph G12 is generated by concatenating them.

For example, it is assumed that a predetermined threshold value is 0.5. In this case, for example, when the score calculated by the above equation (1) is 0.5 or more, the generation device 100 connects the first node and the second node corresponding to the score with an edge. For example, when the score of the first node and the second node connected by the edge in the first graph G11 is 0.5 or more, the generator 100 moves between the first node and the second node corresponding to the score. Maintain connecting edges. Further, for example, when the score of the first node and the second node not connected by the edge in the first graph G11 is 0.5 or more, the generation device 100 includes the first node and the second node corresponding to the score. Connect between them with an edge. Further, for example, when the score of the first node and the second node connected by the edge in the first graph G11 is less than 0.5, the generator 100 sets the score of the first node and the second node corresponding to the score. Release the edges that connect between.

In the example of FIG. 2, since the score of the node N11 and the node N12 not connected by the edge in the first graph G11 is 0.8, which is equal to or higher than the threshold value, the generator 100 and the node N11 in the second graph G12. It is connected to the node N12 by an edge. Further, in the example of FIG. 2, since the score of the node N11 and the node N13 connected by the edge in the first graph G11 is 0.7, which is equal to or higher than the threshold value, the generator 100 also has the node in the second graph G12. Maintain an edge connecting N11 and node N13. Further, in the example of FIG. 2, since the score of the node N21 and the node N22 connected by the edge in the first graph G11 is 0.2, which is less than the threshold value, the generator 100 has the node N21 in the second graph G12. The edge connecting between the node N22 and the node N22 is released.

As described above, the generation device 100 is the first node based on the relationship between the attribute information of the first node, the attribute information of the second node, and the attribute information of the third node among the plurality of nodes in the first graph G11. Generate the second graph G12 with the edge changed between the node and the second node. As a result, the generation device 100 can appropriately generate a graph that connects the subjects on the network based on a desired relationship.

Further, for example, the generation device 100 may cluster the nodes based on the creek (maximum creek) included in the second graph G12. For example, the creek here means a group of nodes in which all the nodes included in the creek are connected. The generation device 100 may cluster the nodes by appropriately using a conventional method such as microclustering. For example, the generation device 100 may cluster the nodes by using the creek calculation process described in Non-Patent Document 1. Further, for example, the generation device 100 may use a process (program) for extracting cliques and maximum creeks in the graph, for example, MACE (MAximal Clique Enumerater). The above is an example, and the generation device 100 may cluster the nodes by any means.

For example, the generation device 100 may cluster a group of nodes in which each node is connected by an edge in the second graph G12 as one cluster. For example, in the generation device 100, the node N11, the node N12, and the node N13 included in the second graph G12 are creeks in which all the nodes are connected by edges. For example, the generation device 100 may cluster the nodes N11, the nodes N12, and the nodes N13 included in the second graph G12 as one cluster. Further, for example, the generation device 100 may cluster the nodes N11, the nodes N12, and the nodes N14 included in the second graph G12 as one cluster. Further, in the second graph G12, when the node N13 and the node N14 are connected by an edge, the generator 100 may cluster the node N11, the node N12, the node N13, and the node N14 as one cluster.

In the above-mentioned example, the case where the first graph in which the nodes corresponding to the buyer are connected by an edge is used, but the generation device 100 is not limited to the buyer but connects the nodes corresponding to various entities by an edge. The first graph may be used. For example, the generation device 100 may use a first graph in which nodes corresponding to a subject (user) in a predetermined SNS (Social Networking Service) are connected by an edge. For example, the generation device 100 may also use a social graph showing the correlation of users (humans) on the network (Web) as the first graph. That is, the generation device 100 may use a graph showing the correlation of users on the network as the first graph. The node may include the entire subject (user) included in the social graph, or may include a part of the subject (user) that satisfies a predetermined condition. Further, for example, a node may be associated with various objects, not limited to the user, as long as it is an object (subject) in the real space. For example, when electronic devices can communicate with each other by various technologies related to so-called IoT (Internet of Things), for example, even if each electronic device such as a wearable terminal is regarded as one subject and each electronic device is associated with a node. good. In this case, the concept corresponding to the subject (user) may include various behavioral subjects other than humans, such as an electronic device used by the user and an electronic device that operates autonomously regardless of the operation of the user. As described above, the generation device 100 is not limited to the user, and may generate the second graph by using the first graph in which the nodes associated with various objects (subjects) are connected by edges. .. That is, the generation device 100 may generate the second graph by using the first graph in which users and electronic devices are mixed. For example, the generation device 100 may generate the second graph by using the first graph in which the node corresponding to the user and the node corresponding to the electronic device are connected by an edge. As described above, the generation device 100 may use the first graph in which the nodes corresponding to any subject are connected by edges. That is, the generation device 100 may use any first graph to generate the second graph.

[2. Generator configuration]
Next, the configuration of the generation device 100 according to the embodiment will be described with reference to FIG. FIG. 3 is a diagram showing a configuration example of the generator according to the embodiment. The generation device 100 is an information processing device that generates a second graph based on the first graph. As shown in FIG. 3, the generation device 100 includes a communication unit 110, a storage unit 120, and a control unit 130. The generation device 100 may have a display unit for displaying various information and an input unit for inputting various information.

(Communication unit 110)
The communication unit 110 is realized by, for example, a NIC or the like. Then, the communication unit 110 is connected to a predetermined network by wire or wirelessly, and transmits / receives information to / from an external information processing device.

(Memory unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory (Flash Memory), or a storage device such as a hard disk or an optical disk. As shown in FIG. 3, the storage unit 120 according to the embodiment includes a transaction information storage unit 121, a first graph information storage unit 122, a category information storage unit 123, an attribute information storage unit 124, and a score information storage unit. It has 125 and a second graph information storage unit 126.

(Transaction Information Storage Unit 121)
The transaction information storage unit 121 according to the embodiment stores transaction information. FIG. 4 is a diagram showing an example of the transaction information storage unit according to the embodiment. In the example shown in FIG. 4, log information of transactions in electronic commerce is stored as transaction information. For example, the transaction information stored in the transaction information storage unit 121 is used to generate the first graph. As shown in FIG. 4, the transaction information storage unit 121 has items such as "transaction ID", "buyer ID (node ID)", "seller ID", "product ID", and "date and time" as transaction information. .. The items are not limited to the above, and the transaction information storage unit 121 may have various items depending on the purpose, such as items related to the category to which the traded product belongs and items related to the transaction amount.

The "transaction ID" indicates identification information for identifying each transaction in an electronic commerce transaction. The "buyer ID (node ID)" indicates identification information for identifying the subject (user) who is the buyer in the transaction identified by the corresponding transaction ID. Further, the "seller ID" indicates identification information for identifying the subject (user) who is the seller in the transaction identified by the corresponding transaction ID. In the example shown in FIG. 4, a case where a node ID is used as the buyer ID is shown, but a user ID or a shop ID associated with the node ID may be used. Further, the node ID corresponds to the code of each node shown in FIG. For example, the buyer identified by the node ID "N11" corresponds to the node N11 shown in FIG. That is, the node identified by the node ID "N11" and the node N11 in FIG. 1 indicate the same node, and the same applies to the other nodes shown in FIG. Further, the "product ID" indicates identification information for identifying the product bought and sold in the transaction identified by the corresponding transaction ID. Further, the "date and time" indicates the date and time when the transaction identified by the corresponding transaction ID was completed.

