JP7122120B2 - Information processing device, information processing method, and information processing program - Google Patents

Description

The present invention relates to an information processing device, an information processing method, and an information processing program.

Conventionally, techniques for extracting various information have been provided. For example, techniques are provided for estimating similarities between concepts in consideration of relative relationships between concepts belonging to other categories.

Japanese Patent Application Laid-Open No. 2007-213151

Masajiro Iwasaki, "Neighborhood Retrieval Using Approximate k-Nearest Neighbor Graph Using Tree Structured Index", Transactions of Information Processing Society of Japan, 2011/2, Vol. 52, No. 2. pp.817-828.

However, with the conventional technology described above, it may be difficult to appropriately extract an object having a predetermined correspondence relationship with a desired object. For example, it is difficult to extract desired information only by estimating similarities between concepts belonging to other categories. For example, it is not always possible to estimate the similarity of information belonging to the same category, and in such cases, it is difficult to extract desired information.

The present application has been made in view of the above, and aims to provide an information processing apparatus, an information processing method, and an information processing program for appropriately extracting an object having a predetermined correspondence relationship with a desired object. .

An information processing apparatus according to the present application includes graph information in which a plurality of nodes corresponding to each of a plurality of objects having people as elements are connected according to the similarity of the plurality of objects, and object information related to one object. an acquiring unit for acquiring, from among the plurality of nodes of the graph information acquired by the acquiring unit, a starting node determined based on a predetermined criterion as a starting point for searching the graph information; and a search unit that extracts, from among the plurality of targets, a corresponding target that has a predetermined correspondence relationship with the one target by searching for information.

According to one aspect of the embodiment, it is possible to appropriately extract an object having a predetermined correspondence relationship with a desired object.

FIG. 1 is a diagram illustrating an example of information processing according to an embodiment. FIG. 2 is a diagram illustrating a configuration example of an information processing system according to the embodiment; FIG. 3 is a diagram illustrating a configuration example of an information processing apparatus according to the embodiment; 4 is a diagram illustrating an example of a target information storage unit according to the embodiment; FIG. FIG. 5 is a diagram illustrating an example of an index information storage unit according to the embodiment; 6 is a diagram illustrating an example of a graph information storage unit according to the embodiment; FIG. 7 is a diagram illustrating an example of a model information storage unit according to the embodiment; FIG. FIG. 8 is a flowchart illustrating an example of information processing according to the embodiment. FIG. 9 is a flowchart illustrating an example of generation processing according to the embodiment. FIG. 10 is a diagram illustrating an example of feature amount extraction according to the embodiment. FIG. 11 is a flowchart illustrating an example of search processing using graph information. FIG. 12 is a diagram conceptually showing group graph information. FIG. 13 is a hardware configuration diagram showing an example of a computer that implements the functions of the information processing apparatus.

Hereinafter, modes for implementing an information processing apparatus, an information processing method, and an information processing program (hereinafter referred to as "embodiments") according to the present application will be described in detail with reference to the drawings. The information processing apparatus, information processing method, and information processing program according to the present application are not limited to this embodiment. Also, in each of the following embodiments, the same parts are denoted by the same reference numerals, and overlapping descriptions are omitted.

(embodiment)
[1. information processing]
An example of information processing according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of information processing according to an embodiment. In FIG. 1, the information processing apparatus 100 retrieves graph data (graph information) in which a plurality of individuals (one person) are graph-structured, as an example of an object whose element is a person. (also referred to as "similar target") is extracted. The object is not limited to an individual (one person), and may be various objects as long as they are human elements. In FIG. 1, the information processing apparatus 100 uses vector data (also referred to as “vector information” or simply “vector”) corresponding to each object to use graph information in which objects are graph-structured. .

The information used by the information processing apparatus 100 is not limited to vectors, and may be information of any format as long as it is information capable of expressing the similarity of each object whose element is a person. For example, the information processing apparatus 100 may use graph information in which objects are graph-structured using predetermined data and values corresponding to each object. For example, the information processing apparatus 100 may use graph information in which an object is graph-structured using a predetermined numerical value (for example, a binary value or a hexadecimal value) generated from each object whose element is a person. . For example, instead of vectors, any form of data may be used as long as the distance (similarity) between data is defined. In the example of FIG. 1, each of the objects whose elements are people is an individual (one person). may be a group (user group) including a plurality of people. That is, the target (object) may be any target as long as it has a person as an element.

[1-1. About graph information]
The information processing apparatus 100 also performs information processing using graph data in which vectors (nodes) are connected by directed edges, as indicated by graph information GR11 in FIG. Graph information as indicated by graph information GR11 in FIG. It may be acquired from another external device such as the providing device 50 (see FIG. 2).

Also, the term "directed edge" used herein means an edge that allows data to be traced in only one direction. In the following description, the source traced by the edge, that is, the node that is the starting point, is the reference source, and the destination that is traced by the edge, that is, the node that is the end point, is the reference destination. For example, a directed edge connecting a predetermined node "A" to a predetermined node "B" indicates an edge having a reference source of node "A" and a reference destination of node "B". Edges connecting nodes are not limited to directed edges, and may be various edges. For example, the edge connecting each node may be a directionless edge connecting the nodes. For example, the edges connecting each node may be mutually referable edges. For example, the edges connecting each node may all be bidirectional edges.

For example, an edge having node "A" as a reference in this way is called an output edge of node "A". Further, for example, an edge having the node "B" as a reference in this way is referred to as an input edge of the node "B". In other words, the term "output edge" and "input edge" as used herein refers to the difference in which of the two nodes to which one directed edge connects is regarded as the center. are the output and input edges. In other words, an output edge and an input edge are relative concepts. For one directed edge, it is an output edge when the reference node is regarded as the center, and the reference node is regarded as the center. is an input edge. Note that, in the present embodiment, directed edges such as output edges and input edges are targeted for edges, and therefore directed edges may be simply referred to as "edges" below.

For example, the information processing apparatus 100 processes nodes corresponding to a huge number of objects (individuals and groups) in units of millions to hundreds of millions, but only some of them are shown in the drawings. In the example of FIG. 1, to simplify the explanation, eight nodes are illustrated to explain the outline of the processing. For example, the information processing apparatus 100 acquires graph information including a plurality of nodes (vectors) such as nodes N1, N2, N3, etc., as indicated by graph information GR11 in FIG. Also, in the example of FIG. 1, each node in the graph information GR11 is connected to a predetermined number of edges (output edges) from the node closest to the node. For example, the predetermined number may be various values such as 2, 5, 10, 100, etc., depending on the purpose and application. For example, if the predetermined number is 2, output edges from the node N1 are connected to the node closest to the node N1 and the two nodes closest to the node N1. Note that the distance as an index indicating similarity may be any distance as long as it can be applied as the distance between vectors (N-dimensional vectors). Various distances may be used.

Also, when "node N* (* is an arbitrary number)" is described in this way, it indicates that the node is identified by the node ID "N*". For example, when "node N1" is described, the node is identified by the node ID "N1".

In graph information GR11 in FIG. 1, node N10 is connected to edge E7, which is a directed edge toward node N7. That is, node N10 is connected to node N7 by edge E7. When "edge E* (* is an arbitrary numerical value)" is described in this way, it indicates that the edge is identified by the edge ID "E*". For example, when "edge E11" is described, the edge is identified by the edge ID "E11". For example, an edge E7 connecting the node N10 as a reference source and the node N7 as a reference destination enables tracing from the node N10 to the node N7. In this case, the directed edge E7 is an output edge when identified with the node N10 as the center, and an input edge when identified with the node N7 as the center. Also, the bidirectional arrows in the graph information GR11 of FIG. 1 indicate that directed edges from both nodes to the other node are connected. For example, a two-way arrow between node N2 and node N451 in graph information GR11 connects two edges: a directed edge from node N2 to node N451 and a directed edge from node N451 to node N2. indicates that

Further, the graph information GR11 in FIG. 1 may be Euclidean space. Graph information GR11 shown in FIG. 1 is a conceptual diagram for explaining distances between vectors, and the graph information GR11 is a multidimensional space. For example, the graph information GR11 shown in FIG. 1 is illustrated in a two-dimensional manner because it is illustrated on a plane, but it is assumed to be a multi-dimensional space such as 100-dimensional space or 1000-dimensional space.

Here, the distance between vector data indicates the similarity of objects, and the closer the distance, the more similar. In this embodiment, the distance of each node in the graph information GR11 is used as the degree of similarity between corresponding objects. For example, it is assumed that the similarity of the object (individual) corresponding to each node is mapped as the distance between nodes in the graph information GR11. For example, it is assumed that the similarity between concepts corresponding to each node is mapped to the distance between each node. Here, in the example shown in FIG. 1, the similarity between objects with short distances between nodes in the graph information GR11 is high, and the similarity between objects with long distances between nodes in the graph information GR11 is low. For example, in the graph information GR11 in FIG. 1, the node identified by the node ID "N35" and the node identified by the node ID "N693" are close to each other, that is, the distance between them is short. Therefore, it indicates that the object corresponding to the node identified by the node ID "N35" and the object corresponding to the node identified by the node ID "N693" have a high degree of similarity.

Further, for example, in the graph information GR11 in FIG. 1, the node identified by the node ID "N7" and the node identified by the node ID "N2" are remote, that is, the distance is long. Therefore, the similarity between the object corresponding to the node identified by the node ID "N7" and the object corresponding to the node identified by the node ID "N2" is low.

[1-2. Vector generation example]
Each node (vector) here corresponds to each object (target). If the target information is numerical data indicating the states of various targets (individuals and groups), it can be treated as it is as vector data. However, if the target information is text, image, or voice, conversion to vector data is required. Therefore, in the example of FIG. 1, the object may be a multi-dimensional (N-dimensional) vector generated from the feature amount extracted from the target information (personal information) of each target (individual).

For example, the information processing apparatus 100 may generate an N-dimensional vector from the target information of each target using a model for extracting features of the target information. In the example of FIG. 1, the information processing apparatus 100 uses the model (model M1) identified by the model ID "M1" as shown in the model information storage unit 124 (see FIG. 7) to obtain the target information of each target. Generate a vector from . As described above, when "model M* (* is an arbitrary numerical value)" is described, it indicates that the model is identified by the model ID "M*". For example, when "model M1" is described, the model is identified by the model ID "M1". Further, as shown in the model information storage unit 124, the model M1 is a model used for use "feature extraction (individual)", that is, for feature extraction from personal data (target information). It indicates that the data is "model data MDT1".

For example, by inputting target information of an arbitrary target (individual) into the model M1, the information processing apparatus 100 calculates the value of each element (neuron) in the model M1, and obtains information similar to the input target information. Output. For example, the information processing apparatus 100 may extract the value of each element (neuron) of the intermediate layer as a feature amount, and generate N-dimensional vector data corresponding to each target (individual).

