JP6865706B2 - Information processing equipment, information processing methods, and information processing programs - 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 types of information have been provided. For example, a technique for estimating the similarity between concepts is provided in consideration of the relative relationships between concepts belonging to other categories. Various services such as item delivery are provided based on the similarities between the concepts estimated in such techniques.

Japanese Unexamined Patent Publication No. 2007-21351

Masajiro Iwasaki "Neighborhood search by approximate k-nearest neighbor graph using tree-structured index", IPSJ Journal, 2011/2, Vol. 52, No. 2. pp.817-828.

However, with the above prior art, it may be difficult to properly extract the item. For example, it is difficult to extract desired information only by estimating the similarity 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 a case, it is difficult to extract desired information.

The present application has been made in view of the above, and an object of the present application is to provide an information processing device, an information processing method, and an information processing program for appropriately extracting items.

The information processing apparatus according to the present application is graph information in which a plurality of nodes corresponding to each of the plurality of items are connected by edges according to the similarity of the plurality of items, and among the plurality of nodes, the said Graph information in which nodes satisfying a predetermined condition from a viewpoint different from similarity are connected by a special edge different from the edge, an acquisition unit for acquiring item information related to one item, and an acquisition unit acquired by the acquisition unit. Among the plurality of items of the graph information, by searching the graph information from the starting node which is the starting point of the search of the graph information determined based on a predetermined criterion, the graph information is searched. , A similar item, which is an item similar to the one item, is extracted as a target item, and when the node reaches the node to which the special edge is connected at the time of searching the graph information, the node and the special edge are connected. It is characterized by including a search unit for extracting other nodes as the target item.

According to one aspect of the embodiment, there is an effect that the item can be appropriately extracted.

FIG. 1 is a diagram showing an example of information processing according to an embodiment. FIG. 2 is a diagram showing a configuration example of an information processing system according to an embodiment. FIG. 3 is a diagram showing a configuration example of the information processing device according to the embodiment. FIG. 4 is a diagram showing an example of the item information storage unit according to the embodiment. FIG. 5 is a diagram showing an example of the index information storage unit according to the embodiment. FIG. 6 is a diagram showing an example of a graph information storage unit according to the embodiment. FIG. 7 is a diagram showing an example of a model information storage unit according to the embodiment. FIG. 8 is a diagram showing an example of the history information storage unit according to the embodiment. FIG. 9 is a flowchart showing an example of information processing according to the embodiment. FIG. 10 is a flowchart showing an example of the generation process according to the embodiment. FIG. 11 is a diagram showing an example of extraction of the feature amount according to the embodiment. FIG. 12 is a flowchart showing an example of a search process using graph information. FIG. 13 is a diagram showing an example of adding a special edge. FIG. 14 is a hardware configuration diagram showing an example of a computer that realizes the functions of the information processing device.

Hereinafter, the information processing apparatus, the information processing method, and the mode for carrying out the information processing program (hereinafter referred to as “the embodiment”) 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 by this embodiment. Further, in each of the following embodiments, the same parts are designated by the same reference numerals, and duplicate description is 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 showing an example of information processing according to an embodiment. In FIG. 1, as an example of an item, the information processing device 100 targets similar items that are similar items by searching graph data (graph information) in which products sold by a predetermined electronic commerce service are graph-structured. The case of extracting as an item is shown. The product referred to here may be any target as long as it can be sold by a predetermined electronic commerce service, and may include a service or the like. Further, in FIG. 1, when the information processing apparatus 100 reaches a node to which a special edge described later is connected at the time of searching graph information, another node connected to the node by the special edge is extracted as a target item. Is shown. In FIG. 1, the information processing apparatus 100 uses graph information in which items are graph-structured using vector data (also referred to as “vector information” or simply “vector”) corresponding to each item. The information used by the information processing device 100 is not limited to a vector, and may be any form of information as long as it can express the similarity of each item. For example, the information processing apparatus 100 may use graph information in which items are graph-structured using predetermined data or values corresponding to each item. For example, the information processing apparatus 100 may use graph information in which items are graph-structured using predetermined numerical values (for example, binary values and hexadecimal values) generated from each item. Further, in the example of FIG. 1, a product (product information) is shown as an example of an item, but the item is not limited to the product (product information) and may be various objects such as article content and video content. That is, the target item (object) may be any item (information) as long as the similarity between the items can be expressed.

[1-1. About graph information]
Further, as shown in the graph information GR11 in FIG. 1, the information processing apparatus 100 performs information processing using graph data in which each vector (node) is connected by a directed edge. The information processing device 100 may generate the graph information as shown in the graph information GR 11 in FIG. 1, and the information processing device 100 may generate the graph information as shown in the graph information GR 11 in FIG. It may be obtained from another external device such as the providing device 50 (see FIG. 2).

Further, the directed edge here means an edge in which data can be traced in only one direction. In the following, the source traced by the edge, that is, the node that is the start point is referred to as the reference source, and the destination that is traced by the edge, that is, the node that is the end point is referred to. For example, the directed edge connected from the predetermined node "A" to the predetermined node "B" indicates that the reference source is the node "A" and the reference destination is the node "B". The edge connecting the nodes is not limited to the directed edge, and may be various edges. For example, the edge connecting the nodes may be an edge having no direction in connecting the nodes. For example, the edge connecting the nodes may be an edge that can be referred to each other. For example, the edges connecting the nodes may all be bidirectional edges.

For example, the edge with the node "A" as the reference source is called the output edge of the node "A". Further, for example, the edge with the node "B" as the reference destination is referred to as the input edge of the node "B". That is, the output edge and the input edge referred to here are differences in which of the two nodes in which the directed edge is connected is regarded as the center, and the one directed edge is regarded as the center. Becomes the output edge and the input edge. That is, the output edge and the input edge are relative concepts, and when one directed edge is regarded as the center of the reference source node, it becomes the output edge, and the reference destination node is regarded as the center. In some cases it becomes the input edge. In the present embodiment, since the edge is a directed edge such as an output edge or an input edge, the directed edge may be simply referred to as an "edge" below.

For example, the information processing apparatus 100 processes a node corresponding to a huge number of items in the unit of millions to hundreds of millions, but only a part thereof is shown in the drawing. In the example of FIG. 1, in order to simplify the explanation, eight nodes are illustrated to explain the outline of the process. For example, as shown in the graph information GR11 in FIG. 1, the information processing apparatus 100 acquires graph information including a plurality of nodes (vectors) as shown in the nodes N1, N2, N3, and the like. Further, in the example of FIG. 1, each node in the graph information GR 11 is connected with edges (output edges) from the one having the shortest distance to the nodes to a predetermined number of nodes. For example, the predetermined number may be various values such as 2, 5, 10, 100, etc., depending on the purpose, application, and the like. For example, when the predetermined number is 2, the output edge is connected from the node N1 to the node having the closest distance from the node N1 and the two nodes having the second closest distance. The distance as an index indicating the degree of similarity may be any distance as long as it can be applied as a distance between vectors (N-dimensional vectors), for example, Euclidean distance, Mahalanobis distance, cosine distance, and the like. Various distances may be used.

Further, when "node N * (* is an arbitrary numerical value)" is described in this way, it indicates that the node is a node identified by the node ID "N *". For example, when described as "node N1", the node is a node identified by the node ID "N1".

Further, in the graph information GR11 in FIG. 1, the node N10 is connected to the edge E7, which is a directed edge toward the node N7. That is, the node N10 is connected to the node N7 by the edge E7. When "edge E * (* is an arbitrary numerical value)" is described in this way, it indicates that the edge is an edge identified by the edge ID "E *". For example, when "edge E11" is described, the edge is an edge identified by the edge ID "E11". For example, the edge E7 connected with the node N10 as the reference source and the node N7 as the reference destination makes it possible to trace from the node N10 to the node N7. In this case, the edge E7, which is a directed edge, becomes an output edge when identified with the node N10 as the center, and becomes an input edge when identified with the node N7 as the center. Further, the double-headed arrow in the graph information GR11 of FIG. 1 indicates that the directed edges from both nodes to the other node are connected. For example, the bidirectional arrow between node N2 and node N451 in the graph information GR11 is a concatenation of two edges, a directed edge from node N2 to node N451 and a directed edge from node N451 to node N2. Indicates that it will be done.

Further, the graph information GR11 in FIG. 1 may be in Euclidean space. Further, the graph information GR 11 shown in FIG. 1 is a conceptual diagram for explaining the distance between each vector, and the graph information GR 11 is a multidimensional space. For example, the graph information GR11 shown in FIG. 1 is shown in a two-dimensional manner for being shown on a plane, but is assumed to be a multidimensional space such as 100 dimensions or 1000 dimensions.

Here, the distance between the vector data indicates the similarity of the items, and the closer the distance is, the more similar the items are. In the present embodiment, the distance of each node in the graph information GR11 is defined as the degree of similarity between the corresponding objects. For example, it is assumed that the similarity of the items corresponding to each node is mapped as the distance between the nodes in the graph information GR11. For example, assume that the similarity between the concepts corresponding to each node is mapped to the distance between each node. Here, in the example shown in FIG. 1, the similarity between the objects having a short distance between the nodes in the graph information GR11 is high, and the similarity between the objects having a long distance between the 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 is short. Therefore, it is shown 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, it is shown that 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" have a low degree of similarity.

[1-2. Vector generation example]
Further, each node (vector) referred to here corresponds to each object (item). If the item information is numerical data indicating the state of various items, it can be treated as it is as vector data. However, if the item information is text, image, or voice, conversion to vector data is required. Therefore, in the example of FIG. 1, the multidimensional (N-dimensional) vector generated by the feature amount extracted from the item information of each item may be an object.

For example, the information processing apparatus 100 may generate an N-dimensional vector from the item information of each item by using a model that extracts the features of the item information. In the example of FIG. 1, as shown in the model information storage unit 124 (see FIG. 7), the information processing apparatus 100 uses the model (model M1) identified by the model ID “M1” to provide item information for each item. Generate a vector from. As described above, when "model M * (* is an arbitrary numerical value)" is described, it means that the model is a model identified by the model ID "M *". For example, when described as "model M1", the model is a model identified by the model ID "M1". Further, as shown in the model information storage unit 124, the model M1 is a model used for the purpose "feature extraction (product)", that is, feature extraction from item data (item information), and a specific model thereof. Indicates that the data is "model data MDT1".

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

Here, FIG. 11 is used to show an example of generating vector data corresponding to each item. FIG. 11 is a diagram showing an example of extraction of the feature amount according to the embodiment. FIG. 11 is a conceptual diagram of the model M1. In FIG. 11, the line showing each connection relationship of each element (neuron) is omitted. As shown in FIG. 11, 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 item information is input. Further, the output layer OL is a layer that outputs the same information as the input item information in response to the input to the input layer IL.

Further, for example, the most compressed compression layer RP in the central portion of the intermediate layer CL is a layer expressing the characteristics of the input item 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, between the input layer IL and the compression layer RP corresponds to a portion that performs a process of compressing the characteristics of the input item 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 the portion where the decoding process is performed. 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 a process of restoring the compressed item information.