For example, in the example shown in FIG. 4, in the transaction identified by the transaction ID "T11", the buyer corresponding to the node identified by the buyer ID (node ID) "N11" is identified by the seller ID "SR11". Indicates that the seller has purchased the product identified by the product ID "G11" at the date and time "2016/2/22 13:05".

(First Graph Information Storage Unit 122)
The first graph information storage unit 122 according to the embodiment stores information related to the first graph (hereinafter, also referred to as “first graph information” or simply “first graph”). FIG. 5 is a diagram showing an example of the first graph information storage unit according to the embodiment. As shown in FIG. 5, the first graph information storage unit 122 stores, as the first graph information, the presence / absence of an edge connecting the nodes corresponding to the buyer identified by the node ID (buyer ID). For example, if there is an edge connecting the nodes, "1" is stored, and if there is no edge connecting the nodes, "0" is stored.

For example, in the example shown in FIG. 5, in the area where the row of the node ID “N11” and the column of the node ID “N12” intersect, the node corresponding to the buyer identified by the node ID “N11” and the node ID “N11” The presence or absence of an edge with the node corresponding to the buyer identified by "N12" is stored. In the example shown in FIG. 5, "0" is stored in the area where the row of the node ID "N11" and the column of the node ID "N12" intersect. That is, it indicates that the edge is not connected between the node corresponding to the buyer identified by the node ID “N11” and the node corresponding to the buyer identified by the node ID “N12”. Further, in the example shown in FIG. 5, "1" is stored in the area where the row of the node ID "N11" and the column of the node ID "N13" intersect. That is, it indicates that an edge is connected between the node corresponding to the buyer identified by the node ID “N11” and the node corresponding to the buyer identified by the node ID “N13”. In this way, it is shown that the edges are connected between the nodes corresponding to the area in which "1" is stored in the first graph information storage unit 122.

Further, in the example shown in FIG. 5, in the area where the row of the node ID “N12” and the column of the node ID “N11” intersect, the row of the node ID “N11” and the column of the node ID “N12” intersect. The same information as the area to be stored is stored. The generation device 100 is an area where the row of the node ID "N11" and the column of the node ID "N12" intersect or the area where the row of the node ID "N12" and the column of the node ID "N11" intersect. Only one of them may be stored. Note that "0" may be stored in the area where the row of the node ID "N11" and the column of the node ID "N11" intersect. The same applies to other node IDs "N12" and the like. Further, in the example shown in FIG. 5, the case where the node ID and the node ID are stored in the form of a matrix is shown as an example, but if the first graph information can be stored, no matter how it is stored. good. For example, the first graph information may be stored in the form of a node ID having an edge for each node ID, that is, a dictionary that stores a list of node IDs linked to each node ID.

(Category information storage unit 123)
The category information storage unit 123 according to the embodiment stores information related to the category (hereinafter, also referred to as “category information”). FIG. 6 is a diagram showing an example of the category information storage unit according to the embodiment. In the example shown in FIG. 6, information about the category that classifies the traded goods is stored. As shown in FIG. 6, the category information storage unit 123 has items of "first layer", "second layer", and "third layer" as category information. The items are not limited to the above, and the category information storage unit 123 may have items such as "fourth layer" and "fifth layer" depending on the hierarchical structure of the category.

The "first layer" includes category C1 of "books and magazines", category C2 of "fashion", and the like. Further, the "second layer" below the category C1 of the "books and magazines" which is the first layer includes the category C11 of the "literary arts" and the category C12 of the "children". Further, the "third layer" below the category C11 of the "literary arts" which is the second layer includes the category C111 of the "fantasy", the category C112 of the "nonfiction", the category C113 of the "history" and the like. The above category hierarchy is an example, and various category hierarchies may be configured depending on the purpose. Further, the category information stored in the category information storage unit 123 may be changed as appropriate. This point will be described later.

(Attribute information storage unit 124)
The attribute information storage unit 124 according to the embodiment stores information related to the attributes of the buyer (user) (hereinafter, also referred to as “attribute information”). FIG. 7 is a diagram showing an example of the attribute information storage unit according to the embodiment. In the example shown in FIG. 7, the attribute information storage unit 124 stores the category information of the product purchased by the buyer (user) as the attribute information. As shown in FIG. 7, the attribute information storage unit 124 stores the number of purchases for each category of the product purchased by the buyer identified by the node ID (buyer ID) as the attribute information.

For example, in the example shown in FIG. 7, the buyer corresponding to the node N11 has purchased the product belonging to the category C11 of "literary arts" five times and has never purchased the product belonging to the category C12 of "children". show. Further, for example, in the example shown in FIG. 7, the buyer corresponding to the node N12 purchased the product belonging to the category C11 of "literary arts" four times and the product belonging to the category C12 of "children" once. Is shown.

(Score information storage unit 125)
The score information storage unit 125 according to the embodiment stores information regarding the score in the combination of each node (hereinafter, also referred to as “score information”). FIG. 8 is a diagram showing an example of the score information storage unit according to the embodiment. In the example shown in FIG. 8, the score calculated for each combination of the first node and the second node is stored. As shown in FIG. 8, the score information storage unit 125 has items of “buyer ID (node ID)” and “score” as score information. Further, the "buyer ID (node ID)" includes the items of "first node" and "second node". The items are not limited to the above, and the score information storage unit 125 may have various items depending on the purpose.

The "first node" and the "second node" indicate a set of nodes for which the corresponding score is calculated. "Score" indicates the score calculated for the corresponding set of nodes.

For example, in the example shown in FIG. 8, it is shown that the score calculated for a set of nodes of node N11 and node N12 is "0.8". Further, for example, in the example shown in FIG. 8, it is shown that the score calculated for a set of nodes of node N11 and node N13 is "0.7". Further, for example, in the example shown in FIG. 8, it is shown that the score calculated for a set of nodes of node N21 and node N22 is "0.2".

(2nd Graph Information Storage Unit 126)
The second graph information storage unit 126 according to the embodiment stores information related to the second graph (hereinafter, also referred to as “second graph information” or simply “second graph”). FIG. 9 is a diagram showing an example of the second graph information storage unit according to the embodiment. As shown in FIG. 9, the second graph information storage unit 126 stores as the second graph information the presence / absence of an edge connecting the nodes corresponding to the buyer identified by the node ID (buyer ID). Specifically, the second graph information storage unit 126 stores the second graph information generated based on the first graph information stored in the first graph information storage unit 122. For example, if there is an edge connecting the nodes, "1" is stored, and if there is no edge connecting the nodes, "0" is stored.

For example, in the example shown in FIG. 9, in the area where the row of the node ID “N11” and the column of the node ID “N12” intersect, the node corresponding to the buyer identified by the node ID “N11” and the node ID “N11” The presence or absence of an edge with the node corresponding to the buyer identified by "N12" is stored. In the example shown in FIG. 9, "1" is stored in the area where the row of the node ID "N11" and the column of the node ID "N12" intersect. That is, it indicates that an edge is connected between the node corresponding to the buyer identified by the node ID “N11” and the node corresponding to the buyer identified by the node ID “N12”. As described above, the second graph information stored in the second graph information storage unit 126 includes the nodes N11 and N12 whose edges are not connected in the first graph information stored in the first graph information storage unit 122. Indicates that the edge is connected between and.