Here, an example of generation of vector data corresponding to each target is shown using FIG. FIG. 10 is a diagram illustrating an example of feature amount extraction according to the embodiment. FIG. 10 is a conceptual diagram of the model M1. Note that in FIG. 10, illustration of lines indicating respective connection relationships of respective elements (neurons) is omitted. As shown in FIG. 10, the model M1 includes an input layer IL, an intermediate layer CL, and an output layer OL. For example, the input layer IL of the model M1 is a layer into which target information is input. Also, the output layer OL is a layer that outputs information similar to the input target information in response to input to the input layer IL.

Also, for example, the most compressed layer RP in the central portion of the intermediate layer CL is a layer that expresses the characteristics of the input target information. For example, in the intermediate layer CL of the model M1, the portion from the input layer IL to the compression layer RP corresponds to the portion where the encoding process is performed. In the intermediate layer CL of the model M1, the portion from the input layer IL to the compression layer RP corresponds to a portion for performing processing for compressing the features of the input target information. For example, in the intermediate layer CL of the model M1, the portion from the compression layer RP to the output layer OL corresponds to a portion that performs decoding processing. In the intermediate layer CL of the model M1, the portion from the compression layer RP to the output layer OL corresponds to a portion that performs processing for restoring compressed target information.

For example, the information processing apparatus 100 may use the information of the neuron NL1, the neuron NL2, etc. included in the compression layer RP as the vector. For example, when target information of a certain target (individual or group) is input, the information processing apparatus 100 converts the calculated value VE1 corresponding to the neuron NL1 and the calculated value VE2 corresponding to the neuron NL2 to the elements of the vector (one dimension value). For example, when target information of a certain target is input, the information processing apparatus 100 may extract the calculated value VE1 corresponding to the neuron NL1 as the first-dimensional element of the target vector. Further, for example, when target information of a certain target is input, the information processing apparatus 100 may extract the calculated value VE2 corresponding to the neuron NL2 as the second-dimensional element of the target vector. In this way, the information processing apparatus 100 may generate a vector corresponding to each target by inputting target information of each target into the model M1. Note that the information processing apparatus 100 may acquire vectors corresponding to each target from another external apparatus such as the information providing apparatus 50 . Note that the information processing apparatus 100 may use the value itself corresponding to each neuron, or may use a value obtained by multiplying the value corresponding to each neuron by a predetermined coefficient, as each element of the vector. In the example of FIG. 1, each element (value) of the vector is an integer to simplify the explanation. good too.

Note that the information processing apparatus 100 may use information of not only the elements (neurons) of the compression layer RP but also other elements (neurons) in the intermediate layer CL for the vectors. For example, the information processing apparatus 100 may use the information of the encoding part neuron NL3, the decoding part neuron NL4, or the like for the vector. For example, when target information of a certain target is input, the information processing apparatus 100 uses the calculated value VE3 corresponding to the neuron NL3 and the calculated value VE4 corresponding to the neuron NL4 as vector elements (one-dimensional values). may be extracted. Note that the above is just an example, and the information processing apparatus 100 may perform feature extraction from target information using various models, not limited to autoencoders. For example, it is possible to generate a model that has been trained to distinguish targets such as start-up members that are suitable for a given purpose and targets that are not suitable, and extract the intermediate layer as vector data. Alternatively, a model may be generated by a similarity learning method such as triplet loss. Also, the information processing apparatus 100 may perform feature extraction without using a model. For example, the information processing apparatus 100 may extract information corresponding to features (features) set by an administrator or the like of the information processing apparatus 100 from the target information and generate a vector. For example, the information processing apparatus 100 extracts, from the target information, attribute information such as the age and gender of a target (individual), and information corresponding to features (features) such as images, voices, and behaviors of the face and the whole image, and extracts vector may be generated.

Further, for example, the information processing device 100 may acquire the model M1 from another external device such as the information providing device 50 . The information processing apparatus 100 may generate the model M1 by inputting the target information TDT1, TDT2, TDT451, etc. of each target stored in the target information storage unit 121 (see FIG. 4). For example, the information processing apparatus 100 may generate the model M1 by inputting the target information TDT1, TDT2, TDT451, etc. of each target. For example, the target information TDT1, TDT2, TDT451, etc. may include attribute information, images, sounds, actions, and the like of the target (individual). For example, target information such as target information TDT1, TDT2, and TDT451 may include various attribute information about the user, such as corresponding individual demographic attribute information and psychographic attribute information. For example, target information such as target information TDT1, TDT2, and TDT451 may include information such as the age, gender, address, and interests of the corresponding individual. Also, for example, the target information may include image information obtained by imaging the corresponding individual. For example, the target information may include image information obtained by imaging the corresponding individual's face or entire body (whole body). For example, the target information may include audio information such as the corresponding individual's speech. For example, the subject information may include various behavioral information regarding the behavior of the corresponding individual. For example, the target information may include action information such as search, content viewing, and purchase of the corresponding individual. Note that the target information such as the target information TDT1, TDT2, and TDT451 may be only career information (character information) related to work such as the career of the target (individual).

The information processing apparatus 100 includes an input layer to which target information of a target is input, an output layer, a first element belonging to any layer from the input layer to the output layer other than the output layer, and a first layer. and a second element whose value is calculated based on the weight of the element and the first element. By performing an operation based on the first element and the weight of the first element, the model M1 may be generated that outputs from the output layer the same information as the information input to the input layer. Note that the information processing apparatus 100 may acquire a model used for vector generation from another external apparatus such as the information providing apparatus 50 or the like.

[1-3. Processing example]
From here, a case where the information processing apparatus 100 acquires target information of one target (individual) from a user and provides information on a target (individual) similar to the one target will be described as an example. In the example of FIG. 1, the terminal device 10 is used by a user U1 who is a predetermined administrator. For example, the user U1 may be a user who conducts business such as organizational organization in a company. The example of FIG. 1 shows a case where user U1 desires information about which individual (person) a target individual (hereinafter referred to as “individual X”) is similar to.

First, the terminal device 10 that has acquired the target information of the individual X transmits the target information TDT11 of the individual X to the information processing device 100 . In the example of FIG. 1, a user U1 who organizes a team (organizational organization) for a project startup in a company wishes to add an individual similar to the individual X as a member of the team. In the example of FIG. 1, individual X is assumed to be an individual who has made significant contributions to projects in other companies. In addition, the individual X may be an individual of the company to which the user U1 belongs. For example, person X may be user U1. Note that when the individual X is the user U1, the user U1 transmits personal information (target information) of the user U1 to the information processing apparatus 100 . For example, after the user U1 collects the target information TDT11 of the person X, the user U1 operates the terminal device 10 so as to transmit the target information TDT11 of the person X to the information processing apparatus 100 . For example, the terminal device 10 acquires the target information TDT11 of the individual X of the individual X from a predetermined information processing device, and transmits the acquired target information TDT11 to the information processing device 100 . In the example of FIG. 1, the target information TDT11 and the like include attribute information such as the age and sex of the target (individual), and information related to images, voices, actions, and the like obtained by imaging the face and the whole. Note that the target information TDT11 and the like may be only career information (character information) related to work such as the career of the target (individual).

Then, the information processing apparatus 100 acquires target information (personal information) regarding one target (step S11). In the example of FIG. 1 , the information processing device 100 acquires target information TDT11 regarding the person X from the terminal device 10 .

Then, the information processing apparatus 100 generates a vector used for searching graph information from target information corresponding to one target. In the example of FIG. 1, the information processing apparatus 100 generates a vector corresponding to the individual X by performing the processing shown in the processing group PS11. The information processing device 100 inputs the target information TDT11 regarding the individual X to the model M1 (step S12). Specifically, the information processing apparatus 100 inputs the target information TDT11 of the individual X to the model M1. Then, the information processing apparatus 100 generates a vector using information in the model M1 after input of the target information TDT11 (step S13). For example, the information processing apparatus 100 generates vector data using each element in the model M1 to which the target information TDT11 is input.

In the example of FIG. 1, the information processing apparatus 100 generates vector data VD11 (also simply referred to as "vector VD11") using the values of the elements in the model M1 to which the target information TDT11 is input. For example, the information processing apparatus 100 calculates a value VE1 (see FIG. 10) corresponding to the neuron NL1 and a value VE2 (see FIG. 10) corresponding to the neuron NL2 of the model M1 when the target information TDT11 of the person X is input. to generate a vector. For example, the information processing apparatus 100 may extract the calculated value VE1 corresponding to the neuron NL1 as the first-dimensional element of the vector VD11 when the target information TDT11 of the individual X is input. Further, for example, when target information of a certain target is input, the information processing apparatus 100 generates a vector VD11 by using the calculated value VE2 corresponding to the neuron NL2 as the second-dimensional element of the vector VD11. In the example of FIG. 1, the information processing apparatus 100 generates a vector VD11 whose first-dimensional element is "35" and whose second-dimensional element is "63."

Then, the information processing device 100 searches for a target (similar target) similar to the individual X (step S14). For example, the information processing apparatus 100 may search for a similar target of the person X by appropriately using various conventional techniques such as the neighborhood search technique disclosed in Non-Patent Document 1. For example, the information processing apparatus 100 uses graph information GR11 in which nodes with high similarity are connected by edges.

In the example of FIG. 1, the information processing apparatus 100 searches for a similar target of the person X using the graph information GR11 and the index information IND11, as shown in the information group INF11. For example, the information processing apparatus 100 acquires graph information GR11 regarding an object (individual) from the graph information storage unit 123 (see FIG. 6). Further, for example, the information processing apparatus 100 stores index information IND11 used for determining a node (hereinafter also referred to as "origin vector") serving as a starting point for searching in the graph information GR11 from the index information storage unit 122 (see FIG. 5). get. Note that the index information IND11 may be generated by the information processing device 100, or the information processing device 100 may acquire the index information IND11 from another external device such as the information providing device 50 or the like.

Then, the information processing apparatus 100 uses the index information IND11 to determine (specify) the origin vector corresponding to one target (query). In the example of FIG. 1, the information processing apparatus 100 uses the index information IND11 to determine (identify) the origin vector corresponding to the person X's vector VD11. That is, the information processing apparatus 100 uses the vector VD11 and the index information IND11 to determine the origin vector in the graph information GR11.

The index information IND11 in FIG. 1 has a hierarchical structure shown in the index information storage section 122 in FIG. For example, the index information IND11 indicates that the nodes (vectors) in the first layer located directly below the root RT are the nodes VT1, VT2, and the like. Also, for example, the index information IND11 indicates that the nodes of the second layer immediately below the node VT2 are the nodes VT2-1 to VT2-4 (not shown). Also, for example, the index information IND11 indicates that the nodes of the third layer immediately below the node VT2-2 are the node N35, the node N451, and the node N693, that is, nodes (vectors) in the graph information GR11.