For example, the information processing apparatus 100 may use information such as neurons NL1 and neurons NL2 included in the compression layer RP as a vector. For example, the information processing apparatus 100 uses the value VE1 corresponding to the calculated neuron NL1 and the value VE2 corresponding to the neuron NL2 as vector elements (values of one dimension) when the item information of a certain item is input. It may be extracted. For example, the information processing apparatus 100 may extract the value VE1 corresponding to the calculated neuron NL1 as the first-dimensional element of the vector of the item when the item information of the item is input. Further, for example, the information processing apparatus 100 may extract the value VE2 corresponding to the calculated neuron NL2 as the second-dimensional element of the vector of the item when the item information of the item is input. In this way, the information processing apparatus 100 may generate a vector corresponding to each item by inputting the item information of each item into the model M1. The information processing device 100 may acquire a vector corresponding to each item from another external device such as the information providing device 50. The information processing apparatus 100 may use the value itself corresponding to each neuron as each element of the vector, or may use a value obtained by multiplying the value corresponding to each neuron by a predetermined coefficient. Further, in the example of FIG. 1, for the sake of simplicity, the case where each element (value) of the vector is an integer is shown, but each element (value) of the vector is a real number including a numerical value after the decimal point. May be good.

The information processing apparatus 100 is not limited to the elements (neurons) of the compression layer RP, and may use information of other elements (neurons) in the intermediate layer CL as a vector. For example, the information processing apparatus 100 may use information such as the neuron NL3 in the encoding portion and the neuron NL4 in the decoding portion as a vector. For example, the information processing apparatus 100 uses the value VE3 corresponding to the calculated neuron NL3 and the value VE4 corresponding to the neuron NL4 as vector elements (values of one dimension) when the item information of a certain item is input. It may be extracted. The above is an example, and the information processing apparatus 100 may extract features from item information using various models, not limited to the autoencoder. For example, a model trained to identify an item suitable for a predetermined purpose such as for sale and an item not suitable for sale may be generated, and the intermediate layer thereof may be extracted as vector data. The model may also be generated by, for example, a method of learning similarity such as triplet loss. Further, the information processing apparatus 100 may perform feature extraction without using a model. For example, the information processing device 100 may be set by the administrator of the information processing device 100 or the like to extract information corresponding to the feature (identity) from the item information and generate a vector. For example, the information processing apparatus 100 may generate a vector by extracting information corresponding to features (features) such as a product category, title, description, image information, manufacturer, and size of an item from the item information.

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 device 100 may generate the model M1 by inputting the item information IDT1, IDT2, IDT451, etc. of each item stored in the item information storage unit 121 (see FIG. 4). For example, the information processing apparatus 100 may generate the model M1 by inputting the item information IDT1, IDT2, IDT451, etc. of each item. For example, the item information IDT1, IDT2, IDT451 and the like may include the product category, title, description, image information, manufacturer, size and the like of the item. For example, the item information (product information) when the product or the like is an item may include information on the product category, title, description, image information, manufacturer, size, and the like. For example, the item information may include information indicating whether the product category of the corresponding product is drinking water, alcoholic beverages, seasonings, or the like. For example, the item information may include information indicating a detailed product category indicating whether the product category of the corresponding product is beer, wine, ketchup, sauce, or the like.

Further, for example, the item information may include image information obtained by capturing an image of the corresponding product. For example, the item information may include information indicating which manufacturer is the manufacturer of the corresponding product. For example, the item information may include information indicating what size the product has, such as the content of the corresponding product. For example, the item information includes information indicating the target of the product, such as information indicating that the target user of the corresponding product is in his twenties and information indicating that the target user of the corresponding product is a woman. But it may be. The above is an example, and any information may be included in the item information. For example, when the item is article content, the item information may include various information such as the title, body, and image of the article. That is, the information processing apparatus 100 may perform processing by appropriately using the item information according to the target to be an item.

The information processing device 100 includes an input layer into which item information of an item is input, an output layer, a first element which is any layer from the input layer to the output layer and belongs to a layer other than the output layer, and a first element. For the item information input to the input layer, including the second element whose value is calculated based on the element and the weight of the first element, each element belonging to each layer other than the output layer is set as the first element. By performing the calculation based on the first element and the weight of the first element, the model M1 that outputs the same information as the information input to the input layer from the output layer may be generated. The information processing device 100 may acquire a model used for vector generation from another external device such as the information providing device 50.

[1-3. Processing example]
From here, processing using graph information will be described. In the example of FIG. 1, after the addition of the special edge is described, the information processing using the graph information including the special edge will be described. The special edge referred to here connects nodes that satisfy predetermined conditions from a different viewpoint (reference) from the viewpoint (reference) for connecting the edges described as "edge E * (* is an arbitrary numerical value)". It may be an edge. For example, a special edge is an edge that connects nodes that satisfy a predetermined condition from a viewpoint (reference) different from the similarity of item information. That is, the special edge is an edge (extended edge) that expands the graph information by introducing an unusual viewpoint (reference) into the graph information.

In the following, the edge added to the graph information GR11 by the processing of steps S11 to S13 may be referred to as an edge special, and the other edges may be referred to as a "normal edge". The edge identified by the edge ID "E * (* is an arbitrary numerical value)" may be described as "normal edge". Further, the edge identified by the edge ID "SE * (* is an arbitrary numerical value)" may be described as "special edge". That is, the edge described as "edge E * (* is an arbitrary numerical value)" is a "normal edge", and the edge described as "special edge SE * (* is an arbitrary numerical value)" is a "special edge". Is.

In the example of FIG. 1, as an example of adding a special edge, a case where a plurality of items in which an item as a product is purchased together by a user are connected by a special edge is shown. That is, in the example of FIG. 1, the information processing device 100 connects a plurality of items to be purchased together at a special edge regardless of the similarity between the items in the user's purchasing behavior as an example. Shown.

First, the information processing device 100 acquires a log showing a predetermined history (step S11). The information processing device 100 acquires historical information indicating the user's behavior in a predetermined period. For example, the information processing device 100 acquires history information from the history information storage unit 125 (see FIG. 8). The predetermined period referred to here may be set to various times depending on the purpose, or may be various periods such as 1 hour, 3 days, 2 weeks, 1 month, and the like. In the example of FIG. 1, the predetermined period is one week including the date and time # 11 to # 13, and may be described as the collection target period.

For example, as shown in the log list LGT11, the information processing apparatus 100 acquires the user's action history in the collection target period including the date and time # 11 to # 13. In the example of FIG. 1, although not shown, the information processing apparatus 100 may acquire history information including information of a user who has performed each action.

"Date and time" indicates the date and time when the corresponding action was performed. The date and time # 11 to date and time # 13 and the like are assumed to be specific dates and times when the corresponding action was performed. "Target item # 1 (purchased product # 1)", "target item # 2 (purchased product # 2)", etc. indicate items that are commonly targeted in the corresponding action. In the example of FIG. 1, "target item # 1 (purchased product # 1)", "target item # 2 (purchased product # 2)", and the like indicate items (products) purchased together in the corresponding action.

In the example shown in FIG. 1, it is shown that item # 35 and item # 831 were purchased together at the date and time # 11 within the collection target period. In addition, it indicates that the item # 1 was purchased alone at the date and time # 11 within the collection target period. It also indicates that item # 35 and item # 831 were purchased together at the date and time # 11 within the collection target period.

Then, the information processing apparatus 100 determines an item having a special relevance based on the history information as shown in the log list LGT 11 (step S12). For example, the information processing apparatus 100 determines in the log list LGT11 that items that are included together as targets in one action are items having a special relationship.

For example, the information processing device 100 determines that, in the behavior of a group of users in a predetermined period, a plurality of items in which the number of co-occurrence as an action target is equal to or greater than a predetermined threshold value are items having a special relationship. For example, the information processing device 100 is an item having a special relationship with a plurality of nodes corresponding to a plurality of items, which are a plurality of products whose number of purchases in a predetermined period is equal to or greater than a predetermined threshold value. Is determined. The special relationship between the items is not limited to the above, and is a viewpoint different from the viewpoint when the graph information is generated, and any relationship may be used as long as it is a significant relationship between the items. ..

For example, the information processing apparatus 100 has a special relationship with a plurality of nodes purchased at a predetermined threshold value (for example, 1000 times or more) within the collection target period including the date and time # 11 to # 13. Determined to be an item. In the example of FIG. 1, the information processing apparatus 100 determines that two items (goods), item # 35 and item # 831, have been purchased in accordance with a predetermined threshold value or more within the collection target period. Then, the information processing apparatus 100 determines that the item # 35 and the item # 831 are items having a special relationship. Therefore, the information processing apparatus 100 determines that a special edge is added between the item # 35 and the item # 831.

Then, the information processing apparatus 100 adds a special edge that connects items having a special relationship. In the example of FIG. 1, the information processing apparatus 100 adds a special edge SE35 and a special edge SE831 as shown in the special edge list SEL11.

The information processing device 100 generates a special edge SE35 connected from the node N35 to the node N831. The information processing apparatus 100 generates a special edge SE35 with the node N35 as the reference source and the node N831 as the reference destination. The information processing device 100 adds (registers) the information of the special edge SE35 to the graph information storage unit 123 (see FIG. 6).

Further, the information processing apparatus 100 generates a special edge SE831 connected from the node N831 to the node N35. The information processing apparatus 100 generates a special edge SE831 with the node N831 as the reference source and the node N35 as the reference destination. The information processing device 100 adds (registers) the information of the special edge SE831 to the graph information storage unit 123 (see FIG. 6).

For example, the information processing apparatus 100 adds a special edge SE35 and a special edge SE831 (not shown) shown by a dotted line to the graph information GR11 as shown in the information group INF11 in FIG. As a result, the information processing apparatus 100 generates graph information in which items having a special relationship are connected by a special edge.

From here, information provision using the graph information GR11 to which the special edge SE35 is added will be described as an example. The case where the information processing apparatus 100 acquires the item information of one item from the user and provides the service based on the item similar to the one item will be described as an example. In the example of FIG. 1, the terminal device 10 is a smartphone used by the user U1, and shows a case where the user U1 is browsing the product information of the product which is an item.

First, the terminal device 10 transmits the item information IDT11 of the item X to the information processing device 100. In the example of FIG. 1, the item information IDT11 and the like include information on the product category, title, description, image information, manufacturer, size, etc. of the item (product). The item information IDT11 and the like may be only the description (character information) of the item (product).

Then, the information processing device 100 acquires the item information regarding one item (step S14). In the example of FIG. 1, the information processing device 100 acquires the item information IDT11 related to the item X from the terminal device 10 of the user U1. The information processing device 100 acquires the item information IDT 11 related to the item X from the terminal device 10 (see FIG. 2) carried by the item X.

Then, the information processing apparatus 100 generates a vector used for searching graph information from the item information corresponding to one item. In the example of FIG. 1, the information processing apparatus 100 generates a vector corresponding to the item X by the processing as shown in the processing group PS11. The information processing device 100 inputs the item information IDT11 related to the item X into the model M1 (step S15). Specifically, the information processing apparatus 100 inputs the item information IDT11 of the item X into the model M1. Then, the information processing apparatus 100 generates a vector by using the information in the model M1 after inputting the item information IDT 11 (step S16). For example, the information processing apparatus 100 generates vector data using each element in the model M1 in which the item information IDT11 is input.