For example, in the example shown in FIG. 9, in the area where the row of the node ID “N21” and the column of the node ID “N22” intersect, the node corresponding to the buyer identified by the node ID “N21” and the node ID “N21” The presence or absence of an edge with the node corresponding to the buyer identified by "N22" is stored. In the example shown in FIG. 9, "0" is stored in the area where the row of the node ID "N21" and the column of the node ID "N22" intersect. That is, it indicates that the edge is not connected between the node corresponding to the buyer identified by the node ID “N21” and the node corresponding to the buyer identified by the node ID “N22”. As described above, the second graph information stored in the second graph information storage unit 126 includes the node N21 and the node N22 in which the edges are connected in the first graph information stored in the first graph information storage unit 122. Indicates that the edge between is released.

Further, in the example shown in FIG. 9, in the area where the row of the node ID "N12" and the column of the node ID "N11" intersect, the row of the node ID "N11" and the column of the node ID "N12" intersect. The same information as the area to be stored is stored. The generation device 100 is an area where the row of the node ID "N11" and the column of the node ID "N12" intersect or the area where the row of the node ID "N12" and the column of the node ID "N11" intersect. Only one of them may be stored. Note that "0" may be stored in the area where the row of the node ID "N11" and the column of the node ID "N11" intersect. The same applies to other node IDs "N12" and the like. Further, in the example shown in FIG. 9, the case where the node ID and the node ID are stored in the form of a matrix is shown as an example, but if the second graph information can be stored, no matter how it is stored. good. For example, the second graph information may be stored in the form of a node ID having an edge for each node ID, that is, a dictionary that stores a list of node IDs linked to each node ID.

(Control unit 130)
Returning to the description of FIG. 3, the control unit 130 is, for example, a controller, and various programs stored in the storage device inside the generation device 100 by the CPU, MPU, or the like (corresponding to an example of the generation program). Is realized by executing RAM as a work area. Further, the control unit 130 is, for example, a controller, and is realized by an integrated circuit such as an ASIC or FPGA.

As shown in FIG. 3, the control unit 130 includes an acquisition unit 131, a first generation unit 132, a second generation unit 133, and a transmission unit 134, and has functions and operations of information processing described below. Realize or execute. The internal configuration of the control unit 130 is not limited to the configuration shown in FIG. 3, and may be another configuration as long as it is a configuration for performing information processing described later. Further, the connection relationship of each processing unit included in the control unit 130 is not limited to the connection relationship shown in FIG. 3, and may be another connection relationship.

(Acquisition unit 131)
The acquisition unit 131 acquires various types of information. The acquisition unit 131 acquires transaction information, first graph information, and the like. For example, the acquisition unit 131 may acquire transaction information from an external device that provides an electronic commerce service. Further, the acquisition unit 131 stores the transaction information acquired from the external device that provides the electronic commerce service in the transaction information storage unit 121. Further, for example, the acquisition unit 131 acquires the attribute information from the attribute information storage unit 124.

In addition, the acquisition unit 131 acquires the first graph information and the like. For example, the acquisition unit 131 acquires a first graph in which a second subject having a predetermined relationship with each first subject connects nodes corresponding to the first subject satisfying a predetermined condition with an edge. For example, the acquisition unit 131 acquires a first graph in which a set of sellers having a predetermined relationship with each buyer connects the nodes corresponding to the buyers satisfying the predetermined conditions with an edge. For example, the acquisition unit 131 acquires a first graph in which among the nodes corresponding to the buyer, the nodes in which the similarity between the sets of sellers who have made transactions a predetermined number of times or more is equal to or higher than a predetermined threshold value are connected by an edge. do. For example, the acquisition unit 131 acquires a first graph in which nodes having a Jaccard index of 0.2 or more between a set of sellers who have made five or more transactions among the nodes corresponding to the buyer are connected by an edge. do. For example, the acquisition unit 131 acquires the first graph in which the attribute information is associated with each node. For example, the acquisition unit 131 acquires the first graph in which the category information of the product having a predetermined relationship with the user corresponding to each node is associated with each node as the attribute information. For example, the acquisition unit 131 acquires the first graph in which the category information of the product purchased by the user corresponding to each node is associated with each node as the attribute information. For example, the acquisition unit 131 acquires the first graph information from the first graph information storage unit 122. Further, for example, the acquisition unit 131 may acquire the first graph information from an external device. Further, the acquisition unit 131 may store the first graph information acquired from the external device in the first graph information storage unit 122. Further, the acquisition unit 131 may acquire the first graph in which the attribute information of the user corresponding to each node is associated with each node as the attribute information. The acquisition unit 131 acquires the first graph in which the attribute information of the first category and the attribute information of the second category are associated with each node. These points will be described later.

(1st generation unit 132)
The first generation unit 132 generates the first graph based on the transaction information. For example, the first generation unit 132 generates a first graph in which a set of sellers having a predetermined relationship with each buyer connects nodes corresponding to buyers satisfying a predetermined condition with an edge. For example, the first generation unit 132 generates the first graph based on the transaction information stored in the transaction information storage unit 121. In FIG. 1, the first generation unit 132 generates the first graph G11 based on the correspondence between the buyer and the seller (user) in the electronic commerce as shown in the transaction list TL11. For example, the first generation unit 132 connects the nodes corresponding to the buyer with the edges where the similarity between the sets of sellers who have made transactions a predetermined number of times or more is equal to or higher than a predetermined threshold value. To generate. In FIG. 1, the first generation unit 132 connects nodes having a Jaccard index of 0.2 or more between a set of sellers who have made five or more transactions among the nodes corresponding to the buyer at an edge. 1 Graph G11 is generated. For example, the first generation unit 132 generates the first graph G11 in which the node N11 and the node N13 are connected by an edge. For example, the first generation unit 132 generates the first graph G11 in which each of the node N21, the node N22, and the node N24 is connected by an edge.

The generation device 100 does not have to have the first generation unit 132 when acquiring the first graph information from the external device.

(2nd generation unit 133)
The second generation unit 133 is a generation unit that generates a second graph based on the first graph. For example, the second generation unit 133 has the first node and the second node based on the relationship between the first node and the second node and the third node among the plurality of nodes in the first graph acquired by the acquisition unit 131. Generate a second graph with modified edges between the two nodes.

In addition, the second generation unit 133 generates the second graph based on the relationship with the third node in which the first node and the second node are connected by an edge. For example, the second generation unit 133 generates the second graph based on the similarity between the first node and the third node and the similarity between the second node and the third node. For example, the second generation unit 133 generates the second graph based on the number of the third nodes in which the first node and the second node are connected by an edge.

In addition, the second generation unit 133 generates the second graph based on the attribute information of the first node and the relationship between the attribute information of the second node and the attribute information of the third node. Further, for example, the second generation unit 133 generates a second graph based on the first graph in which the category information of the product purchased by the user corresponding to each node is associated with each node as the attribute information.