For example, the information processing apparatus 100 determines the origin vector in the graph information GR11 using tree-structured index information as indicated by the index information IND11 in FIG. 1 (step S15). In the example of FIG. 1, after generating the vector VD11, the information processing apparatus 100 traces the index information IND11 from top to bottom to identify the origin vector that is a candidate for the vicinity of the index information IND11. An origin vector corresponding to a search query (one subject) can be determined.

For example, the information processing apparatus 100 may determine the origin vector corresponding to the vector VD11 by tracing the index information IND11 from the root RT to leaf nodes (nodes (vectors) in the graph information GR11). In the example of FIG. 1, for example, the information processing apparatus 100 determines the node N451 as the origin vector by tracing the index information IND11 from the root RT to the node N451. For example, the information processing apparatus 100 may determine the reached leaf node as the origin vector by tracing the index information IND11 from the root RT to the leaf node using various conventional techniques related to the tree structure. For example, the information processing apparatus 100 may determine the origin vector by tracing down the index information IND11 based on the degree of similarity with the vector VD11. For example, the information processing apparatus 100 may determine which of the nodes VT1, VT2, etc. is followed from the root RT based on the degree of similarity between the vector VD11 and the nodes VT1, VT2. For example, the information processing apparatus 100 may determine to trace from the root RT to the node VT2, which has the highest degree of similarity with the vector VD11, among the nodes VT1, VT2, and the like. Further, for example, the information processing apparatus 100 may determine to trace from the node VT2 to the node VT2-2 having the highest degree of similarity with the vector VD11 among the nodes VT2-1 to VT2-4. Further, for example, the information processing apparatus 100 may determine to trace from the node VT2-2 to the node N451, which has the highest degree of similarity with the vector VD11 among the nodes N35, N451, N693, and the like. Note that the example of FIG. 1 shows a case where one origin vector is determined for the sake of simplicity of explanation, but the information processing apparatus 100 may determine a plurality of origin vectors. For example, the information processing apparatus 100 may determine a plurality of vectors (nodes) such as nodes N451, N35, N693, and N2 as origin vectors. Instead of using the index information IND11, one or more nodes may be randomly selected from the graph information GR11 at the start of the search and used as the starting vector, or one or more nodes specified in advance may be used as the starting point. It can also be a vector.

Then, the information processing apparatus 100 extracts a similar target of the individual X by searching the graph information GR11 (step S16). For example, the information processing apparatus 100 extracts nodes located near the node N451 as similar targets. For example, the information processing apparatus 100 extracts a node that is close to the node N451 as a similar target. For example, the information processing apparatus 100 extracts nodes reachable from the node N451 as similar objects by tracing edges starting from the node N451. For example, the information processing apparatus 100 extracts a predetermined number (for example, 2 or 10) of nodes as similar targets. For example, the information processing apparatus 100 may extract a similar target of the person X through a search process as shown in FIG. 11, which will be described later in detail. In the example of FIG. 1, the information processing apparatus 100 extracts the node N451 and the node N35 as similar objects by searching the graph information GR11 with the node N451 as a starting point.

Then, the information processing apparatus 100 provides information on the extracted similar target (step S17). In the example of FIG. 1, the information processing apparatus 100 treats the individual #451 corresponding to the node N451 and the individual #35 corresponding to the node N35 as targets (similar individuals) similar to the individual X. provide to

As described above, the information processing apparatus 100 uses the graph information GR11 and the index information IND11 to extract similar objects of the individual X acquired from the user U1. For example, the information processing apparatus 100 uses the index information IND11 to determine the origin vector in the graph information GR11 when extracting the similar target of the person X. FIG. Then, the information processing apparatus 100 extracts a similar target of the individual X by searching the graph information GR11 with the determined starting vector as a starting point. Accordingly, the information processing apparatus 100 can appropriately extract targets having a predetermined correspondence relationship with a desired target.

For example, even if an object (similar individual) similar to individual X does not seem to be related to individual X at first glance, the information processing apparatus 100 uses the graph information to identify it as a similar individual to individual X. It is possible to appropriately extract an object that can be extracted and that has a predetermined correspondence with a desired object. For example, there is a method of identifying targets suitable for a given purpose, such as a business or startup member, using a method such as deep learning, but if there are a wide variety of factors that affect whether it is suitable for a given purpose, Since there are few examples of each element, there is a problem that learning cannot be performed and it is difficult to improve the identification accuracy. For example, there is a method of using a model generated by a learning method such as deep learning, inputting target information into the model, and identifying whether the target is suitable for a given purpose from the output score. When such elements exist, it is difficult to improve the accuracy of the model because there are few examples of each element, and it is difficult to improve the accuracy of identification. On the other hand, the information processing apparatus 100 appropriately extracts similar objects similar to a certain object by performing a search using graph information in which a plurality of vectors are graph-structured and generated based on features extracted from each element. can do. Therefore, the information processing apparatus 100 can accurately identify for what purpose a certain object is similar to a suitable object.

[1-4. About the target]
In the example of FIG. 1, an individual (single person) is shown as an example of an object whose element is a person. It may be a target to be included as For example, the information processing apparatus 100 may extract a similar target from a group (user group) including a plurality of persons as a target. In this case, the information processing apparatus 100 may extract similar groups (similar groups) from one group. For example, the information processing apparatus 100 may extract similar tissues (similar tissues) from one tissue. For example, the information processing apparatus 100 may extract similar companies (similar companies) for one company.

When the target is a group as described above, for example, the target information may include attribute information such as age and gender of each individual included in the target (group), and images, voices, and behaviors of the face and the whole image. . For example, the target information may include various user attribute information such as demographic attribute information and psychographic attribute information for each individual included in the corresponding target (group). For example, the target information may include information such as age, gender, address, and interests of each individual included in the corresponding target (group). Further, for example, the target information may include image information obtained by imaging each individual included in the corresponding target (group). For example, the target information may include image information obtained by imaging the corresponding individual's face or entire body (whole body). For example, the target information may include audio information such as utterances of each individual included in the corresponding target (group). For example, the target information may include various behavioral information regarding the behavior of each individual included in the corresponding target (group). For example, the target information may include action information such as search, content viewing, and purchase of each individual included in the corresponding target (group). Note that the target information may be only career information (character information) related to work such as the career of each individual included in the target (group).

For example, similar to the graph information GR11, the information processing apparatus 100 may extract similar targets using graph information obtained by structuring vectors generated using the model M2 targeting groups. For example, the information processing apparatus 100 may extract similar targets using graph information in which nodes of each group are connected by edges according to similarity. For example, the information processing apparatus 100 may use graph information GR21 as shown in FIG. 12 to extract similarity information for groups. FIG. 12 is a diagram conceptually showing group graph information.

For example, when the information processing apparatus 100 acquires target information about a group (group Y) that includes a plurality of people as elements, the information processing apparatus 100 displays information on a group (similar target) similar to the group Y as a graph. It may be extracted from the information GR21. For example, the information processing apparatus 100 acquires target information about group Y (target information YDT). For example, the information processing apparatus 100 acquires from the graph information storage unit 123 graph information GR21 for a group including multiple persons as elements. For example, the information processing apparatus 100 acquires graph information GR21 in which a plurality of nodes corresponding to each group including a plurality of persons as elements are connected according to the similarity of the plurality of groups. Then, the information processing apparatus 100 generates a vector YVD from the target information YDT of the group Y using the model M2 stored in the model information storage unit 124 .

Then, the information processing apparatus 100 determines the origin vector to be the node NGP5 (corresponding to the group GP5 in FIG. 12) using the generated vector YVD and the index information. Then, the information processing apparatus 100 extracts similar objects (similar groups) of the group Y by searching the graph information GR21 with the node NGP5, which is the starting vector, as the starting point. For example, the information processing apparatus 100 extracts a predetermined number (for example, two) of nodes as similar objects by searching the graph information GR21 starting from the node NGP5. In the example of FIG. 12, the information processing apparatus 100 extracts the group GP1 (node NGP1) and the group GP5 (node NGP5) included in the similar target range SAR11 as similar targets.

Then, the information processing apparatus 100 provides information about the extracted similar objects. For example, the information processing apparatus 100 provides the terminal device 10 used by the user U1 as groups similar to the group Y, such as the group GP1 and the group GP5 corresponding to the nodes extracted as similar targets.

[1-5. Index information]
The index information (index data) shown in the example of FIG. 1 is an example, and the information processing apparatus 100 may search graph information using various types of index information. Further, for example, the information processing apparatus 100 may generate index data used for searching. For example, the information processing apparatus 100 generates a search index for searching high-dimensional vectors as index data. The high-dimensional vector referred to here may be, for example, a vector with several hundred to several thousand dimensions, or a vector with more dimensions.

For example, the information processing apparatus 100 may generate a search index related to a tree structure (tree structure) as shown in FIG. 1 as index data. For example, the information processing apparatus 100 may generate a search index for a kd-tree (k-dimensional tree) as index data. For example, the information processing apparatus 100 may generate a search index for a VP tree (Vantage-Point tree) as index data.

Further, for example, the information processing apparatus 100 may generate index data having another tree structure. For example, the information processing apparatus 100 may generate various types of index data in which leaves of tree-structured index data are connected to graph data. For example, the information processing apparatus 100 may generate various types of index data in which leaves of the tree-structured index data correspond to nodes in the graph data. When performing a search using such index data, the information processing apparatus 100 may search for graph data from a leaf (node) reached by tracing the index data.

Note that the index data as described above is merely an example, and the information processing apparatus 100 may generate index data having any data structure as long as the query in the graph data can be identified at high speed. . For example, the information processing apparatus 100 generates index data by appropriately using various conventional techniques such as a technique related to binary space division, as long as it is possible to identify a node in graph information corresponding to a query at high speed. may For example, the information processing apparatus 100 may generate index data having any data structure as long as the index is compatible with high-dimensional vector searches. For example, the information processing apparatus 100 may use information about a graph-type search index as described in Non-Patent Document 1 as index information. The information processing apparatus 100 can use the index data and the graph data as described above to enable a more efficient search for a predetermined target.

[1-6. Usage]
Note that the information processing apparatus 100 may extract similar objects for various purposes. For example, as shown in FIG. 1, when forming a team, similar objects that are similar to individuals who have made important contributions to other teams that have achieved high performance (high results) may be extracted. Then, the information processing apparatus 100 may use the similar target information extracted according to the purpose to provide a service corresponding to the purpose. For example, the information processing apparatus 100 may extract similar targets for the purpose of so-called headhunting, in which personnel are scouted from the outside. For example, similar objects may be extracted for the purpose of matching such as matchmaking or matchmaking.