In the example of FIG. 1, the information processing apparatus 100 generates vector data VD11 (also simply referred to as “vector VD11”) by using the value of each element in the model M1 in which the item information IDT11 is input. For example, the information processing apparatus 100 sets a value VE1 (see FIG. 11) corresponding to the neuron NL1 of the model M1 and a value VE2 (see FIG. 11) corresponding to the neuron NL2 when the item information IDT11 of the item X is input. To generate a vector. For example, the information processing apparatus 100 may extract the value VE1 corresponding to the calculated neuron NL1 as the first-dimensional element of the vector VD11 when the item information IDT11 of the item X is input. Further, for example, the information processing apparatus 100 generates the vector VD11 by using the value VE2 corresponding to the calculated neuron NL2 as the second-dimensional element of the vector VD11 when the item information of a certain item is input. In the example of FIG. 1, the information processing apparatus 100 generates a vector VD11 in which the element of the first dimension is "35" and the element of the second dimension is "63".

Then, the information processing device 100 searches for the target item (step S17). For example, the information processing device 100 searches for a target item related to the item X. For example, the information processing apparatus 100 may search for the target item of item X by appropriately using various conventional techniques such as a neighborhood search technique disclosed in Non-Patent Document 1. For example, the information processing apparatus 100 uses the graph information GR11 in which nodes having high similarity are connected by an edge.

In the example of FIG. 1, the information processing apparatus 100 searches for the target item of the item X by 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 item from the graph information storage unit 123 (see FIG. 6). Further, for example, the information processing apparatus 100 uses the index information storage unit 122 (see FIG. 5) to obtain the index information IND11 used for determining the node (hereinafter, also referred to as “starting point vector”) that is the starting point of the search in the graph information GR11. get. The index information IND 11 may be generated by the information processing device 100, or the information processing device 100 may acquire the index information IND 11 from another external device such as the information providing device 50.

Then, the information processing apparatus 100 uses the index information IND11 to determine (specify) the starting point vector corresponding to one item (query). In the example of FIG. 1, the information processing apparatus 100 uses the index information IND11 to determine (specify) the starting point vector corresponding to the vector VD11 of the item X. That is, the information processing apparatus 100 uses the vector VD11 and the index information IND11 to determine the starting point vector in the graph information GR11.

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

For example, the information processing apparatus 100 determines the starting point vector in the graph information GR 11 by using the tree-structured index information as shown in the index information IND 11 in FIG. 1 (step S18). In the example of FIG. 1, the information processing apparatus 100 efficiently generates the vector VD11 and then traces the index information IND11 from top to bottom to identify a starting point vector that is a candidate near the index information IND11. The origin vector corresponding to the search query (one item) can be determined.

For example, the information processing apparatus 100 may determine the starting point vector corresponding to the vector VD11 by tracing the index information IND11 from the root RT to the leaf node (node (vector) in the graph information GR11). In the example of FIG. 1, for example, the information processing apparatus 100 determines the node N451 as a starting point 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 leaf node reached as a starting point vector by tracing the index information IND11 from the root RT to the leaf node by appropriately using various conventional techniques related to the tree structure. For example, the information processing apparatus 100 may determine the starting point vector by tracing the index information IND11 downward based on the similarity with the vector VD11. For example, the information processing apparatus 100 may determine which node, such as node VT1 or VT2, is to be traced from the root RT based on the degree of similarity between the vector VD11 and the nodes VT1 and VT2. For example, the information processing apparatus 100 may determine that the root RT traces to the node VT2 having the highest degree of similarity to the vector VD11 among the nodes VT1, VT2, and the like. Further, for example, the information processing apparatus 100 may determine from the node VT2 to the node VT2-2 having the highest degree of similarity to the vector VD11 among the nodes VT2-1 to VT2-4 and the like. Further, for example, the information processing apparatus 100 may determine that the node VT2-2 traces to the node node N451 having the highest similarity to the vector VD11 among the nodes N35, N451, N693 and the like. In the example of FIG. 1, for the sake of simplicity, the case where one starting point vector is determined is shown, but the information processing apparatus 100 may determine a plurality of starting point vectors. For example, the information processing apparatus 100 may determine a plurality of vectors (nodes) such as nodes N451, N35, N693, and N2 as starting point vectors. It should be noted that one or more nodes may be randomly selected from the graph information GR11 at the start of the search without using the index information IND11 and used as a starting point vector, or one or more nodes specified in advance may be used as the starting point. It may be a vector.

Then, the information processing apparatus 100 extracts the target item of the item X by searching the graph information GR11 (step S19). For example, the information processing device 100 extracts a node located in the vicinity of the node N451 as a target item. For example, the information processing device 100 extracts a node having a short distance from the node N451 as a target item. For example, the information processing apparatus 100 extracts a node that can be reached from the node N451 as a target item by tracing the edge starting from the node N451. For example, the information processing device 100 extracts a predetermined number of nodes (for example, 2 or 10 nodes) as target items. For example, the information processing apparatus 100 may extract the target item of the item X by the search process as shown in FIG. 12, but the details will be described later. In the example of FIG. 1, the information processing apparatus 100 extracts the node N451 and the node N35 as target items by searching the graph information GR11 starting from the node N451.

Here, in step S19, when the information processing apparatus 100 extracts a node (hereinafter, also referred to as “normal extraction node”) by tracing an edge (normal edge) other than the special edge in the graph information GR11, the information processing apparatus 100 extracts the node (hereinafter, also referred to as “normal extraction node”) from the normal extraction node. Determine if a special edge is connected to another edge. For example, the information processing apparatus 100 determines whether or not there is a special edge that connects the normal extraction node and another object with the normal extraction node as a reference source. Then, when the information processing apparatus 100 determines that there is a special edge connecting the normal extraction node and another object (special object) with the normal extraction node as a reference source, it determines that the special object is extracted.

First, when the node N451 is extracted as a target item, the information processing apparatus 100 determines whether or not a special edge is connected to another edge from the node N451. In the example of FIG. 1, when the node N451 which is the starting node is extracted, the information processing apparatus 100 determines whether or not a special edge is connected to another edge from the node N451. For example, the information processing apparatus 100 determines that there are no special edges from the node N451, that is, there are no special edges from the node N451 to other edges. Then, the information processing apparatus 100 determines that the special object related to the node N451 is not extracted, and performs a search at the normal edge.

Next, when the node N35 is extracted as the target item, the information processing apparatus 100 determines whether or not the special edge is connected to another edge from the node N35. In the example of FIG. 1, when the node N35 is extracted from the node N451 by tracing the normal edge, the information processing apparatus 100 determines whether or not the special edge is connected to another edge from the node N35. For example, the information processing apparatus 100 determines that there is a special edge from the node N35 to another edge because the special edge SE35 from the node N35 to the node N831 is connected. Therefore, the information processing apparatus 100 extracts the node N831 to which the special edge SE35 is connected from the node N35 as the target item.

The information processing device 100 does not have to add the special object to the number of extracted objects. That is, the information processing apparatus 100 does not have to use the number of special objects for comparison with a predetermined number. In this case, the information processing apparatus 100 may use the number of objects excluding special objects among the objects extracted as the target items for comparison with a predetermined number. Further, the information processing apparatus 100 does not have to search for a node that can be traced from the node extracted as the target item by connecting the special edges. In the example of FIG. 1, the information processing apparatus 100 does not search for the node that can be traced from the node N831 extracted as the target item by connecting the special edges. In this case, the information processing apparatus 100 performs a search by tracing a normal edge from the node N35. As a result, the information processing apparatus 100 can suppress the extraction of nodes having a low relevance to the item X as target items. The information processing device 100 may extract a node that can be traced from the node N831 as a target item.

Then, the extraction device 100 provides information regarding the extracted target item (step S20). In the example of FIG. 1, the extraction device 100 is a terminal device used by the user U1 with the item # 451 corresponding to the node N451, the item # 35 corresponding to the node N35, and the item # 831 corresponding to the node N831 as target items. Provided to 10.

As described above, the information processing apparatus 100 extracts the target item of the item X acquired from the user U1 by using the graph information GR11 and the index information IND11. For example, the information processing apparatus 100 uses the index information IND11 to determine the starting point vector in the graph information GR11 when extracting the target item of the item X. Then, the information processing apparatus 100 extracts the target item of the item X by searching the graph information GR11 with the determined starting point vector as the starting point. As a result, the information processing apparatus 100 can appropriately extract items.

Further, when the information processing apparatus 100 reaches the node to which the special edge is connected at the time of searching the graph information, the information processing apparatus 100 extracts the node and another node connected by the special edge as the target item, so that the information processing device 100 can be viewed from a different viewpoint. Target items can also be extracted. For example, when the information processing apparatus 100 reaches a node to which a special edge is connected at the time of searching graph information, the information processing apparatus 100 extracts the node and another node connected by the special edge as a target item, so that the information processing device 100 can be viewed from a different viewpoint. Target items can also be extracted.

For example, the information processing apparatus 100 uses graph information to make an item similar to item X (similar item) even when an item similar to item X (similar item) does not seem to have a relationship with item X at first glance. It can be extracted as a target item), and the item can be extracted appropriately. For example, there is a method of identifying an item suitable for a predetermined purpose such as a sale item by a method such as deep learning, but when there are various factors that affect whether or not the item is suitable for a predetermined purpose, each item is used. Since there are few examples of elements, there is a problem that it is difficult to improve the identification accuracy without learning. For example, there is a method of inputting item information into the model using a model generated by a learning method such as deep learning and identifying whether the item is suitable for a predetermined purpose from the output score, but there are various methods. When there are various elements, it is difficult to improve the accuracy of the model and it is difficult to improve the identification accuracy because there are few cases of each element. On the other hand, the information processing apparatus 100 can appropriately extract an item by performing a search using graph-structured graph information of a plurality of vectors generated based on the features extracted from each item.

[1-4. About generation of graph information (addition of special edges)]
As described in the example of FIG. 1, the information processing apparatus 100 can appropriately extract items by connecting nodes satisfying a predetermined condition with a special edge and searching a graph based on the connection. it can. Here, the conventional graph-based index (hereinafter referred to as “conventional graph”) is also effective for searching and recommending (recommendation) of ultra-large-scale items. On the other hand, when a conventional graph is constructed from a certain viewpoint, it is difficult to dynamically (for example, for a limited time) consider new information from another viewpoint when making an item recommendation using the conventional graph. There are challenges.

For example, the following is an example of dynamically (for example, for a limited time) considering new information of the viewpoint. For example, when there is graph information (index) constructed from item content distances (similarities), you may want to consider the distances related to how often they were bought together in the last few hours. For example, by using the user's behavior history, we can extract clusters of items that tend to be bought together depending on the situation, such as when it is raining, when it is hot, when it is cold, or when an earthquake occurs. You may want to reflect that information. However, it takes time to reconstruct the graph information of the conventional graph when the situation changes. In addition, it is difficult and impractical to individually prepare the graph information of the conventional graph assuming all cases in advance.

Therefore, the information processing apparatus 100 heuristically expands the graph information by using information from another viewpoint (distance scale or the like). For example, the information processing apparatus 100 may update the graph information using the distance (similarity) regarding the frequency of being bought together in the last few hours. For example, the information processing apparatus 100 adds an extended edge (special edge) to the product A and the product B whose number of purchases has increased rapidly in the last few hours. For example, when a predetermined celebrity (artist A) talks about a product AA on a television, the information processing device 100 is referred to as a product of the artist A (for example, music content such as a CD (Compact Disc)). The product AA is connected by an extended edge (special edge) on the graph information. The information processing device 100 may perform the operation automatically, or the administrator of the information processing device 100 may perform the operation manually.