In FIG. 2, the second generation unit 133 generates the second graph G12 from the first graph G11. For example, the second generation unit 133 includes the similarity between the third node connected to the first node in the first graph G11 and the first node, and the third node connected to the second node in the first graph G11. The score is calculated based on the similarity with the second node. For example, the second generation unit 133 calculates the score in the combination of the first node and the second node by using the above equation (1).

For example, the second generation unit 133 generates a second graph G12 in which the edge between the first node and the second node is changed based on the score calculated by the above equation (1) and a predetermined threshold value. do. For example, the second generation unit 133 releases the edges connecting the nodes in the first graph G11 based on the score calculated by the above equation (1) and a predetermined threshold value, or newly creates the nodes. The second graph G12 is generated by connecting by edges.

For example, in the example of FIG. 2, when the score calculated by the above equation (1) is 0.5 or more, the second generation unit 133 has an edge between the first node and the second node corresponding to the score. Connect with. For example, when the score of the first node and the second node connected by the edge in the first graph G11 is 0.5 or more, the second generation unit 133 sets the first node and the second node corresponding to the score. Maintain the edges that connect between. Further, for example, when the scores of the first node and the second node not connected by the edge in the first graph G11 are 0.5 or more, the second generation unit 133 has the first node and the second node corresponding to the scores. Connect to the node with an edge. Further, for example, when the score of the first node and the second node connected by the edge in the first graph G11 is less than 0.5, the second generation unit 133 has the first node and the second node corresponding to the score. Release the edge that connects to and.

For example, in the second graph G12, since the score of the node N11 and the node N12 not connected by the edge in the first graph G11 is 0.8, which is equal to or higher than the threshold value, the second generation unit 133 has the node N11 and the node N12 in the second graph G12. Connect with the edge at the edge. Further, for example, in the second graph G11, the score of the node N11 and the node N13 connected by the edge in the first graph G11 is 0.7, which is equal to or higher than the threshold value. Maintain an edge connecting to and from node N13. Further, for example, in the second graph G12, since the score of the node N21 and the node N22 connected by the edge in the first graph G11 is 0.2, which is less than the threshold value, the second generation unit 133 has the node N21 and the node in the second graph G12. The edge connecting with N22 is released.

The second generation unit 133 is not limited to the generation of the second graph described above, and may perform various processes as follows. For example, the second generation unit 133 changes the category corresponding to the attribute information based on the attribute information of the nodes belonging to the predetermined cluster included in the second graph. Further, for example, when the second generation unit 133 has a plurality of attribute information common to the nodes belonging to the predetermined cluster included in the second graph, the second generation unit 133 integrates the categories corresponding to the attribute information. Further, for example, when the second generation unit 133 has attribute information common to the nodes belonging to each of the plurality of clusters included in the second graph, the second generation unit 133 divides the categories corresponding to the attribute information. Further, for example, the second generation unit 133 may generate a second graph based on the first graph in which the attribute information of the user corresponding to each node is associated with each node as the attribute information. Further, for example, the second generation unit 133 is based on the relationship between the attribute information of the first category of the first node, the attribute information of the first category of the second node, and the attribute information of the first category of the third node. , Generate a second graph of the first category. Further, for example, the second generation unit 133 is based on the relationship between the attribute information of the second category of the first node, the attribute information of the second category of the second node, and the attribute information of the second category of the third node. , Generate a second graph of the second category. Further, for example, the second generation unit 133 generates a second graph based on the second graph of the first category and the second graph of the second category. These points will be described later.

(Transmission unit 134)
The transmission unit 134 transmits various information to the external device. For example, the transmission unit 134 may transmit the second graph information generated by the second generation unit 133 to an external information processing device.

[3. Generation process flow]
Next, the procedure of the generation process by the generation apparatus 100 according to the embodiment will be described with reference to FIG. FIG. 10 is a flowchart showing a generation processing procedure by the generation device 100 according to the embodiment.

As shown in FIG. 10, the acquisition unit 131 of the generation device 100 acquires transaction information (step S101). For example, the acquisition unit 131 acquires transaction information from the transaction information storage unit 121. Then, the first generation unit 132 of the generation device 100 generates the first graph based on the transaction information acquired by the acquisition unit 131 (step S102). For example, the first generation unit 132 generates the first graph G11.

After that, the acquisition unit 131 acquires the attribute information (step S103). For example, the acquisition unit 131 acquires attribute information from the attribute information storage unit 124. Then, the second generation unit 133 of the generation device 100 calculates a score for each combination of nodes based on the first graph generated in step S102 and the attribute information acquired in step S103 (step S104). For example, the second generation unit 133 calculates the score in the combination of the first node and the second node by using the above equation (1).

Then, the second generation unit 133 generates a second graph based on the score calculated in step S104 (step S105). For example, the second generation unit 133 generates the second graph G12. In addition, step S104 and step S105 may be repeated. For example, the generation device 100 may repeat a process of calculating a score for each combination of nodes, releasing edges between the nodes, and connecting the nodes with edges.

[4. Modification example]
The generator 100 according to the above-described embodiment may be implemented in various different forms other than the above-described embodiment. Therefore, another embodiment of the generator 100 will be described below.

[5. Configuration of generator with other attribute information]
In the above example, the case where the category information of the product purchased by the buyer is used as the attribute information of the user (buyer) is shown, but the attribute information of the buyer is not limited to the category information of the product and various information is used. May be good. That is, various different categories of attribute information may be appropriately used as the attribute information. Regarding this point, the configuration of the generation device 100A according to the modified example will be described with reference to FIG. FIG. 11 is a diagram showing a configuration example of a generator according to a modified example. In the generation device 100A, the same configurations as those of the generation device 100 according to the embodiment are designated by the same reference numerals, and the description thereof will be omitted.

As shown in FIG. 11, the generation device 100A differs from the generation device 100 in that the storage unit 120A includes the attribute information storage unit 124A, the score information storage unit 125A, and the second graph information storage unit 126A. In the example shown in FIG. 11, although the category information storage unit is not shown, the generation device 100A may have a category information storage unit.

(Attribute information storage unit 124A)
The attribute information storage unit 124A according to the modified example stores the attribute information of the buyer (user). FIG. 12 is a diagram showing an example of the attribute information storage unit according to the modified example. In the example shown in FIG. 12, the demographic attribute (demographic attribute) of the buyer is stored as attribute information in the attribute information storage unit 124A. As shown in FIG. 12, the attribute information storage unit 124A stores demographic attributes related to the gender, age, etc. of the buyer identified by the node ID (buyer ID) as the attribute information. The attribute information is not limited to the demographic attribute, and various information may be included depending on the purpose. For example, the attribute information may include information about psychological attributes (psychographic attributes).

For example, in the example shown in FIG. 12, the buyer corresponding to each node in the vertical row corresponds to "1" when it corresponds to each attribute information in the horizontal column, and "0" when it does not correspond to the attribute information. Remember. For example, in the buyer corresponding to the node N11, the attribute information "male" is "0", the attribute information "female" is "1", the attribute information "20's" is "1", and the attribute information "1". "30's" is "0". That is, it indicates that the buyer corresponding to the node N11 is a female user in her twenties.

Further, the score information storage unit 125A and the second graph information storage unit 126A store the score and the second graph calculated based on the attribute information stored in the attribute information storage unit 124A. Omit.