Further, the information processing apparatus 100 may make a determination about one target to be a query based on the extracted information. For example, the information processing apparatus 100 selects one team (group) as a target, and if the extracted similar groups include a team (group) that exhibits high performance, the one team is a team that is highly likely to succeed. It may be determined that there is For example, the information processing apparatus 100 targets one company (group), and if the extracted similar group includes a rapidly growing company (group), the information processing apparatus 100 determines that the one company is a company with a high probability of success. You can judge. Using such information, the information processing apparatus 100 may select startup members and make investment decisions.

Further, the information processing apparatus 100 may extract similar similar targets using graph information generated for each organization such as each company. For example, the information processing apparatus 100 may extract similar individuals similar to individual ZZZ of B company from individuals in A company using graph information of A company. Then, the information processing apparatus 100 may use the extracted information to conduct headhunting from the A company to the B company.

Further, the information processing apparatus 100 may extract a similar target using information of the user himself/herself who receives information as target information. In this case, the information processing apparatus 100 can provide the user with information indicating what kind of person the user resembles. For example, the information processing apparatus 100 can provide the user with information indicating which person the user resembles among successful persons in the past. For example, the information processing apparatus 100 can provide the user with information indicating a person in a situation similar to that of the user. Also, in this case, the information processing apparatus 100 may provide the user with advice from a person who is in a situation similar to that of the user.

Further, the information processing apparatus 100 may extract similar objects for the purpose of diagnosing a sick person. The information processing apparatus 100 may determine the individual serving as the query based on the extracted information. For example, the information processing apparatus 100 targets one individual, and if the extracted similar individuals include an individual with a predetermined disease (referred to as “disease DSX”), the information processing apparatus 100 determines the possibility that the one individual will have the disease DSX. can be determined to be high.

[1-6-1. Target information according to use]
In the example of FIG. 1, the case where the target information includes attribute information of the target (individual) and various information such as images, voices, and actions has been described as an example. You may use the target information according to the purpose of performing. That is, information assumed to be suitable for the purpose of extracting similar targets by the information processing apparatus 100 may be used as the target information.

For example, if the purpose of extracting similar targets is to extract similar targets related to business (economic activity) such as work, the information processing apparatus 100 may use information related to the target business as target information.

For example, when the information processing apparatus 100 extracts similar targets for the purpose of headhunting, the target information may be biographical information including job duties, work experience, past achievements, and the like of each individual. In this case, when one piece of history information indicating a history desired by the user is acquired as a query, the information processing apparatus 100 uses graph information in which each node is connected based on the similarity of the history information to find a similar target. Extract.

Further, for example, if the purpose of extracting a similar target is to extract a similar target related to matching, the information processing apparatus 100 may use information (image information) relating to the appearance of the target and profile information as target information.

For example, when the information processing apparatus 100 extracts similar targets for the purpose of matching, the target information may be personal information such as image information and profile information of each individual. In this case, when one piece of personal information indicating a partner desired by the user is acquired as a query, the information processing apparatus 100 identifies a similar target using graph information in which each node is connected based on the similarity of the personal information. Extract.

Further, for example, if the purpose of extracting a similar target is to extract a similar target related to diagnosis or the like, the information processing apparatus 100 may use medical chart information of the target as target information.

For example, when the information processing apparatus 100 extracts similar objects for the purpose of diagnosing disease DSX, the object information may be medical record information of each individual. In this case, when medical record information of an individual whose diagnosis is desired is acquired as a query, the information processing apparatus 100 extracts a similar target using graph information in which each node is connected based on the similarity of the medical record information. . Then, when an individual having DSX is included in the extracted similar individuals, the information processing apparatus 100 may determine that one individual is highly likely to have DSX.

[1-6-2. Extraction of dissimilar objects]
In the example of FIG. 1, the graph information GR11 in which nodes with high similarity are connected by edges is used. may be The information processing apparatus 100 may use graph information (referred to as a “dissimilar connection graph”) in which nodes with low similarity among a plurality of objects are connected by edges.

For example, the information processing apparatus 100 may extract a target that is dissimilar to one target using a dissimilar connection graph. A target that is dissimilar to one of the plurality of targets in the dissimilar connection graph may be extracted as a corresponding target. In this case, the information processing apparatus 100 extracts a corresponding target (non-similar target) having a correspondence relationship of low similarity with one target. The information processing apparatus 100 can extract a node with low similarity to a query (one target) by extracting a node to which an edge is connected from a starting node in a dissimilar connection graph. That is, the information processing apparatus 100 can search for a user with a low similarity relationship (long distance) in the graph information. For example, the information processing apparatus 100 can search for a person far from one user by inverting the distance function of the graph structure. For example, the information processing apparatus 100 can search for a person far from one user by using the inverse function of the distance function of the graph structure.

For example, the information processing apparatus 100 may perform predetermined marketing to individuals (users) who are dissimilar to one individual (user) extracted using a dissimilar connection graph. For example, the information processing apparatus 100 provides predetermined information to a user (dissimilar user) who is dissimilar to one user who has provided predetermined information to be posted, such as word of mouth, to the dissimilar user. Accordingly, it is possible to increase the possibility that a wide range of users, such as one user and dissimilar users, will post. Predetermined marketing may be carried out. In this way, the information processing apparatus 100 can spread predetermined information to a wide range of users by providing predetermined information to a wide range of users in the graph information. For example, the information processing apparatus 100 can perform buzz marketing to a wide range of users by providing predetermined information to a wide range of users in the vector space of graph information. Thereby, the information processing apparatus 100 can appropriately perform desired marketing such as buzz marketing.

[2. Configuration of information processing system]
As shown in FIG. 2 , the information processing system 1 includes a terminal device 10 , an information providing device 50 and an information processing device 100 . The terminal device 10, the information providing device 50, and the information processing device 100 are connected via a predetermined network N so as to be communicable by wire or wirelessly. FIG. 2 is a diagram illustrating a configuration example of an information processing system according to the embodiment; The information processing system 1 shown in FIG. 2 may include multiple terminal devices 10 , multiple information providing devices 50 , and multiple information processing devices 100 .

The terminal device 10 is an information processing device used by a user. The terminal device 10 accepts various operations by the user. In addition, below, the terminal device 10 may be described as a user. That is, hereinafter, the user can also be read as the terminal device 10 . The terminal device 10 described above is implemented by, for example, a smart phone, a tablet terminal, a notebook PC (Personal Computer), a desktop PC, a mobile phone, a PDA (Personal Digital Assistant), or the like. For example, the terminal device 10 may be an information processing device in which a predetermined server system is used by an administrator. For example, the terminal device 10 collects target information and transmits the target information to the information processing device 100 .

The information processing apparatus 100 is an information processing apparatus that extracts a similar target that is similar to one target from among a plurality of targets by searching graph information using an origin vector as a starting point. For example, the information processing apparatus 100 acquires target information about one target, and determines a starting vector based on the target information and information about a starting vector that is a starting point for searching graph information.

The information processing device 100 is an information processing device that stores information for providing various information to users and the like. For example, when the information processing device 100 acquires target information of one target (hereinafter also referred to as “query information” or “query”) from the terminal device 10, the information processing device 100 searches for a target (vector information etc.) similar to the query, A search result is provided to the terminal device 10 . In the example of FIG. 1, when the information processing apparatus 100 acquires the target information of one target from the terminal device 10, the information processing device 100 searches for a target similar to the one target, and provides the search result to the terminal device 10 as a similar target. . Further, for example, the data provided by the information processing device 100 to the terminal device 10 may be a target name or target information itself, or may be a URL (Uniform Resource Locator) for referring to corresponding data. It may be information. For example, when the data provided by the information processing device 100 to the terminal device 10 is a target name, it may be an individual name such as a name or a group name such as a company name.

The information providing device 50 is an information processing device that stores information for providing various information to the information processing device 100 . For example, the information providing device 50 may store target information and the like collected from various external devices such as a web server. For example, the information providing device 50 is an information processing device that provides various information such as graph information, index information, and models to the information processing device 100 .

[3. Configuration of Information Processing Device]
Next, the configuration of the information processing apparatus 100 according to the embodiment will be described using FIG. FIG. 3 is a diagram illustrating a configuration example of an information processing apparatus according to the embodiment; As shown in FIG. 3, the information processing apparatus 100 has a communication section 110, a storage section 120, and a control section . The information processing apparatus 100 includes an input unit (for example, a keyboard, a mouse, etc.) that receives various operations from an administrator of the information processing apparatus 100, and a display unit (for example, a liquid crystal display, etc.) for displaying various information. may have.

(Communication unit 110)
The communication unit 110 is realized by, for example, a NIC (Network Interface Card) or the like. The communication unit 110 is connected to a network (for example, network N in FIG. 2) by wire or wirelessly, and transmits and receives information to and from the terminal device 10 and the information providing device 50 .

(storage unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. The storage unit 120 according to the embodiment includes a target information storage unit 121, an index information storage unit 122, a graph information storage unit 123, and a model information storage unit 124, as shown in FIG.

(Target information storage unit 121)
The target information storage unit 121 according to the embodiment stores various types of information about targets (objects). For example, the target information storage unit 121 stores target IDs and vector data. In the example of FIG. 4, target information (personal information) relating to the target (individual) is stored. 4 is a diagram illustrating an example of a target information storage unit according to the embodiment; FIG. The target information storage unit 121 shown in FIG. 4 includes items such as "target ID", "target", "target information", and "vector information".

"Target ID" indicates identification information for identifying a target (object). "Target" indicates the specific name, content, etc. of the target (individual) identified by the target ID. In the example of FIG. 4, the target is indicated by an abstract code such as "individual #1". good.

"Target information" indicates information about a target (individual) identified by the target ID. In the example of FIG. 4, the target information is indicated by an abstract code such as "TDT1", but each target information is attribute information such as age and gender of the target (individual) identified by the target ID It may also include attribute information, such as face and whole image, voice, behavior, and other information related to various objects (individuals). "Vector information" indicates vector information corresponding to a target (object) identified by the target ID. That is, in the example of FIG. 4, vector data (vector information) corresponding to an object is registered in association with an object ID that identifies the object.

For example, in the example of FIG. 4, the target (object) identified by the target ID "TG1" is associated with multidimensional (N-dimensional) vector information of "10, 24, 54, 2...". indicates For example, for individual #1, multidimensional (N-dimensional) vector information of “10, 24, 54, 2 . indicates that

It should be noted that the target information storage unit 121 may store various types of information, not limited to the above, depending on the purpose.

(Index information storage unit 122)
The index information storage unit 122 according to the embodiment stores various information regarding indexes. FIG. 5 is a diagram illustrating an example of an index information storage unit according to the embodiment; Specifically, in the example of FIG. 5, the index information storage unit 122 indicates tree-structured index information. In the example of FIG. 5, the index information storage unit 122 includes items such as "root layer", "first layer", "second layer", and "third layer". It should be noted that not only "first layer" to "third layer" but "fourth layer", "fifth layer", "sixth layer", etc. may be included according to the number of layers of the index.