In this case, the information processing apparatus 100 does not need to reconstruct the entire graph information and can be updated instantly. Further, the information processing apparatus 100 can instantly return to the original graph information by deleting the added edge when the service is provided using the updated graph information and the desired effect is not obtained.

For example, if you narrow down the candidates searched by the conventional graph by another scale, you can think of a two-step extraction method. However, this extraction method is not efficient because the desired target product may not be extracted. In addition, when searching and merging separately for each scale, potential demand may be overlooked because the search (node transition) considering both scales is not considered.

On the other hand, even if the information processing device 100 cannot calculate an appropriate distance from the acquired data, the administrator of the information processing device 100 forcibly "likes this" based on past knowledge (experience knowledge). You can embed information that shows special relationships between nodes, such as "people should like this." For example, the information processing apparatus 100 can also specify relative values such as X times the similarity of the products A and B of the graph information, which is difficult from a single viewpoint. For example, as described above, the information processing apparatus 100 enables a search based on various viewpoints by adding a special edge from a viewpoint (scale) other than the similarity of items, and enables appropriate extraction of items. To.

For example, if the similarity is defined (set) as "the frequency of pages that co-occur (viewed together) within a certain period of time in each user's browsing log", the distance function changes in real time, so the graph changes in real time. It is difficult with the current technology to completely reflect it in the information. Therefore, the information processing apparatus 100 detects a frequent path (shortcut) from the log by another algorithm, adds an edge, and the like, thereby expanding and processing the current graph information in real time (heuristically), and at a predetermined value. The entire graph information may be updated in batch during the time zone (midnight, etc.).

[1-4-1. Special edge]
In the example of FIG. 1, the case where the information processing device 100 determines whether to connect the items with a special edge based on the behavior of the user in a predetermined period is shown, but the information processing device 100 is limited to the period. Instead, it may be determined whether or not the items are connected by a special edge based on various conditions. For example, the information processing device 100 may use a context such as a user's position information. For example, the information processing device 100 may target a user's action group when the user's context satisfies a predetermined condition, and determine whether to connect the items with a special edge based on the information group. .. For example, the information processing device 100 may target a user's action group located in a predetermined area and determine whether to connect the items with a special edge based on the information group. For example, the information processing device 100 may determine that if there is a special relationship between items in the behavior of a user located in a predetermined area, the items are connected by a special edge. Further, even if the information processing device 100 determines that the special edge (special edge X) added based on the predetermined area (X area) is used when the user located in the X area is the information providing destination. Good. For example, when the user located in the X area is the information providing destination, the information processing device 100 extracts the target item by tracing the special edge X at the time of searching. Further, when the user located in the area other than the X area is the information providing destination, the information processing device 100 extracts the target item without following the special edge X at the time of searching. The information processing apparatus 100 may determine whether or not to use the special edge based on various information. In this way, the information processing apparatus 100 may determine the availability of the special edge by using various information. The information processing apparatus 100 may decide whether or not to extract the target item by using the special edge by using the context such as the position information of the user. The information processing apparatus 100 may decide whether or not to extract the target item by using the special edge depending on whether or not the user is located in a predetermined area. For example, the information processing apparatus 100 may determine whether to extract the target item using the special edge based on the attribute information of the user who is the information providing destination.

For example, the information processing apparatus 100 may store information indicating a target user attribute in association with a special edge. In this case, the information processing apparatus 100 compares the user attribute associated with the special edge with the user attribute of the information providing destination user, and the user attribute of the user matches the user attribute associated with the special edge. , You may decide to use that special edge.

For example, in the information processing device 100, when a special edge (referred to as "special edge Y") whose user attribute is a male in his twenties is included in the graph information and the user to whom the information is provided is a male in his twenties, it is special. The target item is extracted using the edge Y. On the other hand, in the information processing device 100, even if the graph information includes a special edge Y whose user attribute is a man in his twenties, the special edge Y is when the user to whom the information is provided is a woman in his twenties. Extract the target item without using.

Further, for example, the information processing apparatus 100 may determine the addition of a special edge by using various information, not limited to the user's purchase information. For example, the information processing device 100 may determine the addition of a special edge by using the user's behavior information including the posted information posted by the user in a predetermined service. For example, the information processing device 100 determines that a special edge is added between item # 22 and item # 33 when item # 22 and item # 33 co-occur in the user's posted information more than a predetermined threshold value. You may.

For example, the information processing device 100 may determine the addition of a special edge by using the posted information posted by the user on the SNS. For example, the information processing apparatus 100 may determine the addition of a special edge by using the posted information posted by the user on Twitter (registered trademark), Facebook (registered trademark), or the like. For example, the information processing device 100 may determine the addition of a special edge by using the user's behavior information including the character information posted by the user on Twitter. For example, the information processing apparatus 100 may determine the addition of a special edge by using the user's action information including the information input by the user in the message service. For example, the information processing apparatus 100 may determine the addition of a special edge by using the user's behavior information including the information input by the user in a message service such as LINE (registered trademark). For example, the information processing device 100 may determine the addition of a special edge by using the user's behavior information including the information input in the e-mail. The information processing device 100 may determine the addition of a special edge by using various behavior information of the user, not limited to the posted information.

Next, the addition of a special edge to the graph information will be described with reference to FIG. FIG. 13 is a diagram showing an example of adding a special edge. First, the information processing device 100 acquires a log showing a predetermined history (step S51). The information processing device 100 acquires historical information indicating the user's behavior in a predetermined period. For example, the information processing device 100 acquires history information from the history information storage unit 125 (see FIG. 8). In the example of FIG. 13, the predetermined period is 3 months including the date and time # 51 to # 53, and may be described as the collection target period.

For example, as shown in the log list LGT51, the information processing apparatus 100 acquires the user's action history in the collection target period including the date and time # 51 to # 53 and the like. In the example of FIG. 13, although not shown, the information processing apparatus 100 may acquire history information including information of a user who has performed each action.

"Date and time" indicates the date and time when the corresponding action was performed. The date and time # 51 to date and time # 53 and the like are assumed to be specific dates and times when the corresponding action was performed. "Target item # 1", "target item # 2", etc. indicate items that are commonly targeted in the corresponding action. In the example of FIG. 13, "target item # 1", "target item # 2", and the like indicate items (products) purchased together in the corresponding action.

In the example shown in FIG. 13, it is shown that the item # 51 and the item # 555 were purchased together at the date and time # 51 within the collection target period. It also indicates that item # 51 and item # 555 were purchased together at the date and time # 51 within the collection target period. It also indicates that item # 555 and item # 51 were purchased together at the date and time # 51 within the collection target period.

Then, the information processing apparatus 100 determines an item having a special relevance based on the history information as shown in the log list LGT51 (step S52). For example, the information processing apparatus 100 has a special relationship with a plurality of nodes purchased at a predetermined threshold value (for example, 5000 times or more) within the collection target period including the date and time # 51 to # 53 and the like. Determined to be an item. In the example of FIG. 13, the information processing apparatus 100 determines that two items (goods), item # 51 and item # 555, have been purchased in accordance with a predetermined threshold value or more within the collection target period. Then, the information processing apparatus 100 determines that the item # 51 and the item # 555 are items having a special relationship. Therefore, the information processing apparatus 100 determines that a special edge is added between the item # 51 and the item # 555.

Then, the information processing apparatus 100 adds a special edge that connects items having a special relationship. In the example of FIG. 13, the information processing apparatus 100 adds a special edge SE51 and a special edge SE555.

The information processing device 100 generates a special edge SE51 connected from the node N51 to the node N555. The information processing device 100 generates a special edge SE51 with the node N51 as the reference source and the node N555 as the reference destination. For example, the information processing apparatus 100 adds (registers) the information of the special edge SE51 to the graph information storage unit 123 (see FIG. 6).

Further, the information processing apparatus 100 generates a special edge SE555 connected from the node N555 to the node N51. The information processing apparatus 100 generates a special edge SE555 with node N555 as a reference source and node N51 as a reference destination. The information processing device 100 adds (registers) the information of the special edge SE555 to the graph information storage unit 123 (see FIG. 6).

For example, the information processing apparatus 100 adds the special edge SE51 and the special edge SE555 shown by the dotted line to the graph information GR21 in FIG. As a result, the information processing apparatus 100 generates graph information in which items having a special relationship are connected by a special edge. The information processing apparatus 100 may remove the special edge SE51 and the special edge SE555 from the graph information GR 21 when a predetermined period has elapsed from the addition of the special edge SE51 and the special edge SE555. When a predetermined period has elapsed from the addition of the special edge SE51 or the special edge SE555, the information processing apparatus 100 may delete the special edge SE51 or the special edge SE555 from the graph information storage unit 123 (see FIG. 6).

In this way, as shown in the graph information GR21, the information processing device 100 makes a special relationship based on the user's behavior even if the items are far apart from each other, that is, the items have low similarity in the item information. Judge and connect with a special edge. As a result, the information processing device 100 can generate graph information adaptive to the user's behavior and provide the user with adaptive information.

[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 for graph information using various index information. Further, for example, the information processing apparatus 100 may generate index data to be used at the time of search. For example, the information processing apparatus 100 generates a search index for searching a high-dimensional vector as index data. The high-dimensional vector referred to here may be, for example, a vector having several hundred dimensions to several thousand dimensions, or a vector having 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 related to a kd tree (k-dimensional tree) as index data. For example, the information processing apparatus 100 may generate a search index related to 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 index data in which the leaf of the index data of the tree structure is connected to the graph data. For example, the information processing apparatus 100 may generate various index data in which the leaf of the index data of the tree structure corresponds to the node in the graph data. Further, when the information processing apparatus 100 searches using such index data, the information processing apparatus 100 may search the graph data from the leaf (node) reached by tracing the index data.

The index data as described above is 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 specified 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 partitioning if it is possible to identify a node in the graph information corresponding to the query at high speed. You may. For example, the information processing apparatus 100 may generate index data having any data structure as long as it is an index capable of searching for a high-dimensional vector. For example, the information processing apparatus 100 may use information on a graph-type search index as described in Non-Patent Document 1 as index information. The information processing apparatus 100 can enable a more efficient search for a predetermined item by using the index data and the graph data as described above.

[2. Information processing system configuration]
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 to each other via a predetermined network N so as to be communicable by wire or wirelessly. FIG. 2 is a diagram showing a configuration example of an information processing system according to an embodiment. The information processing system 1 shown in FIG. 2 may include a plurality of terminal devices 10, a plurality of information providing devices 50, and a plurality of information processing devices 100.

The terminal device 10 is an information processing device used by the user. The terminal device 10 accepts various operations by the user. In the following, the terminal device 10 may be referred to as a user. That is, in the following, the user can be read as the terminal device 10. The terminal device 10 described above is realized by, for example, a smartphone, 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 an administrator uses a predetermined server system. For example, the terminal device 10 collects item information and transmits the item information to the information processing device 100.

The information processing device 100 is an information processing device that performs various types of information processing. The information processing device 100 is an information processing device that extracts similar items that are similar to one item among a plurality of items as target items by searching graph information. The information processing device 100 is an information processing device that, when a node to which a special edge is connected is reached at the time of searching graph information, extracts another node connected to the node by the special edge as a target item. For example, the information processing apparatus 100 acquires item information related to one item, and determines the starting point vector based on the item information and the information related to the starting point vector that is the starting point for searching the graph information.