Second generating unit 133 according to the modified example, the first exponent u _i and the second index v _i of the expression (2), during the information indicating the attribute information storage unit 124A, i dimension of the attribute information of the node u The score may be calculated using the value stored in the area of.

For example, when node u is node N11 and i is 1, u ₁ indicates a value corresponding to a one-dimensional element of the attribute information of node N11, for example, the attribute information "male". For example, a node u is the node N11, when i is 1, u _1, during the information indicating the attribute information storage unit 124A, 1-dimensional element attribute information of the node u, for example, the attribute information "male" The corresponding value is "0". Further, for example, when the node u is the node N11 and the i is 2, the u ₂ is a two-dimensional element of the attribute information of the node u in the information shown in the attribute information storage unit 124A, for example, the attribute information “female”. It becomes the value "1" corresponding to. The second generation unit 133 according to the modified example may calculate the score by using the value stored in the attribute information storage unit 124A as described above, and generate the second graph.

The generation device 100A may store the category information corresponding to the attribute information. For example, when the attribute information includes the occupation of the user, the generation device 100A may have a category information storage unit that stores the category information or the like indicating the hierarchical structure of the occupation. Further, in the above-described example, the generation device 100A shows an example in which the storage unit 120A does not have the category information storage unit 123 or the attribute information storage unit 124, but the generation device 100A has the category information storage unit 123 in the storage unit 120A. And the attribute information storage unit 124 may be provided. Further, the generation device 100A may have a score information storage unit 125 and a second graph information storage unit 126 corresponding to the category information storage unit 123 and the attribute information storage unit 124. That is, the generation device 100 according to the embodiment and the generation device 100A according to the modified example may be integrated and used as one generation device.

[6. Change category]
The generation device 100 and the generation device 100A may change the category according to, for example, the result of clustering of the nodes included in the second graph. This point will be described with reference to FIGS. 13 and 14. For example, the following category change processing may be performed by the generation device 100 or the second generation unit 133 of the generation device 100A. In the following, a case where the generation device 100 performs the processing will be shown as an example.

[6-1. Category integration]
First, the integration of categories will be described with reference to FIG. FIG. 13 is a diagram showing an example of category integration. For example, when the second generation unit 133 of the generation device 100 has a plurality of attribute information common to the nodes belonging to the predetermined cluster included in the second graph, the second generation unit 133 integrates the categories corresponding to the attribute information.

In the example shown in FIG. 13, the generator 100 uses the category information as shown in the category information CLT11-1. The "first layer" in the category information CLT11-1 includes the category C3 and the like of the "first category C". Further, the "second layer" below the category C3 of the "first category C" which is the first layer includes the category C31 of the "second category _1" and the like. Further, in the "third layer" lower than the category C31 of the "second category _1" which is the second layer, the category C311 of the "third category _1", the category C312 of the "third category _1", and the "third". Category C313 of "Category_3", category C314 of "Third category_4", and the like are included.

In the example shown in FIG. 13, the generation device 100 clusters each node based on the connection between the nodes in the generated second graph. For example, as shown in the cluster list LT11, the generation device 100 clusters the node N111, the node N112, and the node N113 connected by the edge in the generated second graph as the cluster CL11. Further, for example, the generation device 100 clusters another node group connected by an edge in the generated second graph as a cluster CL12.

In the example shown in FIG. 13, as shown in the cluster list LT11, the node N111, the node N112, and the node N113 belonging to the cluster CL11 are common in that the values of the category C311 and the category C313 are large. Therefore, the generation device 100 integrates the category C311 and the category C313 into one category (step S21). For example, the generation device 100 integrates the category C311 and the category C313 into the category C311-3 of the new category "third category_I-3" as shown in the category information CLT11-2 in FIG.

As a result, the generation device 100 can generate a second graph that reflects the attribute information of the node by using the category information in which the category is changed based on the result of clustering. Therefore, the generation device 100 can perform more appropriate clustering. In the example of FIG. 13, for the sake of simplicity, an example of integrating the lower layers of category C31, which is the same second layer and second classification _1, is shown, but the generator 100 is in the same cluster. Categories with different upper hierarchies may be integrated based on the similarity of the attribute information of the clustered nodes. Further, the above integration is an example, and the generator 100 may integrate the categories based on various conditions. For example, the generator 100 may integrate the two categories if more than half (50%) of the buyers in a cluster purchase the same two categories of goods.

[6-2. Category division]
Next, the division of categories will be described with reference to FIG. FIG. 14 is a diagram showing an example of division of categories. For example, when the second generation unit 133 of the generation device 100 has attribute information common to the nodes belonging to each of the plurality of clusters included in the second graph, the second generation unit 133 divides the categories corresponding to the attribute information.

In the example shown in FIG. 14, the generator 100 uses the category information as shown in the category information CLT21-1. The "first layer" in the category information CLT21-1 includes the category C4 and the like of the "first category C". Further, the "second layer" below the category C4 of the "first category C" which is the first layer includes the category C41 of the "second category _1" and the like. Further, in the "third layer" lower than the category C41 of the "second category _1" which is the second layer, the category C411 of the "third category _1", the category C412 of the "third category _1", and the "third". Category C413 and the like of "Category_3" are included.

In the example shown in FIG. 14, the generation device 100 clusters each node based on the connection between the nodes in the generated second graph. For example, as shown in the cluster list LT21, the generation device 100 clusters the node N211, the node N212, and the node N213 connected by the edge in the generated second graph as the cluster CL21. Further, for example, as shown in the cluster list LT21, the generation device 100 clusters the nodes N221 and the nodes N222 connected by the edges in the generated second graph as the cluster CL22.

In the example shown in FIG. 14, as shown in the cluster list LT21, the nodes N211 to N213 belonging to the cluster CL21 and the nodes N221 to N222 belonging to the cluster CL22 are common in that the value of the category C411 is large. Therefore, the generation device 100 divides the category C411 into two categories (step S26). For example, as shown in the category information CLT21-2 in FIG. 14, the generator 100 has a category C411-1 of the “third category_I-1” and a category C411-2 of the “third category_I-2”. Category C411 is divided into two new categories. For example, in the generator 100, the attribute information of the nodes N211 to N213 belonging to the cluster CL21 is included in the category C411-1 of the "third category_I-1", and the category C411-2 of the "third category_I-2" includes the attribute information. Category C411 is divided so that the attribute information of the nodes N221 to N222 belonging to the cluster CL22 is included.

As a result, the generation device 100 can generate a second graph that reflects the attribute information of the node by using the category information in which the category is changed based on the result of clustering. Therefore, the generation device 100 can perform more appropriate clustering. In the example of FIG. 14, for the sake of simplicity, an example of arranging the divided categories in the lower layers of the category C41, which is the second classification _1 of the same second layer, is shown, but the generator 100 , The divided category may be arranged as a lower layer category of the upper layer different from the upper layer before the division.