The “root hierarchy” indicates the root (top) hierarchy that is the starting point for determining the origin node using the index. "First layer" stores information for identifying (specifying) a node (node or vector in graph information) belonging to the first layer of the index. A node stored in the "first layer" is a node corresponding to a layer directly connected to the root of the index.

"Second layer" stores information that identifies (specifies) a node (node or vector in graph information) belonging to the second layer of the index. A node stored in the "second layer" is a node corresponding to the immediately lower layer connected to a node in the first layer. "Third layer" stores information for identifying (specifying) a node (node or vector in graph information) belonging to the third layer of the index. A node stored in the "third layer" is a node corresponding to the immediately lower layer linked to a node in the second layer.

For example, in the example shown in FIG. 5, the index information storage unit 122 stores information corresponding to the index information IND11 in FIG. For example, the index information storage unit 122 indicates that the nodes of the first layer are the nodes VT1 to VT3. Numerical values in parentheses under each node indicate vector values corresponding to each node.

Also, for example, the index information storage unit 122 indicates that the nodes of the second layer immediately below the node VT2 are the nodes VT2-1 to VT2-4. Also, for example, the index information storage unit 122 indicates that the nodes in the third layer immediately below the node VT2-2 are the nodes (vectors) in the graph information GR11 of the nodes N35, N451, and N693.

It should be noted that the index information storage unit 122 may store various types of information, not limited to the above, depending on the purpose.

(Graph information storage unit 123)
The graph information storage unit 123 according to the embodiment stores various types of information regarding graph information. 6 is a diagram illustrating an example of a graph information storage unit according to the embodiment; FIG. In the example of FIG. 6, the graph information storage unit 123 has items such as "node ID", "target ID", and "edge information". "Edge information" includes information such as "edge ID" and "reference destination".

“Node ID” indicates identification information for identifying each node (object) in graph data. "Target ID" indicates identification information for identifying a target (object).

"Edge information" indicates information about edges connected to the corresponding node. In the example of FIG. 6, "edge information" indicates the case where the edge is a directed edge, and indicates information regarding the output edge output from the corresponding node. "Edge ID" indicates identification information for identifying an edge connecting nodes. "Reference destination" indicates information indicating a reference destination (node) connected by an edge. That is, in the example of FIG. 6, for a node ID that identifies a node, a reference destination ( node) are associated and registered.

For example, the example of FIG. 6 indicates that the node (vector) identified by the node ID "N1" corresponds to the target (object) identified by the target ID "TG1". It also indicates that the edge identified by the edge ID "E11" is connected to the node (vector) identified by the node ID "N25" from the node identified by the node ID "N1". That is, the example of FIG. 6 indicates that the node (vector) identified by the node ID "N1" can be traced to the node (vector) identified by the node ID "N25".

It should be noted that the graph information storage unit 123 may store various types of information, not limited to the above, depending on the purpose. For example, the graph information storage unit 123 may store lengths of edges connecting nodes (vectors). That is, the graph information storage unit 123 may store information indicating distances between nodes (vectors).

(Model information storage unit 124)
The model information storage unit 124 according to the embodiment stores information about models. For example, the model information storage unit 124 stores model information (model data) generated by the generation process. 7 is a diagram illustrating an example of a model information storage unit according to the embodiment; FIG. The model information storage unit 124 shown in FIG. 7 includes items such as "model ID", "usage", and "model data". Although only the models M1 and M2 are illustrated in FIG. 7, a large number of model information such as M21 and M22 may be stored according to each application (prediction target).

"Model ID" indicates identification information for identifying a model. For example, the model identified by model ID "M1" corresponds to model M1 shown in the example of FIG. "Use" indicates the use of the corresponding model. Also, "model data" indicates the data of the associated corresponding model. For example, "model data" includes information including nodes in each layer, functions adopted by each node, connection relations of nodes, and connection coefficients set for connections between nodes.

For example, in the example shown in FIG. 7, the model (model M1) identified by the model ID "M1" is used for "feature extraction (individual)" and is used to extract features from input target information. indicates that It also indicates that the model data of the model M1 is the model data MDT1.

The model M1 (model data MDT1) includes an input layer in which target information (personal information) of a target (individual) is input, an output layer, and any layer from the input layer to the output layer, except for the output layer. Each layer other than the output layer includes a first element belonging to the layer and a second element whose value is calculated based on the weight of the first element and the first element, and for the target information input to the input layer with each element belonging to the first element as the first element, and by performing an operation based on the first element and the weight of the first element, the computer functions to output the same information as the information input to the input layer from the output layer It is a model for

It is also assumed that the models M1, M2, etc. are implemented by a neural network having one or more intermediate layers, such as a DNN (Deep Neural Network). In this case, for example, the first element included in the models M1 and M2 corresponds to any node of the input layer or the intermediate layer. Also, the second element corresponds to the next node, which is a node to which the value is transmitted from the node corresponding to the first element. Also, the weight of the first element corresponds to the connection coefficient, which is the weight considered for the value transmitted from the node corresponding to the first element to the node corresponding to the second element.

Here, it is assumed that the models M21, M22, etc. are realized by a regression model represented by "y=a1*x1+a2*x2+...+ai*xi". In this case, for example, the first elements included in the models M21 and M22 correspond to input data (xi) such as x1 and x2. Also, the weight of the first element corresponds to the coefficient ai corresponding to xi. Here, the regression model can be viewed as a simple perceptron with an input layer and an output layer. When each model is regarded as a simple perceptron, the first element can be regarded as a node of the input layer, and the second element can be regarded as a node of the output layer.

It should be noted that the model information storage unit 124 may store various types of model information, not limited to the above, depending on the purpose.

(control unit 130)
Returning to the description of FIG. 3, the control unit 130 is a controller, and is stored in a storage device inside the information processing apparatus 100 by, for example, a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). Various programs (corresponding to an example of an information processing program) are executed by using the RAM as a work area. Also, the control unit 130 is a controller, and is implemented by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). The control unit 130 performs information processing in accordance with the models M1, M2, etc. stored in the model information storage unit 124 to create an input layer to which target information of a target is input, an output layer, and layers from the input layer to the output layer. A first element belonging to any layer other than the output layer and a second element whose value is calculated based on the first element and the weight of the first element, and is input to the input layer The same information as the information input to the input layer is obtained by performing an operation based on the first element and the weight of the first element, with each element belonging to each layer other than the output layer as the first element. is output from the output layer.

As shown in FIG. 3, the control unit 130 includes an acquisition unit 131, a generation unit 132, a determination unit 133, a search unit 134, and a provision unit 135, and has information processing functions and actions described below. realize or perform Note that 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 performs information processing described later.

(Acquisition unit 131)
Acquisition unit 131 acquires various types of information. For example, the acquisition unit 131 acquires various information from the storage unit 120 . For example, the acquisition unit 131 acquires various information from the target information storage unit 121, the index information storage unit 122, the graph information storage unit 123, the model information storage unit 124, and the like. The acquisition unit 131 also acquires various types of information from an external information processing device. The acquisition unit 131 acquires various types of information from the terminal device 10 and the information providing device 50 .

The acquiring unit 131 acquires graph information in which a plurality of nodes corresponding to each of a plurality of targets having people as elements are connected according to the similarity of the plurality of targets, and target information about one target. The acquiring unit 131 acquires graph information in which a plurality of vectors representing characteristics of each of a plurality of targets are connected according to similarity. The acquiring unit 131 acquires graph information in which a plurality of vectors corresponding to each of a plurality of targets are connected according to similarity, and target information regarding one target. The acquisition unit 131 acquires graph information in which a plurality of vectors, each element of which is a feature amount extracted from each of a plurality of targets using a predetermined model, are connected according to similarity. By inputting information about a plurality of targets into a predetermined model, the acquiring unit 131 obtains graph information in which a plurality of vectors whose elements are the feature amounts of each of the extracted targets are connected according to similarity. to get

The acquisition unit 131 acquires target information about one target from the terminal device 10 used by the user. The acquiring unit 131 acquires graph information in which a plurality of nodes corresponding to each of a plurality of subjects each of which is an individual are connected, and target information regarding one subject which is one individual. The acquisition unit 131 acquires graph information in which a plurality of nodes corresponding to each of a plurality of targets, each of which is a group including a plurality of people, are connected, and target information regarding one target, which is one group.

For example, the acquisition unit 131 acquires a plurality of nodes (vectors) to be searched for data. For example, the acquiring unit 131 acquires a directed edge group including a plurality of nodes and a plurality of directed edges connecting each of the plurality of nodes.

For example, the acquisition unit 131 acquires graph information (graph data) from an external information processing device. For example, the acquisition unit 131 acquires graph information from the graph information storage unit 123 . For example, the acquisition unit 131 acquires graph information. In the example of FIG. 1, the acquisition unit 131 acquires graph information GR11.

For example, the acquisition unit 131 acquires index information (index data) from an external information processing device. For example, the acquisition unit 131 acquires index information from the index information storage unit 122 . For example, the acquisition unit 131 acquires tree-structured index information. In the example of FIG. 1, the acquisition unit 131 acquires index information IND11.

Also, the acquisition unit 131 acquires target information about one target from the terminal device 10 used by the user. For example, the acquisition unit 131 acquires target information of one target as a search query. For example, the acquisition unit 131 acquires a search query regarding an object (individual). The acquisition unit 131 acquires the target information TDT11 regarding the person X from the terminal device 10 used by the user U1.

For example, the acquisition unit 131 acquires target information (personal information) regarding one target. For example, the acquisition unit 131 acquires the target information TDT11 regarding the person X from the terminal device 10 . For example, the acquisition unit 131 acquires the graph information GR11 about the object from the graph information storage unit 123. FIG. Further, for example, the acquiring unit 131 acquires index information IND11 used for determining a node (originating vector) serving as a starting point for searching in the graph information GR11 from the index information storage unit 122 (see FIG. 5).

(Generating unit 132)
The generator 132 generates various types of information. For example, the generation unit 132 generates a model shown in the model information storage unit 124 using learning data (target information) stored in the target information storage unit 121 . For example, the generation unit 132 generates a model (autoencoder) that outputs information similar to the input target information based on the learning data acquired by the acquisition unit 131 . For example, the generation unit 132 generates a model (autoencoder) that outputs information similar to the input target information, using the input target information itself as correct information.