The information processing device 100 is an information processing device in which information for providing various information to a user or the like is stored. For example, when the information processing device 100 acquires the item information (hereinafter, also referred to as "query information" or "query") of one item from the terminal device 10, it searches for an item (vector information or the like) similar to the query. The search result is provided to the terminal device 10. In the example of FIG. 1, when the information processing device 100 acquires the item information of one item from the terminal device 10, it searches for an item similar to the one item and provides the search result to the terminal device 10 as a similar item. .. Further, for example, the data provided by the information processing device 100 to the terminal device 10 may be the item name or the item information itself of the item, or for referring to the corresponding data such as a URL (Uniform Resource Locator). It may be information.

The information providing device 50 is an information processing device in which information for providing various information to the information processing device 100 is stored. For example, the information providing device 50 may store item information or 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 a model to the information processing device 100.

[3. Information processing device configuration]
Next, the configuration of the information processing apparatus 100 according to the embodiment will be described with reference to FIG. FIG. 3 is a diagram showing a configuration example of the information processing device according to the embodiment. As shown in FIG. 3, the information processing device 100 includes a communication unit 110, a storage unit 120, and a control unit 130. The information processing device 100 includes an input unit (for example, a keyboard, a mouse, etc.) that receives various operations from the administrator of the information processing device 100, and a display unit (for example, a liquid crystal display, etc.) for displaying various information. You may have.

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

(Memory unit 120)
The storage unit 120 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory (Flash Memory), or a storage device such as a hard disk or an optical disk. As shown in FIG. 3, the storage unit 120 according to the embodiment includes an item information storage unit 121, an index information storage unit 122, a graph information storage unit 123, a model information storage unit 124, and a history information storage unit 125. Have.

(Item information storage unit 121)
The item information storage unit 121 according to the embodiment stores various information related to the item (object). For example, the item information storage unit 121 stores item IDs and vector data. In the example of FIG. 4, item information about the item is stored. FIG. 4 is a diagram showing an example of the item information storage unit according to the embodiment. The item information storage unit 121 shown in FIG. 4 includes items such as "item ID", "item", "item information", and "vector information".

The "item ID" indicates identification information for identifying an item (object). Further, the "item" indicates a specific name, content, or the like of the item identified by the item ID. In the example of FIG. 4, the item is indicated by an abstract code such as "item # 1", but includes character information for identifying the item (product) such as "A product" in parentheses. May be.

"Item information" indicates information about an item identified by an item ID. In the example of FIG. 4, the item information is indicated by an abstract code such as "IDT1", but each item information is the product category, title, description, image information, manufacturer, and size of the item identified by the item ID. May include information about various items such as. The "vector information" indicates the vector information corresponding to the item (object) identified by the item ID. That is, in the example of FIG. 4, the vector data (vector information) corresponding to the object is associated and registered with the item ID that identifies the item (object).

For example, in the example of FIG. 4, the item (object) identified by the item ID "IM1" is associated with the multidimensional (N-dimensional) vector information of "10, 24, 54, 2 ...". Is shown. For example, for item # 1, multidimensional (N-dimensional) vector information of "10, 24, 54, 2 ..." indicating the characteristics of item # 1 is extracted from item information IDT1 by model M1 or the like. Show that. Further, in the example of FIG. 4, the index value of item # 1 is shown to be “VL1”.

The item information storage unit 121 is not limited to the above, and various information may be stored depending on the purpose.

(Index information storage unit 122)
The index information storage unit 122 according to the embodiment stores various information related to the index. FIG. 5 is a diagram showing an example of the index information storage unit according to the embodiment. Specifically, in the example of FIG. 5, the index information storage unit 122 shows index information having a tree structure. In the example of FIG. 5, the index information storage unit 122 includes items such as "root hierarchy", "first hierarchy", "second hierarchy", and "third hierarchy". In addition, not limited to "first layer" to "third layer", "fourth layer", "fifth layer", "sixth layer" and the like may be included depending on the number of layers of the index.

The "route hierarchy" indicates a hierarchy of routes (top level) that is a starting point for determining a starting node using an index. The "first layer" stores information for identifying (identifying) a node (vector in node or graph information) belonging to the first layer of the index. The node stored in the "first layer" is a node corresponding to the layer directly connected to the root of the index.

The "second layer" stores information for identifying (identifying) a node (vector in node or graph information) belonging to the second layer of the index. The node stored in the "second layer" is a node corresponding to the immediately lower layer connected to the node of the first layer. The "third layer" stores information for identifying (identifying) a node (a vector in node or graph information) belonging to the third layer of the index. The node stored in the "third layer" is a node corresponding to the immediately lower layer connected to the node of 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 nodes VT1 to VT3 and the like. The numerical value in parentheses below each node indicates the value of the vector corresponding to each node.

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

The index information storage unit 122 is not limited to the above, and various information may be stored depending on the purpose.

(Graph Information Storage Unit 123)
The graph information storage unit 123 according to the embodiment stores various information related to the graph information. FIG. 6 is a diagram showing an example of a graph information storage unit according to the embodiment. In the example of FIG. 6, the graph information storage unit 123 has items such as "node ID", "item ID", "edge information", and "special edge information". Further, the "edge information" and "special edge information" include information such as "edge ID" and "reference destination".

The "node ID" indicates identification information for identifying each node (item) in the graph data. Further, the "item ID" indicates identification information for identifying an item (object).

Further, "edge information" indicates information about an edge connected to the corresponding node. "Edge information" indicates information about an edge that connects nodes according to the similarity of a plurality of items. For example, "edge information" usually indicates information about an edge. In the example of FIG. 6, “edge information” indicates a case where the edge is a directed edge, and indicates information about an output edge output from the corresponding node. Further, the "edge ID" indicates identification information for identifying an edge connecting the nodes. Further, "reference destination" indicates information indicating a reference destination (node) connected by an edge. That is, in the example of FIG. 6, the reference destination (output edge) to which the information for identifying the object (item) corresponding to the node and the directed edge (output edge) from the node are connected to the node ID for identifying the node. Node) is associated and registered.

Further, "special edge information" indicates information about the special edge connected to the corresponding node. "Special edge information" indicates information on an edge (special edge) that connects nodes that satisfy a predetermined condition from a viewpoint different from the viewpoint of connecting normal edges. In the example of FIG. 6, “special edge information” indicates a case where the special edge is a directed edge, and indicates information about an output edge output from the corresponding node. Further, the "edge ID" indicates identification information for identifying a special edge connecting the nodes. Further, "reference destination" indicates information indicating a reference destination (node) connected by a special edge. That is, in the example of FIG. 6, the reference destination (output edge) to which the information for identifying the object (item) corresponding to the node and the directed edge (output edge) from the node are connected to the node ID for identifying the node. Node) is associated and registered.

For example, in the example of FIG. 6, it is shown that the node (vector) identified by the node ID “N1” corresponds to the item (object) identified by the item ID “IM1”.

Further, from the node identified by the node ID "N1", the edge (normal edge) identified by the edge ID "E11" is connected to the node (vector) identified by the node ID "N25". Shown. In the example of FIG. 6, it is shown that the node (vector) identified by the node ID “N1” can be traced to the node (vector) identified by the node ID “N25”. That is, it is shown that the node N1 can be traced to the node N25 by the edge E11 which is a normal edge.

Further, the node identified by the node ID "N1" indicates that the special edge is not connected to another node. In the example of FIG. 6, it is shown that there is no special edge whose reference source is the node (vector) identified by the node ID “N1”. That is, it is shown that there is no node that can be traced by the special edge from the node N1.

Further, from the node identified by the node ID "N35", the edge (normal edge) identified by the edge ID "E351" is connected to the node (vector) identified by the node ID "N693". Shown. In the example of FIG. 6, it is shown that the node (vector) identified by the node ID “N35” can be traced to the node (vector) identified by the node ID “N693”. That is, it is shown that the node N65 can be traced to the node N693 by the edge E351 which is a normal edge.

Further, from the node identified by the node ID "N35", it is shown that the special edge identified by the edge ID "SE35" is connected to the node (vector) identified by the node ID "N831". In the example of FIG. 6, it is shown that the node (vector) identified by the node ID “N35” can be traced to the node (vector) identified by the node ID “N831”. That is, it is shown that the node N35 can be traced to the node N831 by the special edge SE35.

The graph information storage unit 123 is not limited to the above, and various information may be stored depending on the purpose. For example, the graph information storage unit 123 may store the length of the edge connecting each node (vector). That is, the graph information storage unit 123 may store information indicating the distance between each node (vector).

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

The "model ID" indicates identification information for identifying the model. For example, the model identified by the model ID "M1" corresponds to the model M1 shown in the example of FIG. "Use" indicates the use of the corresponding model. Further, the "model data" indicates the data of the corresponding corresponding model associated with the model data. For example, "model data" includes information including nodes in each layer, functions adopted by each node, connection relationships 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" has a purpose of "feature extraction (product)" and is used for extracting features from input item information. Show that. Further, it is shown that the model data of the model M1 is the model data MDT1.

The model M1 (model data MDT1) includes an input layer into which item information of an item is input, an output layer, and a first element that is any layer from the input layer to the output layer and belongs to a layer other than the output layer. , A second element whose value is calculated based on the first element and the weight of the first element, and each element belonging to each layer other than the output layer with respect to the item information input to the input layer is first. As an element, it is a model for making a computer function so as to output the same information as the information input to the input layer from the output layer by performing an operation based on the first element and the weight of the first element.

Further, it is assumed that the models M1, M2 and the like are realized by a neural network having one or a plurality of 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. Further, the second element corresponds to a node in the next stage, which is a node to which a value is transmitted from a node corresponding to the first element. Further, the weight of the first element corresponds to a connection coefficient which is a 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 and the like are realized by the regression model represented by "y = a1 * x1 + a2 * x2 + ... + ai * xi". In this case, for example, the first element included in the models M21 and M22 corresponds to input data (xi) such as x1 and x2. Further, the weight of the first element corresponds to the coefficient ai corresponding to xi. Here, the regression model can be regarded as a simple perceptron having an input layer and an output layer. When each model is regarded as a simple perceptron, the first element corresponds to any node of the input layer, and the second element can be regarded as the node of the output layer.

The model information storage unit 124 is not limited to the above, and may store various model information depending on the purpose.

(History information storage unit 125)
The history information storage unit 125 according to the embodiment stores various information related to the history. FIG. 8 is a diagram showing an example of the history information storage unit according to the embodiment. The history information storage unit 125 stores information indicating the history of various actions of a plurality of users (user groups). For example, the history information storage unit 125 stores information indicating the date and time when the user performed an action and the target item. In the example of FIG. 8, the history information storage unit 125 stores the date and time when the user purchased the product and the product (purchased product) to be purchased. In the history information storage unit 125 shown in FIG. 8, "history ID", "date and time", "user", "target item # 1 (purchased product # 1)", "target item # 2 (purchased product # 2)" It has items such as.

"Target item # 1 (purchased product # 1)" and "target item # 2 (purchased product # 2)" indicate items that are the target of action, but "target item # 1 (purchased product # 1)" , Not limited to "target item # 2 (purchased product # 2)", only the number of target items may be stored, for example, "target item # 3 (purchased product # 3)", "target item # 4 (purchased product # 4)". # 4) ”etc. may be included. In FIG. 8, for the sake of simplicity, only the user's behavior regarding the purchase of the product is shown, but the user's behavior is not limited to the purchase of the product, but the browsing of item information (product information) and the item (product information). Various actions such as searching for products) shall be included. For example, when the action is browsing item information, information indicating the browsed item is stored in "target item # 1" and "target item # 2" in the history information storage unit 125. Further, for example, when the action is a search for item information, the "target item # 1" and "target item # 2" in the history information storage unit 125 contain information indicating the items extracted together with the search. It will be remembered.