[7. ensemble〕
In the above example, the generator 100 and the generator 100A show the case where the second graph is generated based on one attribute information, but the generator 100 and the generator 100A have, for example, the attribute information of a plurality of categories. A second graph may be generated based on. In other words, when the generation device 100 and the generation device 100A have a plurality of attribute information (vectors) in the node, they do not combine them and treat them as one vector, but clustering corresponding to each attribute information. It may be done and integrated to produce one result (second graph or cluster). For example, the process of generating the second graph based on the attribute information of a plurality of categories may be performed by the generation device 100 or the second generation unit 133 of the generation device 100A. This point will be described with reference to FIG. FIG. 15 is a diagram showing an example of generation of a second graph in which a plurality of second graphs are integrated. In the following, a case where the generation device 100 performs the processing will be shown as an example. In the example of FIG. 15, the generation device 100 uses, for example, the category information of the product purchased by the buyer (user) corresponding to each node as the attribute information A of the first category. Further, in FIG. 15, the generation device 100 uses, for example, information on the demographic attribute of the buyer (user) corresponding to each node as the attribute information B of the second category.

In the example shown in FIG. 15, the generator 100 uses the first graph information as shown in the first graph information NL31. For example, in the example shown in FIG. 15, "0" is stored in the area where the row of the node ID "N311" and the column of the node ID "N312" intersect. That is, no edge is connected between the node corresponding to the buyer identified by the node ID "N311" and the node corresponding to the buyer identified by the node ID "N312". Further, in the example shown in FIG. 15, an edge is connected between the node corresponding to the buyer identified by the node ID “N311” and the node corresponding to the buyer identified by the node ID “N313”. ..

In FIG. 15, the generation device 100 generates the second graph using the first graph information NL31 and the attribute information A (step S31). Then, in the example shown in FIG. 15, the generation device 100 clusters each node based on the connection between the nodes in the second graph of the generated attribute information A. For example, as shown in the cluster list LT311, the generation device 100 clusters the node N311, the node N312, and the node N313 connected by the edge in the generated second graph as the cluster CL311.

Further, in FIG. 15, the generation device 100 generates the second graph by using the first graph information NL31 and the attribute information B (step S32). Then, in the example shown in FIG. 15, the generation device 100 clusters each node based on the connection between the nodes in the second graph of the generated attribute information B. For example, as shown in the cluster list LT312, the generation device 100 clusters the nodes N311 and the nodes N312 connected by the edges in the generated second graph as the cluster CL321. Further, for example, as shown in the cluster list LT312, the generation device 100 clusters the nodes N314 and the nodes N315 connected by the edges in the generated second graph as the cluster CL322.

Then, the generation device 100 generates a second graph (hereinafter, also referred to as “integrated second graph”) based on the second graph of the first category and the second graph of the second category (step S33). In FIG. 15, the generation device 100 generates an integrated second graph based on the second graph of the attribute information A and the second graph of the attribute information B. Specifically, the generation device 100 generates an integrated second graph including the clusters shown in the cluster list LT31 based on the clusters shown in the cluster list LT311 and the clusters shown in the cluster list LT312. ..

In the example of FIG. 15, the generation device 100 generates an integrated second graph by a process of obtaining an intersection of a cluster as shown in the cluster list LT311 and a cluster as shown in the cluster list LT312. For example, the generation device 100 generates an integrated second graph in which nodes N311 and nodes N312 are clustered as one cluster CL31 as shown in the cluster list LT31. That is, the generation device 100 generates an integrated second graph in which the node N311 and the node N312 are connected by an edge.

The generator 100 may generate the integrated second graph by a process of obtaining the union of the clusters as shown in the cluster list LT311 and the clusters as shown in the cluster list LT312. In this case, the generation device 100 generates an integrated second graph in which the node N311, the node N312, and the node N313 are clustered as one cluster, and the node N314 and the node N315 are clustered as one cluster. That is, the generation device 100 generates an integrated second graph in which the node N311, the node N312, and the node N313 are connected by an edge, and the node N314 and the node N315 are connected by an edge.

As described above, the generation device 100 can generate the second graph reflecting the plurality of attribute information of each node by generating the integrated second graph using the attribute information of the plurality of categories. Therefore, the generation device 100 can perform more appropriate clustering. When performing the process of integrating the plurality of second graphs described above, the generation device 100 and the generation device 100A include a category information storage unit corresponding to each category, an attribute information storage unit corresponding to each attribute information, and a score information storage unit. Or a second graph information storage unit may be provided. Further, the generation device 100 and the generation device 100A may have an integrated second graph information storage unit that stores an integrated second graph.

The above-mentioned generation of the integrated second graph is an example, and the generation device 100 may generate the integrated second graph by an appropriate method according to various purposes. For example, the generation device 100 sets 0,1 bits for whether or not it belongs to a cluster in the second graph for each attribute information, and adds up the clustering results in the second graph for all attribute information to create a heat map. You may create it. In this case, the generation device 100 may extract an integrated cluster that integrates the clusters of the second graph corresponding to each attribute information using a predetermined threshold value. Further, the generator 100 creates and integrates a heat map by obtaining the center of gravity of the cluster based on the result of clustering in the second graph for each attribute information and assigning a higher score to the node as it is closer to the center of gravity. You may extract the cluster. In this way, for example, the generation device 100 may extract the integrated cluster by using the result of clustering in the second graph of each attribute information. For example, when the generation device 100 overlaps the clusters in the second graph of certain attribute information with the clusters in the second graph of other attribute information and the nodes having a predetermined threshold value or more, the overlapping node group is selected. It may be extracted as an integrated cluster. By the process of extracting such an integrated cluster, the generation device 100 may generate an integrated second graph. The above is an example, and the generation device 100 may generate the integrated second graph by any standard.

[8. effect〕
As described above, the generator 100 according to the embodiment and the generator 100A according to the modified example have the acquisition unit 131 and the generation unit (“second generation unit 133” in the embodiment; the same applies hereinafter). Have. The acquisition unit 131 acquires a first graph including a plurality of nodes corresponding to each of the subjects on the network and an edge connecting the nodes having a predetermined correspondence relationship. The generation unit is located between the first node and the second node based on the relationship between the first node and the second node and the third node among the plurality of nodes in the first graph acquired by the acquisition unit 131. Generates a second graph with modified edges.

As a result, the generator 100 according to the embodiment and the generator 100A according to the modified example are located between the first node and the second node based on the relationship between the first node and the second node and the third node. By generating the second graph with the edge changed, it is possible to appropriately generate a graph that connects the subjects on the network based on the desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the generation unit is based on the relationship with the third node in which the first node and the second node are connected by an edge. Generate a second graph.

As a result, the generator 100 according to the embodiment and the generator 100A according to the modified example are on the network based on the relationship with the third node in which the first node and the second node are connected by an edge. A graph that connects the subjects based on the desired relationship can be appropriately generated.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the generation unit is based on the degree of similarity between the first node and the third node and the degree of similarity between the second node and the third node. To generate the second graph.

As a result, the generation device 100 according to the embodiment and the generation device 100A according to the modified example are based on the degree of similarity between the first node and the third node and the degree of similarity between the second node and the third node. A graph that connects the above subjects based on the desired relationship can be appropriately generated.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the generation unit has a second generation unit based on the number of third nodes in which the first node and the second node are connected by an edge. Generate a graph.

As a result, the generator 100 according to the embodiment and the generator 100A according to the modified example have the main body on the network based on the number of the third nodes in which the first node and the second node are connected by an edge. Graphs that are concatenated based on the desired relationships can be appropriately generated.