For example, the generation unit 132 generates the model M1 and the like, and stores the generated model M1 and the like in the model information storage unit 124 . Note that the generation unit 132 may generate the model M1 using any learning algorithm. For example, the generator 132 generates the model M1 using a learning algorithm such as a neural network. As an example, when the generation unit 132 generates the model M1 or the like using a neural network, the model M1 or the like includes an input layer including one or more neurons, an intermediate layer including one or more neurons, and one or more neurons. and an output layer containing

The generation unit 132 generates an input layer to which target information of a target is input, an output layer, a first element belonging to any layer from the input layer to the output layer other than the output layer, and a first element and a second element whose value is calculated based on the weight of the first element. By performing calculations based on the one element and the weight of the first element, a model is generated in which the same information as that input to the input layer is output from the output layer.

For example, the generator 132 generates a model based on learning data. For example, the generator 132 generates a model based on learning data. For example, the generation unit 132 generates a model by performing learning using the target information TDT1, TDT2, etc. in the target information storage unit 121 as learning data (teacher data).

For example, when the target information TDT1 is input, the generation unit 132 performs learning processing so that the model M1 outputs information similar to the target information TDT1. For example, the generation unit 132 performs learning processing so that the model M1 outputs information similar to the target information TDT2 when the target information TDT2 is input.

Note that the generation unit 132 is not limited to the model M1 as an autoencoder, and uses various learning algorithms to generate a model M21 corresponding to the target "individual: business suitability" and a model M22 corresponding to the target "group: business success". may be generated. For example, the generation unit 132 generates the models M21, M22, etc. using learning algorithms such as neural networks, support vector machines (SVM), clustering, and reinforcement learning. As an example, when the generation unit 132 generates the models M21, M22, etc. using a neural network, the models M21, M22, etc. include an input layer including one or more neurons, an intermediate layer including one or more neurons, and a and an output layer containing the above neurons.

The generation unit 132 generates a model and stores the generated model in the model information storage unit 124 . Specifically, the generation unit 132 generates a first element belonging to an input layer to which the target information of the target is input, an output layer, and any layer from the input layer to the output layer other than the output layer. and a second element whose value is calculated based on the first element and the weight of the first element. By performing calculations based on the first element and the weight of the first element as one element, a model is generated in which the output layer outputs a score value used for predicting whether a target is suitable for a predetermined purpose. For example, the generation unit 132 includes an input layer to which target information of a certain target is input, an output layer, a first element belonging to any layer from the input layer to the output layer other than the output layer, and a second element whose value is calculated based on the first element and the weight of the first element. , the calculation based on the first element and the weight of the first element generates a model M21 that outputs from the output layer the value of the score used to predict whether the target is suitable for a given purpose. do. For example, the generating unit 132 generates the model M21 so as to output a score of "1" when personal target information suitable for sales is input. For example, the generating unit 132 generates the model M21 so as to output a score of "0" when personal target information unsuitable for sales is input. Note that the information processing apparatus 100 may not have the generation unit 132 when acquiring a model from another external apparatus such as the information providing apparatus 50 .

In the example of FIG. 1, the generation unit 132 generates an N-dimensional vector from the target information of each target using a model for extracting features of the target information. The generation unit 132 generates a vector from the target information (personal information) of each target (individual) using the model M1 shown in the model information storage unit 124 (FIG. 7).

The generation unit 132 also generates a vector used for searching graph information from target information corresponding to one target. In the example of FIG. 1, the generation unit 132 generates a vector corresponding to the individual X by performing the processing shown in the processing group PS11. The generation unit 132 inputs the target information TDT11 regarding the individual X to the model M1. Specifically, the generation unit 132 inputs target information TDT11 including attribute information such as the age and gender of the target (individual) of the individual X, and information such as an image of the face and the whole image, voice, behavior, etc. to the model M1. do. Then, the generation unit 132 generates a vector using information in the model M1 after the input of the target information TDT11. For example, the generator 132 generates vector data using each element in the model M1 to which the target information TDT11 is input.

In the example of FIG. 1, the generator 132 generates the vector VD11 using the values of the elements in the model M1 to which the target information TDT11 is input. For example, the generating unit 132 uses the value VE1 (see FIG. 10) corresponding to the neuron NL1 and the value VE2 (see FIG. 10) corresponding to the neuron NL2 of the model M1 when the target information TDT11 of the person X is input. to generate a vector. For example, when the target information TDT11 of the individual X is input, the generation unit 132 may extract the calculated value VE1 corresponding to the neuron NL1 as the first-dimensional element of the vector VD11. Further, for example, when the target information of the target is input, the generating unit 132 generates the vector VD11 by using the calculated value VE2 corresponding to the neuron NL2 as the second-dimensional element of the vector VD11. In the example of FIG. 1, the generator 132 generates a vector VD11 whose first-dimensional element is "35" and whose second-dimensional element is "63."

For example, the generation unit 132 may generate an N-dimensional vector from the target information of each target using the model M21. For example, the generation unit 132 inputs target information corresponding to the target to the model M21 and generates a vector from the target information of each target. For example, the generation unit 132 inputs target information of a target to the model M21, calculates the value of each element (neuron) in the model M21, and outputs a score corresponding to the input target information. For example, the generation unit 132 may extract the value of each element (neuron) in the intermediate layer of the model M21 as a feature amount, and generate N-dimensional vector data corresponding to each object.

(Determination unit 133)
The determination unit 133 determines various types of information. The determination unit 133 determines the origin node based on the index information used for determining the origin node. For example, the determination unit 133 determines the starting vector based on the target information acquired by the acquiring unit 131 and the information about the starting vector that is the starting point for searching the graph information. For example, the determination unit 133 determines the starting vector based on the target information and the information that serves as a reference for determining the starting vector for searching the graph information. Further, the determination unit 133 determines the origin vector based on the index information used for determining the origin vector. Further, the determination unit 133 determines the origin vector based on the tree-structured index information.

In the example of FIG. 1, the determination unit 133 uses the index information IND11 to determine (specify) the origin vector corresponding to one target (query). For example, the determination unit 133 uses the index information IND11 to determine (specify) the origin vector corresponding to the person X's vector VD11. The determining unit 133 determines the starting vector in the graph information GR11 using the vector VD11 and the index information IND11.

For example, the determining unit 133 determines the origin vector in the graph information GR11 using tree-structured index information as indicated by the index information IND11 in FIG. In the example of FIG. 1, after generating the vector VD11, the determination unit 133 traces the index information IND11 from top to bottom to identify the origin vector that is a candidate for the neighborhood of the index information IND11, thereby efficiently searching. Determine the origin vector corresponding to the query (one object).

For example, the determining unit 133 determines the origin vector corresponding to the vector VD11 by tracing the index information IND11 from the root RT to leaf nodes (nodes (vectors) in the graph information GR11). In the example of FIG. 1, the determining unit 133 determines the node N451 as the origin vector by tracing the index information IND11 from the root RT to the node N451.

(Search unit 134)
The search unit 134 searches for various types of information. For example, the search unit 134 searches various information using the information acquired by the acquisition unit 131 . For example, the search unit 134 extracts various information. For example, the search unit 134 uses information acquired by the acquisition unit 131 to extract various types of information. For example, the search unit 134 performs various determination processes. For example, the search unit 134 uses information acquired by the acquisition unit 131 to determine various types of information. For example, the search unit 134 searches various information stored in the storage unit 120 . For example, the acquisition unit 131 searches various types of information stored in the target information storage unit 121, the index information storage unit 122, the graph information storage unit 123, the model information storage unit 124, and the like. For example, the search unit 134 performs calculation processing using various information. For example, the search unit 134 uses information about vectors to calculate the distance between the vectors. For example, the search unit 134 may use information on the length (distance) of edges connecting nodes (vectors) stored in the graph information storage unit 123, or may use vector information on each node ( It is also possible to calculate information on the length (distance) of the edge connecting the vectors) and use the information on the calculated length (distance).

The retrieving unit 134 retrieves graph information starting from a starting node, which is determined based on a predetermined criterion, among the plurality of nodes of the graph information acquired by the acquiring unit 131. extracts a corresponding target that has a predetermined correspondence relationship with one of the plurality of targets. The search unit 134 extracts corresponding objects starting from the starting node determined by the determining unit 133 . The search unit 134 extracts targets similar to one target from among the plurality of targets as corresponding targets.

The retrieving unit 134 retrieves the graph information from among the plurality of vectors of the graph information acquired by the acquiring unit 131, starting from the starting vector that is the starting point for searching the graph information determined based on a predetermined criterion. , to extract a similar object (similar individual) that is similar to one object from among a plurality of objects. The search unit 134 extracts various information. For example, the retrieval unit 134 retrieves graph information using the starting vector determined by the determination unit 133 as a starting point, thereby extracting a similar target that is similar to one target from among the plurality of targets.

In the example of FIG. 1, the search unit 134 searches for objects (individuals) similar to person X. In FIG. For example, the search unit 134 extracts a similar target of the individual X by searching the graph information GR11. The search unit 134 extracts nodes located near the node N451 as similar objects. For example, the search unit 134 extracts a node that is close to the node N451 as a similar target. For example, the search unit 134 extracts nodes reachable from the node N451 as similar objects by tracing edges starting from the node N451. For example, the search unit 134 extracts a predetermined number (for example, 2 or 10) of nodes as similar targets. For example, the search unit 134 extracts a similar target of the individual X through search processing as shown in FIG. In the example of FIG. 1, the search unit 134 extracts the node N451 and the node N35 as similar objects by searching the graph information GR11 with the node N451 as the starting point.

The search unit 134 extracts a target dissimilar to one target from among the plurality of targets as a corresponding target. For example, the search unit 134 may extract a target that is dissimilar to one target using a dissimilar connection graph. A target that is dissimilar to one of the plurality of targets in the dissimilar connection graph may be extracted as a corresponding target. In this case, the search unit 134 extracts a corresponding target (non-similar target) having a correspondence relationship of low similarity with one target.

(Providing unit 135)
The providing unit 135 provides various information. For example, the providing unit 135 transmits various information to the terminal device 10 and the information providing device 50 . For example, the providing unit 135 distributes various types of information to the terminal device 10 and the information providing device 50 . For example, the providing unit 135 provides various information to the terminal device 10 and the information providing device 50 . The providing unit 135 provides a predetermined service based on the correspondence target extracted by the searching unit 134 . The providing unit 135 provides an information providing service regarding the correspondence target. The providing unit 135 provides the terminal device 10 with information on the correspondence target. The providing unit 135 provides a predetermined service based on the similar objects extracted by the searching unit 134 . In addition, the providing unit 135 provides an information providing service regarding similar objects. The providing unit 135 provides the terminal device 10 with information on the similar target.

For example, the providing unit 135 provides object IDs corresponding to queries as search results. For example, the providing unit 135 provides the information providing device 50 with the object ID selected by the searching unit 134 . The providing unit 135 provides the information providing device 50 with the object ID selected by the searching unit 134 as information indicating the vector corresponding to the query. Further, the providing unit 135 may provide the model generated by the generating unit 132 to an external information processing device.