The "history ID" indicates identification information for identifying the history. "Date and time" indicates the date and time when the corresponding action was performed. In the example of FIG. 8, the date and time are indicated by an abstract code such as "date and time # 11", but the corresponding action was performed such as "20:45:38 on March 13, 2018". It may be a specific date and time. “User” refers to a user who has performed the corresponding action.

"Target item # 1 (purchased product # 1)", "target item # 2 (purchased product # 2)", etc. indicate items that are commonly targeted in the corresponding action. In the example of FIG. 8, "target item # 1 (purchased product # 1)", "target item # 2 (purchased product # 2)", and the like indicate items (products) purchased together in the corresponding action.

In the example shown in FIG. 8, the history identified by the history ID “LG11” indicates that the user U5 purchases the product at the date and time # 11. Further, at the date and time # 11, the user U5 indicates that the item # 35 and the item # 831 have been purchased together.

The history information storage unit 125 is not limited to the above, and may store various information 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 device 100 by, for example, a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). It is realized by executing various programs (corresponding to an example of an information processing program) using the RAM as a work area. Further, the control unit 130 is a controller, and is realized by, for example, an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). The control unit 130 has an input layer, an output layer, and an input layer to an output layer into which item information of an item is input by information processing according to models M1, M2, etc. stored in the model information storage unit 124. It includes a first element which is any layer and belongs to a layer other than the output layer, and a second element whose value is calculated based on the weights of the first element and the first element, and is input to the input layer. Information similar to the information input to the input layer by performing an operation based on the weights of the first element and 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 functions and operations of information processing described below. To realize or execute. The internal configuration of the control unit 130 is not limited to the configuration shown in FIG. 3, and may be another configuration as long as it is a configuration for performing information processing described later.

(Acquisition unit 131)
The 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 item information storage unit 121, the index information storage unit 122, the graph information storage unit 123, the model information storage unit 124, the history information storage unit 125, and the like. In addition, the acquisition unit 131 acquires various types of information from an external information processing device. The acquisition unit 131 acquires various information from the terminal device 10 and the information providing device 50.

The acquisition unit 131 is graph information in which a plurality of nodes corresponding to each of the plurality of items are connected by edges according to the similarity of the plurality of items, and is a viewpoint different from the similarity among the plurality of nodes. Acquires graph information in which nodes that satisfy the predetermined conditions are connected by a special edge different from the edge. The acquisition unit 131 acquires item information related to one item. The acquisition unit 131 acquires graph information in which a plurality of vectors corresponding to each of the plurality of items are concatenated according to the similarity of the plurality of items, and item information relating to one item. The acquisition unit 131 acquires graph information in which a plurality of vectors indicating the characteristics of each of the plurality of items are concatenated according to their similarity.

The acquisition unit 131 acquires graph information in which a plurality of vectors whose elements are feature quantities extracted from each of a plurality of items using a predetermined model are concatenated according to the similarity. By inputting information about a plurality of items into a predetermined model, the acquisition unit 131 is a graph information in which a plurality of vectors whose elements are the feature amounts of the plurality of items to be extracted are concatenated according to the similarity. To get.

The acquisition unit 131 acquires graph information in which nodes having a predetermined commonality in the behavior of the user group among a plurality of items are connected by a special edge. The acquisition unit 131 acquires graph information in which nodes having a predetermined commonality in the behavior of the user group among a plurality of items are connected by a special edge. The acquisition unit 131 acquires graph information in which a plurality of nodes corresponding to a plurality of items, which are a plurality of products whose number of purchases in a predetermined period is equal to or greater than a predetermined threshold value, are connected by a special edge. To do. The acquisition unit 131 acquires graph information in which nodes that reach in common at the time of a search of a predetermined threshold value or more among a plurality of items are connected by a special edge.

For example, the acquisition unit 131 acquires a plurality of nodes (vectors) that are items for data search. For example, the acquisition unit 131 acquires a plurality of nodes and a group of directed edges including 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 the 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 the index information of the tree structure type. In the example of FIG. 1, the acquisition unit 131 acquires the index information IND11.

In addition, the acquisition unit 131 acquires item information related to one item from the terminal device 10 used by the user. For example, the acquisition unit 131 acquires the item information of one item as a search query. For example, the acquisition unit 131 acquires a search query related to an item. The acquisition unit 131 acquires the item information IDT11 related to the item X from the terminal device 10 used by the user U1.

For example, the acquisition unit 131 acquires item information related to one item. For example, the acquisition unit 131 acquires the item information IDT11 related to the item X from the terminal device 10. For example, the acquisition unit 131 acquires the graph information GR11 regarding the item from the graph information storage unit 123. Further, for example, the acquisition unit 131 acquires the index information IND11 used for determining the node (starting point vector) that is the starting point of the search in the graph information GR 11 from the index information storage unit 122 (see FIG. 5).

(Generator 132)
The generation unit 132 generates various information. For example, the generation unit 132 uses the learning data (item information) stored in the item information storage unit 121 to generate a model as shown in the model information storage unit 124. For example, the generation unit 132 generates a model (autoencoder) that outputs the same information as the input item information based on the learning data acquired by the acquisition unit 131. For example, the generation unit 132 generates a model (autoencoder) that outputs the same information as the input item information, using the input item information itself as the correct answer 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. The generation unit 132 may generate the model M1 using any learning algorithm. For example, the generation unit 132 generates the model M1 by using a learning algorithm such as a neural network. As an example, when the generation unit 132 generates a model M1 or the like using a neural network, the model M1 or the like includes an input layer containing one or more neurons, an intermediate layer containing one or more neurons, and one or more neurons. It has an output layer including.

The generation unit 132 includes an input layer into which item information of an item is input, an output layer, a first element which is any layer from the input layer to the output layer and belongs to a layer other than the output layer, and a first element. A second element whose value is calculated based on the weight of the first element and the second element, and each element belonging to each layer other than the output layer is set as the first element with respect to the item information input to the input layer. By performing an operation based on the weights of one element and the first element, a model that outputs the same information as the information input to the input layer from the output layer is generated.

For example, the generation unit 132 generates a model based on the training data. For example, the generation unit 132 generates a model by learning using the item information IDT1, IDT2, etc. in the item information storage unit 121 as learning data (teacher data).

For example, the generation unit 132 performs a learning process so that when the item information IDT1 is input, the model M1 outputs the same information as the item information IDT1. For example, the generation unit 132 performs a learning process so that when the item information IDT2 is input, the model M1 outputs the same information as the item information IDT2.

The generation unit 132 is not limited to the model M1 as an autoencoder, and may generate a model M21 corresponding to an item (entrance), a model M22 corresponding to an item (entry), and the like by using various learning algorithms. .. For example, the generation unit 132 generates models M21, M22, and the like using learning algorithms such as a neural network, a support vector machine (SVM), clustering, and reinforcement learning. As an example, when the generation unit 132 generates models M21, M22, etc. using a neural network, the models M21, M22, etc. include an input layer containing one or more neurons and an intermediate layer containing one or more neurons. It has an output layer containing the above neurons. The information processing device 100 does not have to have the generation unit 132 when acquiring a model from another external device such as the information providing device 50.

In the example of FIG. 1, the generation unit 132 generates an N-dimensional vector from the item information of each item by using a model that extracts the features of the item information. The generation unit 132 generates a vector from the item information of each item by using the model M1 shown in the model information storage unit 124 (FIG. 7).

Further, the generation unit 132 generates a vector used for searching graph information from the item information corresponding to one item. In the example of FIG. 1, the generation unit 132 generates the vector corresponding to the item X by the processing as shown in the processing group PS11. The generation unit 132 inputs the item information IDT11 regarding the item X into the model M1. Specifically, the generation unit 132 inputs the item information IDT11 including information such as the product category, title, description, image information, manufacturer, and size of the item X into the model M1. Then, the generation unit 132 generates a vector by using the information in the model M1 after the input of the item information IDT11. For example, the generation unit 132 generates vector data using each element in the model M1 in which the item information IDT11 is input.

In the example of FIG. 1, the generation unit 132 generates the vector VD11 by using the value of each element in the model M1 in which the item information IDT11 is input. For example, the generation unit 132 uses the value VE1 (see FIG. 11) corresponding to the neuron NL1 of the model M1 and the value VE2 (see FIG. 11) corresponding to the neuron NL2 when the item information IDT11 of the item X is input. To generate a vector. For example, the generation unit 132 may extract the value VE1 corresponding to the calculated neuron NL1 as the first-dimensional element of the vector VD11 when the item information IDT11 of the item X is input. Further, for example, the generation unit 132 generates the vector VD11 by using the value VE2 corresponding to the calculated neuron NL2 as the second-dimensional element of the vector VD11 when the item information of the item is input. In the example of FIG. 1, the generation unit 132 generates the vector VD11 such that the element of the first dimension is “35” and the element of the second dimension is “63”.

(Decision unit 133)
The determination unit 133 determines various 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 point vector based on the item information acquired by the acquisition unit 131 and the information regarding the starting point vector that is the starting point for searching the graph information. For example, the determination unit 133 determines the starting point vector based on the item information and the information that serves as a reference for determining the starting point vector for searching the graph information. Further, the determination unit 133 determines the starting point vector based on the index information used for determining the starting point vector. Further, the determination unit 133 determines the starting point vector based on the index information of the tree structure type.

In the example of FIG. 1, the determination unit 133 uses the index information IND11 to determine (specify) the origin vector corresponding to one item (query). For example, the determination unit 133 uses the index information IND11 to determine (specify) the starting point vector corresponding to the vector VD11 of the item X. The determination unit 133 determines the starting point vector in the graph information GR 11 by using the vector VD11 and the index information IND11.

For example, the determination unit 133 determines the starting point vector in the graph information GR 11 by using the tree-structured index information as shown in the index information IND11 in FIG. In the example of FIG. 1, after generating the vector VD11, the determination unit 133 efficiently searches by tracing the index information IND11 from top to bottom to identify a starting point vector that is a candidate near the index information IND11. Determine the origin vector corresponding to the query (one item).

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

In the example of FIG. 1, when the search unit 134 determines that the determination unit 133 has a special edge connecting the normal extraction node and another object (special object) with the normal extraction node as a reference source, the special Decide to extract the object.

(Search unit 134)
The search unit 134 searches for various types of information. For example, the search unit 134 searches for various types of 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 extracts various information using the information acquired by the acquisition unit 131. For example, the search unit 134 performs various determination processes. For example, the search unit 134 determines various types of information using the information acquired by the acquisition unit 131. For example, the search unit 134 searches for various information stored in the storage unit 120. For example, the acquisition unit 131 searches for various information stored in the item information storage unit 121, the index information storage unit 122, the graph information storage unit 123, the model information storage unit 124, the history information storage unit 125, and the like. For example, the search unit 134 performs a calculation process using various information. For example, the search unit 134 calculates the distance between the vectors using the information about the vectors. For example, the search unit 134 may use information on the length (distance) of the edge connecting each node (vector) stored in the graph information storage unit 123, or each node (from the vector information of each node). Information on the length (distance) of the edges connecting the vectors) may be calculated, and the calculated length (distance) information may be used.