Further, in the generator 100 according to the embodiment and the generator 100A according to the modified example, the acquisition unit 131 is a first subject in which the second subject having a predetermined relationship with each first subject satisfies a predetermined condition. Acquire the first graph in which the corresponding nodes are connected by an edge.

As a result, in the generator 100 according to the embodiment and the generator 100A according to the modified example, the second subject having a predetermined relationship with each first subject corresponds to the first subject satisfying the predetermined condition. From the first graph in which the above are connected by an edge, a graph in which the subjects on the network are connected based on a desired relationship can be appropriately generated.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 is subject to a predetermined condition that a set of sellers, which are the second main body, having a predetermined relationship with the buyer, which is the first main body, is a predetermined condition. Acquire the first graph in which the nodes corresponding to the buyers satisfying the above conditions are connected by an edge.

As a result, the generator 100 according to the embodiment and the generator 100A according to the modified example are nodes in which a set of sellers having a predetermined relationship with each buyer corresponds to a buyer who satisfies a predetermined condition among the nodes corresponding to the buyer. From the first graph in which the intervals are connected by edges, it is possible to appropriately generate a graph in which the subjects on the network are connected based on a desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 determines the degree of similarity between the sets of sellers who have made transactions a predetermined number of times or more among the nodes corresponding to the buyer. The first graph in which the nodes that are equal to or greater than the threshold value of are connected by an edge is acquired.

As a result, in the generator 100 according to the embodiment and the generator 100A according to the modified example, the similarity between the set of sellers who have made transactions a predetermined number of times or more among the nodes corresponding to the buyer is equal to or higher than a predetermined threshold value. From the first graph in which the nodes are connected by an edge, a graph in which the subjects on the network are connected based on a desired relationship can be appropriately generated.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 acquires the first graph in which the attribute information is associated with each node. In addition, the generation unit generates the second graph based on the relationship between the attribute information of the first node, the attribute information of the second node, and the attribute information of the third node.

As a result, the generation device 100 according to the embodiment and the generation device 100A according to the modified example are on the network based on the relationship between the attribute information of the first node, the attribute information of the second node, and the attribute information of the third node. It is possible to appropriately generate a graph that connects the subjects in the above based on the desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 associates the category information of the product having a predetermined relationship with the user corresponding to each node as the attribute information with each node. The first graph is acquired.

As a result, in the generator 100 according to the embodiment and the generator 100A according to the modified example, the first graph in which the category information of the product having a predetermined relationship with the user corresponding to each node is associated with each node as the attribute information. Therefore, it is possible to appropriately generate a graph that connects the subjects on the network based on the desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 is the first in which the category information of the product purchased by the user corresponding to each node is associated with each node as the attribute information. Get the graph.

As a result, the generator 100 according to the embodiment and the generator 100A according to the modified example are networked from the first graph in which the category information of the product purchased by the user corresponding to each node is associated with each node as attribute information. A graph that connects the above subjects based on the desired relationship can be appropriately generated.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 acquires the first graph in which the attribute information of the user corresponding to each node is associated with each node as the attribute information. ..

As a result, the generation device 100 according to the embodiment and the generation device 100A according to the modified example identify the subject on the network from the first graph in which the attribute information of the user corresponding to each node is associated with each node as the attribute information. Graphs that are concatenated based on the desired relationships can be appropriately generated.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the generation unit changes the category corresponding to the attribute information based on the attribute information of the nodes belonging to the predetermined cluster included in the second graph. do.

As a result, the generator 100 according to the embodiment and the generator 100A according to the modified example change the category corresponding to the attribute information based on the attribute information of the nodes belonging to the predetermined cluster included in the second graph. , It is possible to appropriately generate a graph that connects subjects on a network based on a desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, when there is a plurality of attribute information common to the nodes belonging to the predetermined cluster included in the second graph, the generation unit corresponds to the attribute information. Integrate the categories you want to do.

As a result, the generation device 100 according to the embodiment and the generation device 100A according to the modified example generate a second graph reflecting the attribute information of the node by using the category information whose category is changed based on the result of clustering. be able to. Therefore, the generators 100 and 100A can perform more appropriate clustering. As a result, the generators 100 and 100A can appropriately generate a graph that connects the subjects on the network based on a desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, if the generation unit has attribute information common to the nodes belonging to each of the plurality of clusters included in the second graph, the attribute information includes the attribute information. Split the corresponding category.

As a result, the generation device 100 according to the embodiment and the generation device 100A according to the modified example generate a second graph reflecting the attribute information of the node by using the category information whose category is changed based on the result of clustering. be able to. Therefore, the generators 100 and 100A can perform more appropriate clustering. As a result, the generators 100 and 100A can appropriately generate a graph that connects the subjects on the network based on a desired relationship.

Further, in the generation device 100 according to the embodiment and the generation device 100A according to the modified example, the acquisition unit 131 provides a first graph in which the attribute information of the first category and the attribute information of the second category are associated with each node. get. Further, the generation unit is based on the relationship between the attribute information of the first category of the first node, the attribute information of the first category of the second node, and the attribute information of the first category of the third node. The second graph is generated, and the second category is based on the relationship between the attribute information of the second category of the first node, the attribute information of the second category of the second node, and the attribute information of the second category of the third node. The second graph of is generated, and the second graph is generated based on the second graph of the first category and the second graph of the second category.

As a result, the generation device 100 according to the embodiment and the generation device 100A according to the modified example reflect the plurality of attribute information of each node by generating the integrated second graph using the attribute information of a plurality of categories. Two graphs can be generated. Therefore, the generators 100 and 100A can perform more appropriate clustering. That is, the generation devices 100 and 100A can appropriately generate a graph that connects the subjects on the network based on a desired relationship.

[9. Hardware configuration]
The generation device 100 according to the above-described embodiment and the generation device 100A according to a modified example are realized by, for example, a computer 1000 having a configuration as shown in FIG. FIG. 16 is a hardware configuration diagram showing an example of a computer that realizes the function of the generator. The computer 1000 has a CPU 1100, a RAM 1200, a ROM 1300, an HDD 1400, a communication interface (I / F) 1500, an input / output interface (I / F) 1600, and a media interface (I / F) 1700.

The CPU 1100 operates based on a program stored in the ROM 1300 or the HDD 1400, and controls each part. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started, a program depending on the hardware of the computer 1000, and the like.

The HDD 1400 stores a program executed by the CPU 1100, data used by such a program, and the like. The communication interface 1500 receives data from another device via a predetermined network N and sends the data to the CPU 1100, and transmits the data generated by the CPU 1100 to the other device via the predetermined network N.

The CPU 1100 controls an output device such as a display or a printer, and an input device such as a keyboard or a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. Further, the CPU 1100 outputs the generated data to the output device via the input / output interface 1600.

The media interface 1700 reads a program or data stored in the recording medium 1800 and provides the program or data to the CPU 1100 via the RAM 1200. The CPU 1100 loads the program from the recording medium 1800 onto the RAM 1200 via the media interface 1700, and executes the loaded program. The recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. And so on.