In the example of FIG. 1 , the providing unit 135 provides information about similar targets extracted by the searching unit 134 . For example, the providing unit 135 provides individual #451 corresponding to the node N451 and individual #35 corresponding to the node N35 as objects (individuals) similar to the individual X to the terminal device 10 used by the user U1.

[4. Information processing flow]
Next, the procedure of information processing by the information processing system 1 according to the embodiment will be described with reference to FIG. FIG. 8 is a flowchart illustrating an example of information processing according to the embodiment.

As shown in FIG. 8, the information processing apparatus 100 obtains target information about one target whose element is a person (step S101). In the example of FIG. 1 , the information processing device 100 acquires target information TDT11 regarding the person X from the terminal device 10 .

The information processing apparatus 100 acquires graph information about an object whose element is a person (step S102). For example, the information processing apparatus 100 acquires the graph information GR11 regarding the target (individual) from the graph information storage unit 123 .

Then, the information processing apparatus 100 uses the model to generate a vector from the target information regarding one target (step S103). In the example of FIG. 1, the information processing apparatus 100 uses the model M1 stored in the model information storage unit 124 to generate the vector VD11 from the target information TDT11.

Then, the information processing apparatus 100 determines the origin vector using the generated vector and the index information (step S104). In the example of FIG. 1, the information processing apparatus 100 uses the vector VD11 and the index information IND11 stored in the index information storage unit 122 to determine the origin vector to be the node N451.

Then, the information processing apparatus 100 extracts a corresponding target that has a predetermined correspondence relationship with the one target by searching the graph information (step S105). For example, the information processing apparatus 100 extracts targets similar to one target as corresponding targets by searching graph information. In the example of FIG. 1, the information processing apparatus 100 extracts the node N451 and the node N35 as similar objects by searching the graph information GR11 with the node N451 as a starting point.

Then, the information processing apparatus 100 provides information on the extracted similar target (step S106). In the example of FIG. 1, the information processing apparatus 100 treats the individual #451 corresponding to the node N451 and the individual #35 corresponding to the node N35 as targets (similar individuals) similar to the individual X. provide to

[5. Generation process flow]
Next, a procedure of generation processing by the information processing system 1 according to the embodiment will be described with reference to FIG. 9 . FIG. 9 is a flowchart illustrating an example of generation processing according to the embodiment.

As shown in FIG. 9, the information processing apparatus 100 acquires learning data (step S201). For example, the information processing device 100 acquires learning data from the target information storage unit 121 . For example, the information processing apparatus 100 acquires the target information TDT1, TDT2, etc. from the target information storage unit 121 as learning data.

After that, the information processing apparatus 100 generates a model based on the learning data (step S202). For example, the information processing apparatus 100 uses learning data from the target information storage unit 121 to generate the model M1. For example, the information processing apparatus 100 generates the model M1 so that information (target information) similar to information (target information) input to the input layer is output from the output layer. For example, the information processing apparatus 100 generates a model M1 as an autoencoder that receives target information (personal information) of a target (individual) as input.

[6. Search example]
Here, an example of retrieval using the graph information described above will be shown. The search using graph information (graph data) is not limited to the following, and may be performed by various procedures. This point will be described with reference to FIG. 11 as an example. FIG. 11 is a flowchart illustrating an example of search processing using graph data (graph information). The search processing described below is performed by the search unit 134 of the information processing apparatus 100, for example. Also, the object referred to below may be read as a vector or a node. In the following description, the information processing apparatus 100 performs the search process, but the search process may be performed by another apparatus. For example, the information processing apparatus 100 uses vector data generated from target information of one target as a search query. For example, the information processing apparatus 100 searches for graph data starting from an origin vector determined based on vector data generated from object information of one object and index information. In the example of FIG. 1, the information processing apparatus 100 searches for the graph information GR11 starting from the node N451, which is the starting vector determined based on the vector VD11 of the person X and the index information IND11.

Here, the neighborhood object set N(G, y) is the set of neighborhood objects associated by the edge attached to node y. "G" may be predetermined graph data (for example, graph information GR11, etc.). For example, the information processing apparatus 100 executes k-nearest neighbor search processing.

For example, the information processing apparatus 100 sets the radius r of the hypersphere to ∞ (infinity) (step S300), and extracts the subset S from the existing object set (step S301). For example, the information processing apparatus 100 may extract, as the subset S, an object (node) selected as a root node (origin vector). In the example of FIG. 1, the information processing apparatus 100 may extract the node N451, which is the origin vector, etc., as the subset S. In the example of FIG. Also, for example, a hypersphere is a virtual sphere that indicates a search range. The objects included in the object set S extracted in step S301 are also included in the initial set of the object set R of the search results.

Next, the information processing apparatus 100 extracts an object having the shortest distance from the object y, where y is the search query object, among the objects included in the object set S, and sets it as an object s (step S302). In the example of FIG. 1, the information processing apparatus 100 extracts the object with the shortest distance from the vector VD11, which is the search query object, from among the objects included in the object set S, and sets it as the object s. For example, the information processing apparatus 100 extracts the node N451 having the shortest distance from the vector VD11 among the objects included in the object set S, and sets it as the object s. For example, when only the object (node) selected as the root node (originating vector) is an element of the object set S, the information processing apparatus 100 extracts the root node (originating vector) as the object s. . Next, the information processing apparatus 100 excludes the object s from the object set S (step S303).

Next, the information processing apparatus 100 determines whether or not the distance d(s, y) between the object s and the object y exceeds r(1+ε) (step S304). Here, ε is an expansion factor, and r(1+ε) is a value that indicates the radius of the search range (only nodes within this range are searched; making it larger than the search range increases the accuracy). be. When the distance d(s, y) between the object s and the object y exceeds r(1+ε) (step S304: Yes), the information processing device 100 outputs the object set R as a neighborhood object set of the object y (step S305), the process ends.

If the distance d(s, y) between the object s and the search query object y does not exceed r(1+ε) (step S304: No), the information processing apparatus 100 calculates the neighborhood object set N(G, s) of the object s. One object not included in the object set C is selected from the objects that are elements of , and the selected object u is stored in the object set C (step S306). The object set C is provided for convenience in order to avoid duplicate searches, and is set to an empty set at the start of processing.

Next, the information processing apparatus 100 determines whether or not the distance d(u, y) between the object u and the object y is r(1+ε) or less (step S307). If the distance d(u, y) between the object u and the object y is less than or equal to r(1+ε) (step S307: Yes), the information processing apparatus 100 adds the object u to the object set S (step S308). If the distance d(u, y) between the object u and the object y is not less than r(1+ε) (step S307: No), the information processing apparatus 100 performs determination (processing) in step S309.

Next, the information processing apparatus 100 determines whether or not the distance d(u, y) between the object u and the object y is less than or equal to r (step S309). When the distance d(u, y) between the object u and the object y exceeds r (step S309: No), the information processing apparatus 100 performs determination (processing) of step S315. If the distance d(u, y) between the object u and the object y is not equal to or less than r (step S309: No), the information processing apparatus 100 performs determination (processing) in step S315.

If the distance d(u, y) between the object u and the object y is less than or equal to r (step S309: Yes), the information processing apparatus 100 adds the object u to the object set R (step S310). Then, the information processing apparatus 100 determines whether or not the number of objects included in the object set R exceeds ks (step S311). The predetermined number ks is an arbitrarily determined natural number. For example, various settings such as ks=2 and ks=10 may be used. When the number of objects included in the object set R does not exceed ks (step S311: No), the information processing apparatus 100 performs determination (processing) in step S313.

If the number of objects included in the object set R exceeds ks (step S311: Yes), the information processing apparatus 100 selects the object that is the longest (farthest) from the object y among the objects included in the object set R, It is excluded from the object set R (step S312).

Next, the information processing apparatus 100 determines whether or not the number of objects included in the object set R matches ks (step S313). If the number of objects included in the object set R does not match ks (step S313: No), the information processing apparatus 100 performs determination (processing) in step S315. If the number of objects included in the object set R matches ks (step S313: Yes), the information processing apparatus 100 determines that the object y has the longest (farthest) distance among the objects included in the object set R. The distance between the object and object y is set to a new r (step S314).

Then, the information processing apparatus 100 determines whether or not all objects have been selected from objects that are elements of the neighborhood object set N(G, s) of the object s and stored in the object set C (step S315). . If all the objects that are elements of the neighborhood object set N(G, s) of the object s have not been selected and stored in the object set C (step S315: No), the information processing apparatus 100 proceeds to step S306. and repeat the process.

When all objects are selected from the objects that are elements of the neighborhood object set N(G, s) of the object s and stored in the object set C (step S315: Yes), the information processing apparatus 100 returns the object set S is an empty set (step S316). If the object set S is not an empty set (step S316: No), the information processing apparatus 100 returns to step S302 and repeats the process. If the object set S is an empty set (step S316: Yes), the information processing apparatus 100 outputs the object set R and ends the process (step S317). For example, the information processing apparatus 100 may provide objects (nodes) included in the object set R as search results corresponding to the search query (input object y) to the terminal apparatus 10 or the like that performed the search. In the example of FIG. 1, the information processing apparatus 100 provides the node N451 and the node N35 included in the object set R as search results corresponding to the search query (the vector VD11 of the individual X) to the terminal device 10 or the like that performed the search. You may For example, the information processing apparatus 100 provides a person #451 corresponding to the node N451 and a person #35 corresponding to the node N35 as objects (individuals) similar to the person X to the terminal device 10 used by the user U1.

[7. effect〕
As described above, the information processing apparatus 100 according to the embodiment has the acquisition unit 131 and the search unit 134 . The acquiring unit 131 acquires graph information in which a plurality of nodes corresponding to each of a plurality of targets having people as elements are connected according to the similarity of the plurality of targets, and target information about one target. The retrieving unit 134 retrieves graph information starting from a starting node, which is determined based on a predetermined criterion, among the plurality of nodes of the graph information acquired by the acquiring unit 131. extracts a corresponding target that has a predetermined correspondence relationship with one of the plurality of targets.

As described above, the information processing apparatus 100 according to the embodiment searches the graph information with the starting vector as the starting point, and extracts a corresponding target having a predetermined correspondence relationship with one of the plurality of targets. , it is possible to appropriately extract a target having a predetermined correspondence with a desired target.

Further, the information processing apparatus 100 according to the embodiment has a determination unit 133 . The determination unit 133 determines the origin node based on the index information used for determining the origin node. The search unit 134 extracts corresponding objects starting from the starting node determined by the determining unit 133 .

As described above, the information processing apparatus 100 according to the embodiment appropriately extracts a target having a predetermined correspondence with a desired target by determining the starting vector based on the index information used for determining the starting vector. be able to.