The search unit 134 searches for graph information from a plurality of nodes of the graph information acquired by the acquisition unit 131, starting from the starting node that is the starting point for searching the graph information determined based on a predetermined criterion. Extracts the target item, which is an item similar to one item, out of a plurality of items. The search unit 134 extracts the target item starting from the starting point node determined by the determining unit 133. When the search unit 134 reaches the node to which the special edge is connected at the time of searching the graph information, the search unit 134 extracts the node and another node connected by the special edge as the target item.

The search unit 134 searches the graph information starting from the starting point vector that is the starting point for searching the graph information determined based on a predetermined criterion among the plurality of vectors of the graph information acquired by the acquisition unit 131. , Extract the target item which is an item similar to one item from a plurality of items. The search unit 134 extracts various information. For example, the search unit 134 searches the graph information with the starting point vector determined by the determining unit 133 as the starting point, and extracts similar items that are similar to one item among a plurality of items as target items. ..

In the example of FIG. 1, the search unit 134 searches for the target item. For example, the search unit 134 searches for the target item related to the item X. For example, the search unit 134 may search for the target item of item X by appropriately using various conventional techniques such as a neighborhood search technique disclosed in Non-Patent Document 1. For example, the search unit 134 uses the graph information GR11 in which nodes having high similarity are connected by an edge.

In the example of FIG. 1, the search unit 134 searches for the target item of the item X by using the graph information GR11 and the index information IND11 as shown in the information group INF11. The search unit 134 extracts the node N451 and the node N35 as target items by searching the graph information GR11 starting from the node N451.

When a node (normal extraction node) is extracted by tracing an edge (normal edge) other than the special edge in the graph information GR11, the search unit 134 determines whether or not the special edge is connected to another edge from the normal extraction node. judge. For example, the search unit 134 determines whether or not there is a special edge that connects the normal extraction node and other objects with the normal extraction node as a reference source. Then, the search unit 134 determines that there is a special edge that connects the normal extraction node and another object (special object) with the normal extraction node as the reference source.

When the node N451 is extracted as the target item, the search unit 134 determines whether or not a special edge is connected to another edge from the node N451. In the example of FIG. 1, when the node N451 which is the starting node is extracted, the search unit 134 determines whether or not the special edge is connected to another edge from the node N451. For example, the search unit 134 determines that there are no special edges from the node N451, that is, there are no special edges from the node N451 to other edges. Then, the search unit 134 determines that the special object related to the node N451 is not extracted, and performs a search at the normal edge.

When the node N35 is extracted as the target item, the search unit 134 determines whether or not a special edge is connected to another edge from the node N35. In the example of FIG. 1, when the node N35 is extracted by tracing the normal edge from the node N451, the search unit 134 determines whether or not the special edge is connected to another edge from the node N35. For example, the search unit 134 determines that there is a special edge from the node N35 to another edge because the special edge SE35 from the node N35 to the node N831 is connected. Therefore, the search unit 134 extracts the node N831 to which the special edge SE35 is connected from the node N35 as the target item.

Further, the search unit 134 does not have to search for the node that can be traced from the node extracted as the target item by concatenating the special edges. In the example of FIG. 1, the search unit 134 does not have to search for the node that can be traced from the node N831 extracted as the target item by connecting the special edges. In this case, the search unit 134 performs a search by tracing a normal edge from the node N35. The search unit 134 may extract a node that can be traced from the node N831 as a target item.

(Providing section 135)
The providing unit 135 provides various types of 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 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 target item extracted by the search unit 134. In addition, the providing unit 135 provides an information providing service regarding the target item. The providing unit 135 provides the terminal device 10 with information regarding the target item.

The providing unit 135 provides a predetermined service based on the target item extracted by the search unit 134. The providing unit 135 provides a predetermined service to the terminal device 10. The providing unit 135 provides an information providing service regarding the target item.

For example, the provider 135 provides the object ID corresponding to the query as a search result. For example, the providing unit 135 provides the object ID selected by the search unit 134 to the information providing device 50. The providing unit 135 provides the information providing device 50 with the object ID selected by the search unit 134 as information indicating a 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 is a terminal device used by the user U1 with the item # 451 corresponding to the node N451, the item # 35 corresponding to the node N35, and the item # 831 corresponding to the node N831 as target items. Provided to 10.

[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. 9 is a flowchart showing an example of information processing according to the embodiment.

As shown in FIG. 9, the information processing apparatus 100 acquires item information related to one item (step S101). In the example of FIG. 1, the information processing device 100 acquires the item information IDT11 related to the item X from the terminal device 10.

The information processing device 100 acquires graph information related to a plurality of items (step S102). For example, the information processing apparatus 100 acquires graph information GR11 regarding an item from the graph information storage unit 123. In the example of FIG. 1, the information processing apparatus 100 acquires the graph information GR 11 including the special edge SE 35 from the graph information storage unit 123.

Then, the information processing apparatus 100 generates a vector from the item information about one item by using the model (step S103). In the example of FIG. 1, the information processing apparatus 100 generates the vector VD11 from the item information IDT11 by using the model M1 stored in the model information storage unit 124.

Then, the information processing apparatus 100 determines the starting point vector using the generated vector and the index information (step S104). In the example of FIG. 1, the information processing apparatus 100 determines the starting point vector at the node N451 by using the vector VD11 and the index information IND11 stored in the index information storage unit 122.

Then, the information processing apparatus 100 extracts similar items of one item as target items by searching the graph information, and when the node reaches the node to which the extended edges are connected at the time of searching the graph information, the information processing device 100 and the node are connected. Other nodes connected by the extended edge are extracted as target items (step S105). For example, the information processing apparatus 100 extracts an item similar to one item as a target item by searching the graph information. In the example of FIG. 1, when the node N35 is extracted, the information processing apparatus 100 extracts the node N831 which is the reference destination of the special edge SE35 whose reference source is the node N35 as the target item.

Then, the information processing device 100 provides a service based on the target item (step S106). In the example of FIG. 1, the information processing device 100 provides information indicating items # 451, item # 35, and item # 831 extracted as target items to the terminal device 10 used by the user U1.

[5. Generation process flow]
Next, the procedure of the generation process by the information processing system 1 according to the embodiment will be described with reference to FIG. FIG. 10 is a flowchart showing an example of the generation process according to the embodiment.

As shown in FIG. 10, the information processing apparatus 100 acquires the learning data (step S201). For example, the information processing device 100 acquires learning data from the item information storage unit 121. For example, the information processing device 100 acquires item information IDT1, IDT2, and the like as learning data from the item information storage unit 121.

After that, the information processing apparatus 100 generates a model based on the training data (step S202). For example, the information processing device 100 generates the model M1 from the item information storage unit 121 using the learning data. For example, the information processing apparatus 100 generates the model M1 so as to output the same information (item information) as the information (item information) input to the input layer from the output layer. For example, the information processing apparatus 100 generates a model M1 as an autoencoder that inputs item information of an item.

[6. Search example]
Here, an example of a search using the graph information described above is 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. 12 as an example. FIG. 12 is a flowchart showing an example of a search process using graph data (graph information). The search process described below is performed by, for example, the search unit 134 of the information processing apparatus 100. In addition, the objects referred to below may be read as vectors and nodes. In the following description, it is assumed that the information processing device 100 performs the search process, but the search process may be performed by another device. For example, the information processing apparatus 100 uses vector data generated from the item information of one item as a search query. For example, the information processing apparatus 100 searches graph data starting from a starting point vector determined based on vector data generated from item information of one item and index information. In the example of FIG. 1, the information processing apparatus 100 searches the graph information GR11 starting from the node N451 which is the starting point vector determined based on the vector VD11 of the item X and the index information IND11.

Here, the neighborhood object set N (G, y) is a set of neighborhood objects associated with the edge assigned to the node y. “G” may be predetermined graph data (for example, graph information GR11 or the like). For example, the information processing apparatus 100 executes the k-nearest neighbor search process.

For example, the information processing apparatus 100 sets the radius r of the hypersphere to ∞ (infinity) (step S300) and extracts a subset S from the existing object set (step S301). For example, the information processing apparatus 100 may extract an object (node) selected as a root node (starting point vector) as a subset S. In the example of FIG. 1, the information processing apparatus 100 may extract the node N451 or the like, which is a starting point vector, as a subset S. Further, for example, a hypersphere is a virtual sphere indicating 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 result at the same time.

Next, the information processing apparatus 100 extracts the object having the shortest distance from the object y from the objects included in the object set S, where y is the search query object, and sets it as the object s (step S302). In the example of FIG. 1, the information processing apparatus 100 extracts the object having the shortest distance from the vector VD11, which is the search query object, among the objects included in the object set S, and sets them as objects 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 uses it as the object s. For example, in the information processing apparatus 100, when only the object (node) selected as the root node (starting point vector) is an element of the object set S, the root node (starting point vector) is extracted as the object s as a result. .. Next, the information processing apparatus 100 excludes the objects 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 (exceeds) r (1 + ε) (step S304). Here, ε is an extension element, and r (1 + ε) is a value indicating the radius of the search range (searching only the nodes within this range. The accuracy can be improved by making it larger than the search range). is there. When the distance d (s, y) between the object s and the object y exceeds r (1 + ε) (step S304: Yes), the information processing apparatus 100 outputs the object set R as a set of objects in the vicinity of the object y (step S304: Yes). Step S305), the process is terminated.

When 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 determines the neighboring object set N (G, s) of the object s. One object that is not included in the object set C is selected from the objects that are the elements of the above, 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 search, 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). When the distance d (u, y) between the object u and the object y is r (1 + ε) or less (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 or equal to r (1 + ε) (step S307: No), the information processing apparatus 100 determines (processes) 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 r or less (step S309). When the distance d (u, y) between the object u and the object y is r (step S309: No), the information processing apparatus 100 determines (processes) step S315. When the distance d (u, y) between the object u and the object y is not r or less (step S309: No), the information processing apparatus 100 determines (processes) step S315.

When the distance d (u, y) between the object u and the object y is r or less (step S309: Yes), the information processing apparatus 100 adds the object u to the object set R (step S310).

Here, after step S310, the information processing apparatus 100 determines whether or not there is a special edge that connects the object u and another object with the object u as a reference source. When the information processing apparatus 100 determines that there is a special edge that connects the object u and another object (special object) with the object u as a reference source, it determines that the special object is extracted. The information processing device 100 may add a special object to the object set R, or may store information on the special object separately from the object set R. Further, the information processing apparatus 100 does not have to use the number of special objects for comparison with ks described later. For example, when the information processing apparatus 100 extracts a special object, the number of special objects may be added to the number of objects, or the number of special objects may not be added to the number of objects. For example, the information processing apparatus 100 may perform the comparison process with ks in steps S311 and S313 for the number of normal objects included in the object set R excluding the special object.

For example, when the information processing apparatus 100 adds a special object to the object set R, the information processing device 100 may manage the special object so that it can be distinguished from other objects (normal objects) by giving a specific flag or the like to the special object. Good. In this case, the information processing apparatus 100 may use the number of objects (normal objects) other than the special objects included in the object set R as the number of objects included in the object set R. Further, the information processing apparatus 100 does not have to include the special object included in the object set R in the target object in step S312 or step S314. For example, the information processing apparatus 100 may perform the processing of steps S312 and S314 only on the normal objects included in the object set R excluding the special objects.