For example, when the computer 1000 functions as the generator 100 according to the embodiment, the CPU 1100 of the computer 1000 realizes the function of the control unit 130 by executing the program loaded on the RAM 1200. The CPU 1100 of the computer 1000 reads and executes these programs from the recording medium 1800, but as another example, these programs may be acquired from another device via a predetermined network N.

Although some of the embodiments of the present application have been described in detail with reference to the drawings, these are examples, and various modifications are made based on the knowledge of those skilled in the art, including the embodiments described in the disclosure line of the invention. It is possible to practice the present invention in other improved forms.

[10. others〕
Further, among the processes described in each of the above embodiments, all or a part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed. It is also possible to automatically perform all or part of the above by a known method. In addition, the processing procedure, specific name, and information including various data and parameters shown in the above document and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the illustrated information.

Further, each component of each of the illustrated devices is a functional concept, and does not necessarily have to be physically configured as shown in the figure. That is, the specific form of distribution / integration of each device is not limited to the one shown in the figure, and all or part of the device is functionally or physically distributed / physically in arbitrary units according to various loads and usage conditions. Can be integrated and configured. For example, the device for generating the first graph and the device for generating the second graph may be separate bodies.

In addition, the above-described embodiments can be appropriately combined as long as the processing contents do not contradict each other.

Further, the above-mentioned "section, module, unit" can be read as "means" or "circuit". For example, the acquisition unit can be read as an acquisition means or an acquisition circuit.

100 Generator 121 Transaction information storage unit 122 First graph information storage unit 123 Category information storage unit 124 Attribute information storage unit 125 Score information storage unit 126 Second graph information storage unit 130 Control unit 131 Acquisition unit 132 First generation unit 133 2 Generator 134 Transmitter

Claims (21)

A second of the plurality of nodes, which is between a plurality of nodes corresponding to each of the first subjects on the network and a node having a predetermined correspondence relationship and which satisfies a predetermined condition. A first graph including edges connecting nodes whose similarity between a set of subjects is equal to or higher than a predetermined threshold, and an acquisition unit for acquiring a first graph in which attribute information is associated with each node.
A generation unit that generates a second graph with the edge changed based on the attribute information of the nodes belonging to a predetermined cluster that classifies the plurality of nodes included in the first graph acquired by the acquisition unit.
A generator characterized by being equipped with. The generator
Based on the attribute information of the node belonging to the predetermined cluster included in the first graph, the second graph in which the edge is changed is generated based on the category information in which the category corresponding to the attribute information is changed. The generator according to claim 1. The generator
When there is a plurality of attribute information common to the nodes belonging to the predetermined cluster included in the first graph, a second graph in which the edge is changed is generated based on the category information in which the categories corresponding to the attribute information are integrated. The generator according to claim 2, wherein the generator is characterized by the above. The generator
When there is attribute information common to the nodes belonging to each of the plurality of clusters included in the first graph, the second graph in which the edge is changed is displayed based on the category information obtained by dividing the category corresponding to the attribute information. The generator according to claim 2 or 3, wherein the generator is generated. The acquisition unit
Any one of claims 1 to 4, wherein as attribute information, the first graph in which the target category information having a predetermined relationship with the user corresponding to each node is associated with each node is acquired. The generator described in the section. The acquisition unit
The generator according to claim 5, wherein the first graph in which the category information of the product purchased by the user corresponding to each node is associated with each node is acquired as the attribute information. The acquisition unit
The generator according to claim 5, wherein the first graph in which the attribute information of the user corresponding to each node is associated with each node is acquired as the attribute information. The acquisition unit
The generator according to claim 7, wherein the first graph in which the category information of the occupation of the user corresponding to each node is associated with each node is acquired as the attribute information. The acquisition unit
Any one of claims 5 to 8, wherein the first graph in which the category information whose category is changed according to the hierarchical structure for classifying the target as the attribute information is associated with each node is acquired. The generator described in. The generator
Among the plurality of nodes in the first graph acquired by the acquisition unit, between the first node and the second node based on the relationship between the first node and the second node and the third node. The generator according to any one of claims 1 to 9, wherein the second graph is generated by changing the edge of the above. The generator
The generator according to claim 10, wherein the second graph is generated based on the relationship between the first node and the third node connected by an edge. The generator
The eleventh claim is characterized in that the second graph is generated based on the similarity between the first node and the third node and the similarity between the second node and the third node. Generator. The generator
The generator according to claim 12, wherein the second graph is generated based on the number of the third nodes in which the first node and the second node are connected by an edge. The generator
Any of claims 10 to 13, wherein the second graph is generated based on the attribute information of the first node and the relationship between the attribute information of the second node and the attribute information of the third node. The generator according to item 1. The acquisition unit
Acquire the first graph in which the attribute information of the first category and the attribute information of the second category are associated with each node.
The generator
A second graph of the first category is generated based on the relationship between the attribute information of the first category of the first node, the attribute information of the first category of the second node, and the attribute information of the first category of the third node. , Generates the second graph of the second category based on the relationship between the attribute information of the second category of the first node and the attribute information of the second category of the second node and the attribute information of the second category of the third node. The generator according to any one of claims 10 to 14, wherein the second graph is generated based on the second graph of the first category and the second graph of the second category. .. The acquisition unit
Claim 1 is characterized in that a second subject having a predetermined relationship with each first subject acquires the first graph in which nodes corresponding to the first subject satisfying a predetermined condition are connected by an edge. The generator according to any one of 1 to 15. The acquisition unit
A claim characterized in that the information about the seller who is the second entity that has made a transaction with the buyer who is the first entity acquires the first graph in which the nodes corresponding to the buyer satisfying a predetermined condition are connected. Item 16. The generator according to item 16. The acquisition unit
Acquiring the first graph in which a set of sellers who are the second subject having a predetermined relationship with the buyer who is the first subject connects the nodes corresponding to the buyers who satisfy the predetermined conditions at an edge. The generator according to claim 17. The acquisition unit
Among the nodes corresponding to the buyer, the first graph in which the nodes in which the similarity between the sets of sellers who have made transactions a predetermined number of times or more is equal to or higher than a predetermined threshold is connected by an edge is acquired. The generator according to claim 18. It ’s a computer-executed generation method.
A second of the plurality of nodes, which is between a plurality of nodes corresponding to each of the first subjects on the network and a node having a predetermined correspondence relationship and which satisfies a predetermined condition. A first graph including edges connecting nodes whose similarity between a set of subjects is equal to or higher than a predetermined threshold, and an acquisition step of acquiring a first graph in which attribute information is associated with each node.
A generation step of generating a second graph in which the edge is changed based on the attribute information of the nodes belonging to a predetermined cluster that classifies the plurality of nodes included in the first graph acquired by the acquisition step, and a generation step.
A generation method characterized by including. A second of the plurality of nodes, which is between a plurality of nodes corresponding to each of the first subjects on the network and a node having a predetermined correspondence relationship and which satisfies a predetermined condition. A first graph including edges connecting nodes whose similarity between a set of subjects is equal to or higher than a predetermined threshold, and an acquisition procedure for acquiring a first graph in which attribute information is associated with each node.
A generation procedure for generating a second graph in which the edge is changed based on the attribute information of a node belonging to a predetermined cluster that classifies a plurality of nodes included in the first graph acquired by the acquisition procedure, and a generation procedure.
A generator characterized by having a computer execute.

Images