Further, in the information processing apparatus 100 according to the embodiment, the determining unit 133 determines the origin vector based on the tree-structured index information.

As described above, the information processing apparatus 100 according to the embodiment can appropriately extract a target having a predetermined correspondence with a desired target by determining the origin vector based on the tree-structured index information. can.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 acquires graph information in which a plurality of vectors representing characteristics of each of a plurality of targets are connected according to similarity.

As described above, the information processing apparatus 100 according to the embodiment obtains graph information in which a plurality of vectors representing characteristics of each of a plurality of targets are connected according to similarity, thereby obtaining a desired target and a predetermined correspondence. Objects with relationships can be appropriately extracted.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 connects a plurality of vectors whose elements are feature amounts extracted from each of a plurality of targets using a predetermined model according to similarity. get graph information.

As described above, the information processing apparatus 100 according to the embodiment provides graph information in which a plurality of vectors whose elements are feature amounts extracted from each of a plurality of targets using a predetermined model are connected according to similarity. By obtaining , it is possible to appropriately extract an object having a predetermined correspondence with a desired object.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 inputs information about a plurality of targets into a predetermined model, and obtains a plurality of vectors whose elements are feature amounts of each of the extracted targets. obtains graph information connected according to similarity.

As described above, the information processing apparatus 100 according to the embodiment inputs information about a plurality of targets into a predetermined model, so that a plurality of vectors whose elements are feature amounts of each of the plurality of targets to be extracted are similar to each other. By acquiring the graph information linked according to the gender, it is possible to appropriately extract the target having a predetermined correspondence with the desired target.

The information processing apparatus 100 according to the embodiment also has a providing unit 135 . The providing unit 135 provides a predetermined service based on the correspondence target extracted by the searching unit 134 .

As described above, the information processing apparatus 100 according to the embodiment provides a predetermined service based on the extracted corresponding target (for example, a similar target), thereby obtaining information about a target having a predetermined correspondence relationship with a desired target. It is possible to appropriately provide the services used.

In addition, in the information processing apparatus 100 according to the embodiment, the providing unit 135 provides an information providing service regarding correspondence targets.

As described above, the information processing apparatus 100 according to the embodiment can appropriately provide a service using information about an object having a predetermined correspondence relationship with a desired object by providing an information providing service about the corresponding object. can.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 acquires target information regarding one target from the terminal device 10 used by the user. The providing unit 135 provides the terminal device 10 with information on the correspondence target.

As described above, the information processing apparatus 100 according to the embodiment acquires target information related to one target from the terminal device 10 used by the user, and provides the terminal device 10 with information related to the target to be handled. It is possible to appropriately provide services using information on similar objects).

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 obtains graph information in which a plurality of nodes each corresponding to each of a plurality of subjects, each of which is an individual, are connected, and Get target information.

As described above, the information processing apparatus 100 according to the embodiment combines graph information in which a plurality of nodes each corresponding to each of a plurality of subjects, each of which is an individual, are connected, and target information regarding one subject, which is one individual. By acquiring the information, it is possible to appropriately extract individuals having a predetermined correspondence with the desired individual.

In addition, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 obtains graph information in which a plurality of nodes corresponding to each of a plurality of targets, each of which is a group including a plurality of people, is connected, and Get object information about one object.

As described above, the information processing apparatus 100 according to the embodiment provides graph information in which a plurality of nodes corresponding to each of a plurality of targets, each of which is a group including a plurality of people, is connected, and one target that is one group. By acquiring the target information regarding, it is possible to appropriately extract a group having a predetermined correspondence with a desired group.

Further, in the information processing apparatus 100 according to the embodiment, the search unit 134 extracts targets similar to one target from among the plurality of targets as corresponding targets.

As described above, the information processing apparatus 100 according to the embodiment extracts a target similar to one target from among a plurality of targets as a corresponding target, thereby appropriately selecting a target similar to a desired target (similar target). can be extracted. In this case, the information processing apparatus 100 can appropriately extract targets (similar targets) that are highly similar to the desired target.

Further, in the information processing apparatus 100 according to the embodiment, the search unit 134 extracts a target dissimilar to one target from among the plurality of targets as a corresponding target.

As described above, the information processing apparatus 100 according to the embodiment extracts a target dissimilar to one target from among a plurality of targets as a corresponding target, thereby extracting a target dissimilar to the desired target (dissimilar target). Can be properly extracted. In this case, the information processing apparatus 100 can appropriately extract targets having low similarity to the desired target (dissimilar targets).

[8. Hardware configuration]
The information processing apparatus 100 according to the embodiments described above is implemented by a computer 1000 configured as shown in FIG. 13, for example. FIG. 13 is a hardware configuration diagram showing an example of a computer that implements the functions of the information processing apparatus. The computer 1000 includes a CPU 1100, a RAM 1200, a ROM (Read Only Memory) 1300, a HDD (Hard Disk Drive) 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 programs stored in the ROM 1300 or HDD 1400 and controls each section. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started up, a program depending on the hardware of the computer 1000, and the like.

The HDD 1400 stores programs executed by the CPU 1100, data used by the programs, and the like. Communication interface 1500 receives data from other devices via network N, sends the data to CPU 1100, and transmits data generated by CPU 1100 to other devices via network N. FIG.

The CPU 1100 controls output devices such as displays and printers, and input devices such as keyboards and mice, through an input/output interface 1600 . CPU 1100 acquires data from an input device via input/output interface 1600 . CPU 1100 also outputs the generated data to an output device via input/output interface 1600 .

Media interface 1700 reads programs or data stored in recording medium 1800 and provides them to CPU 1100 via RAM 1200 . CPU 1100 loads such a program from recording medium 1800 onto RAM 1200 via 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 a PD (Phase change rewritable disc), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. etc.

For example, when the computer 1000 functions as the information processing apparatus 100 according to the embodiment, the CPU 1100 of the computer 1000 executes a program or data (for example, model M1 (model data MDT1)) loaded on the RAM 1200 to It implements the functions of the control unit 130 . CPU 1100 of computer 1000 reads these programs or data (for example, model M1 (model data MDT1)) from recording medium 1800 and executes them. may be obtained.

As described above, some of the embodiments of the present application have been described in detail based on the drawings, but these are merely examples, and various modifications, It is possible to carry out the invention in other forms with modifications.

[9. others〕
Further, among the processes described in the above embodiments, all or part of the processes described as being automatically performed can be manually performed, or the processes described as being performed manually can be performed manually. All or part of this can also be done automatically by known methods. In addition, information including processing procedures, specific names, various data and parameters shown in the above documents and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each drawing is not limited to the illustrated information.

Also, each component of each device illustrated is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution and integration of each device is not limited to the one shown in the figure, and all or part of them can be functionally or physically distributed and integrated in arbitrary units according to various loads and usage conditions. Can be integrated and configured.

Further, each process described in each embodiment described above can be appropriately combined within a range in which the contents of the process are not inconsistent.

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

1 information processing system 100 information processing device 121 target information storage unit 122 index information storage unit 123 graph information storage unit 124 model information storage unit 130 control unit 131 acquisition unit 132 generation unit 133 determination unit 134 search unit 135 provision unit 10 terminal device 50 Information providing device N network

Claims (9)

Graph information in which a plurality of nodes corresponding to each of a plurality of subjects, each of which is an individual, are linked according to the similarity of the plurality of subjects, and one individual who has made an important contribution in one company an acquisition unit that acquires target information about a target from a terminal device used by a user who organizes a project in a company different from the one company;
retrieving the graph information starting from a starting point node determined based on a predetermined criterion among the plurality of nodes of the graph information obtained by the obtaining unit and serving as a starting point for searching the graph information; , a search unit for extracting, from among the plurality of targets , a corresponding target similar to the individual who has made an important contribution to the one company, which is the one target;
The user who organizes the project uses information indicating that the person who is the target of correspondence extracted by the search unit is a person similar to the person who has made an important contribution to the one company. a providing unit that transmits to the terminal device to
An information processing device comprising: a determination unit that determines the origin node based on index information used to determine the origin node;
further comprising
The search unit is
The information processing apparatus according to claim 1, wherein the correspondence target is extracted with the starting node determined by the determination unit as a starting point. The decision unit
3. The information processing apparatus according to claim 2, wherein the starting node is determined based on the tree-structured index information. The acquisition unit
4. The method according to any one of claims 1 to 3, wherein graph information in which the plurality of nodes are connected is acquired according to the similarity of the plurality of vectors corresponding to each of the plurality of objects. Information processing equipment. The acquisition unit
5. The information processing apparatus according to claim 4, wherein the graph information is acquired in which the plurality of vectors indicating the characteristics of each of the plurality of targets are connected according to similarity. The acquisition unit
5. Obtaining the graph information in which the plurality of vectors whose elements are feature amounts extracted from each of the plurality of targets using a predetermined model are connected according to similarity. Or the information processing apparatus according to claim 5 . The acquisition unit
By inputting information about the plurality of targets into a predetermined model, the plurality of vectors whose elements are feature amounts of each of the plurality of targets to be extracted form the graph information linked according to similarity. 7. The information processing apparatus according to any one of claims 4 to 6, characterized in that: A computer-executed information processing method comprising:
Graph information in which a plurality of nodes corresponding to each of a plurality of subjects, each of which is an individual, are linked according to the similarity of the plurality of subjects, and one individual who has made an important contribution in one company an acquisition step of acquiring target information about a target from a terminal device used by a user who organizes a project in a company different from the one company;
retrieving the graph information starting from a starting point node determined based on a predetermined criterion among the plurality of nodes of the graph information obtained by the obtaining step, and serving as a starting point for searching the graph information; , a search step of extracting, from among the plurality of targets , a corresponding target similar to the individual who has made an important contribution to the one company, which is the one target;
The user who organizes the project uses information indicating that the person who is the target of correspondence extracted by the search step is a person similar to the person who has made an important contribution to the one company. a providing step of transmitting to the terminal device to
An information processing method characterized by including Graph information in which a plurality of nodes corresponding to each of a plurality of subjects, each of which is an individual, are linked according to the similarity of the plurality of subjects, and one individual who has made an important contribution in one company an acquisition procedure for acquiring target information about a target from a terminal device used by a user who organizes a project in a company different from the one company;
Searching for the graph information starting from a starting point node determined based on a predetermined criterion among the plurality of nodes of the graph information obtained by the obtaining procedure. , a search procedure for extracting, from among the plurality of targets , a corresponding target similar to the individual who has made an important contribution to the one company, which is the one target;
The user who organizes the project uses information indicating that the person who is the target of correspondence extracted by the search procedure is a person similar to the person who has made an important contribution to the one company. a providing procedure for transmitting to the terminal device to
An information processing program characterized by causing a computer to execute

Images