Further, the information processing apparatus 100 may add a special object to a set (special set) different from the object set R. In this case, the information processing apparatus 100 may output both the object included in the object set R and the object (special object) included in the special set as the search result in step S317.

The information processing device 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 determines (processes) step S313.

When the number of objects included in the object set R exceeds ks (step S311: Yes), the information processing apparatus 100 selects the object having the longest (far) distance from the object y among the objects included in the object set R. Exclude 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). When the number of objects included in the object set R does not match ks (step S313: No), the information processing apparatus 100 determines (processes) step S315. When the number of objects included in the object set R matches ks (step S313: Yes), the information processing apparatus 100 has the longest distance (far) from the object y among the objects included in the object set R. The distance between the object and the object y is set to a new r (step S314).

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

When all the objects are selected from the objects that are the elements of the object set N (G, s) in the vicinity of the objects s and stored in the object set C (step S315: Yes), the information processing apparatus 100 sets the object set S. Determines whether or not 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. When 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 device 100 may provide an object (node) included in the object set R to the terminal device 10 or the like that has performed the search as a search result corresponding to the search query (input object y). In the example of FIG. 1, the information processing device 100 provides the node N451 and the node N35 included in the object set R to the searched terminal device 10 and the like as the search result corresponding to the search query (vector VD11 of the item X). You may. For example, the information processing device 100 provides the terminal device 10 used by the user U1 with the item # 451 corresponding to the node N451, the item # 35 corresponding to the node N35, and the item # 1 corresponding to the node N1 as target items. To do.

[7. effect〕
As described above, the information processing apparatus 100 according to the embodiment includes an acquisition unit 131 and a search unit 134. The acquisition unit 131 is graph information in which a plurality of nodes corresponding to each of the plurality of items are connected by edges according to the similarity of the plurality of items, and is a viewpoint different from the similarity among the plurality of nodes. The graph information in which the nodes satisfying the predetermined conditions are connected by a special edge different from the edge and the item information related to one item are acquired. The search unit 134 searches for graph information from a plurality of nodes of the graph information acquired by the acquisition unit 131, starting from the starting node that is the starting point for searching the graph information determined based on a predetermined criterion. When a similar item, which is an item similar to one item, is extracted as a target item from a plurality of items and a node to which a special edge is connected is reached at the time of searching graph information, the node and the special edge are used. Extract other concatenated nodes as target items.

As described above, the information processing apparatus 100 according to the embodiment appropriately extracts the target item including a similar item of one item from the plurality of items by searching the graph information with the starting point vector as the starting point. Can be done.

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 the target item starting from the starting point node determined by the determining unit 133.

As described above, the information processing apparatus 100 according to the embodiment can appropriately extract items by determining the starting point vector based on the index information used for determining the starting point vector.

Further, in the information processing apparatus 100 according to the embodiment, the determination unit 133 determines the starting point vector based on the index information of the tree structure type.

As described above, the information processing apparatus 100 according to the embodiment can appropriately extract items by determining the starting point vector based on the index information of the tree structure type.

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

As described above, the information processing apparatus 100 according to the embodiment appropriately extracts items by acquiring graph information in which a plurality of vectors indicating the characteristics of each of the plurality of items are concatenated according to their similarity. Can be done.

Further, in the information processing apparatus 100 according to the embodiment, in the acquisition unit 131, a plurality of vectors having feature quantities extracted from each of the plurality of items using a predetermined model are connected according to the similarity. Get graph information.

As described above, in the information processing apparatus 100 according to the embodiment, graph information in which a plurality of vectors whose elements are feature quantities extracted from each of a plurality of items using a predetermined model are concatenated according to their similarity. Items can be properly extracted by acquiring.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 inputs information about a plurality of items into a predetermined model, and a plurality of vectors having the feature amounts of the plurality of items to be extracted as elements. Gets the graph information concatenated according to the similarity.

As described above, in the information processing apparatus 100 according to the embodiment, by inputting information about a plurality of items into a predetermined model, a plurality of vectors having features of each of the plurality of items extracted are similar. Items can be appropriately extracted by acquiring graph information concatenated according to gender.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 acquires graph information in which nodes having a predetermined commonality in the behavior of the user group among a plurality of items are connected by a special edge.

As described above, the information processing apparatus 100 according to the embodiment obtains graph information in which nodes having a predetermined commonality in the behavior of a group of users are connected by a special edge among a plurality of items to obtain the item. It can be extracted appropriately.

Further, in the information processing apparatus 100 according to the embodiment, the acquisition unit 131 acquires graph information in which nodes having a predetermined commonality in the behavior of the user group among a plurality of items are connected by a special edge.

As described above, the information processing apparatus 100 according to the embodiment obtains graph information in which nodes having a predetermined commonality in the behavior of a group of users are connected by a special edge among a plurality of items to obtain the item. It can be extracted appropriately.

Further, in the information processing device 100 according to the embodiment, the acquisition unit 131 is located between a plurality of nodes corresponding to a plurality of items which are a plurality of products whose number of purchases in a predetermined period is equal to or greater than a predetermined threshold value. Gets the graph information connected by the special edge.

As described above, the information processing apparatus 100 according to the embodiment is specially provided between a plurality of nodes corresponding to a plurality of items which are a plurality of products whose number of purchases in a predetermined period is equal to or greater than a predetermined threshold value. Items can be appropriately extracted by acquiring the graph information connected by the edges.

Further, the information processing device 100 according to the embodiment has a providing unit 135. The providing unit 135 provides a predetermined service based on the target item extracted by the search unit 134.

As described above, the information processing apparatus 100 according to the embodiment can appropriately provide a service using information on similar items by providing a predetermined service based on the extracted target item.

Further, in the information processing device 100 according to the embodiment, the acquisition unit 131 acquires item information related to one item from the terminal device used by the user. The providing unit 135 provides a predetermined service to the terminal device 10.

As described above, the information processing device 100 according to the embodiment acquires the item information about one item from the terminal device 10 used by the user, and provides the terminal device 10 with a predetermined service to provide information about similar items. It is possible to appropriately provide services using.

Further, in the information processing apparatus 100 according to the embodiment, the providing unit 135 provides an information providing service regarding the target item.

As described above, the information processing apparatus 100 according to the embodiment can appropriately provide the service using the information about the similar item by providing the information providing service about the target item.

[8. Hardware configuration]
The information processing apparatus 100 according to the above-described embodiment is realized by, for example, a computer 1000 having a configuration as shown in FIG. FIG. 14 is a hardware configuration diagram showing an example of a computer that realizes the functions of the information processing device. The computer 1000 includes a CPU 1100, a RAM 1200, a ROM (Read Only Memory) 1300, an HDD (Hard Disk Drive) 1400, a communication interface (I / F) 1500, an input / output interface (I / F) 1600, and a media interface (I / F). ) Has 1700.

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

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

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

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

For example, when the computer 1000 functions as the 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. The function of the control unit 130 is realized. The CPU 1100 of the computer 1000 reads and executes these programs or data (for example, model M1 (model data MDT1)) from the recording medium 1800, but as another example, these programs from another device via the network N. May be obtained.

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

[9. Others]
Further, among the processes described in the above-described embodiment, all or a part of the processes described as being automatically performed can be manually performed, or the processes described as being manually performed can be performed. All or part of it can be done automatically by a known method. In addition, the processing procedure, specific name, and information including various data and parameters shown in the above document and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the illustrated information.

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

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

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

1 Information processing system 100 Information processing device 121 Item information storage unit 122 Index information storage unit 123 Graph information storage unit 124 Model information storage unit 125 History information storage unit 130 Control unit 131 Acquisition unit 132 Generation unit 133 Decision unit 134 Search unit 135 Provided Department 10 Terminal device 50 Information providing device N network

Claims (15)

A plurality of nodes corresponding to each of the plurality of items are graph information connected by edges according to the similarity of the plurality of items, and among the plurality of nodes, a predetermined viewpoint different from the similarity is determined. Graph information in which nodes satisfying the above conditions are connected by a special edge different from the edge, an acquisition unit for acquiring item information related to one item, and an acquisition unit.
By searching the graph information from the plurality of nodes of the graph information acquired by the acquisition unit, starting from the starting node that is the starting point for searching the graph information determined based on a predetermined criterion. When a similar item, which is an item similar to the one item, is extracted as a target item from the plurality of items, and the node reaches the node to which the special edge is connected at the time of searching the graph information, the node and the node are used. A search unit that extracts other nodes connected by the special edge as the target item, and
An information processing device characterized by being equipped with. A determination unit that determines the origin node based on the index information used to determine the origin node.
With more
The search unit
The information processing apparatus according to claim 1, wherein the target item is extracted from the starting point node determined by the determining unit. The decision unit
The information processing apparatus according to claim 2, wherein the starting node is determined based on the index information of the tree structure type. The acquisition unit
The invention according to any one of claims 1 to 3, wherein the 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 items. Information processing device. The acquisition unit
The information processing apparatus according to claim 4, wherein the plurality of vectors showing the characteristics of each of the plurality of items acquire the graph information in which the plurality of vectors are connected according to the similarity. The acquisition unit
4. Claim 4 characterized in that the plurality of vectors having feature quantities extracted from each of the plurality of items using a predetermined model are concatenated according to the similarity to obtain the graph information. Alternatively, the information processing apparatus according to claim 5. The acquisition unit
By inputting information about the plurality of items into a predetermined model, the graph information in which the plurality of vectors having the feature amounts of the plurality of items extracted as elements are concatenated according to the similarity is obtained. The information processing apparatus according to any one of claims 4 to 6, wherein the information processing apparatus is to be acquired. The acquisition unit
The invention according to any one of claims 1 to 7, wherein among the plurality of items, nodes having a predetermined commonality in the behavior of the user group acquire graph information connected by the special edge. Information processing equipment. The acquisition unit
In the behavior of the user group in a predetermined period, graph information in which a plurality of nodes corresponding to a plurality of items whose co-occurrence as a target of the behavior is equal to or more than a predetermined threshold value are connected by the special edge is displayed. The information processing apparatus according to claim 8, wherein the information processing apparatus is to be acquired. The acquisition unit
Acquiring graph information in which a plurality of nodes corresponding to a plurality of items, which are a plurality of products whose number of purchases in the predetermined period is equal to or greater than a predetermined threshold value, are connected by the special edge. The information processing apparatus according to claim 9. A provider that provides a predetermined service based on the target item extracted by the search unit,
The information processing apparatus according to any one of claims 1 to 10, further comprising. The acquisition unit
The item information about the one item is acquired from the terminal device used by the user, and the item information is acquired.
The providing part
The information processing device according to claim 11, wherein the predetermined service is provided to the terminal device. The providing part
The information processing device according to claim 11 or 12, wherein an information providing service is provided based on the target item. It is an information processing method executed by a computer.
A plurality of nodes corresponding to each of the plurality of items are graph information connected by edges according to the similarity of the plurality of items, and among the plurality of nodes, a predetermined viewpoint different from the similarity is determined. Graph information in which nodes satisfying the above conditions are connected by a special edge different from the edge, and an acquisition process for acquiring item information related to one item.
By searching the graph information from the plurality of nodes of the graph information acquired in the acquisition step, starting from the starting node that is the starting point for searching the graph information determined based on a predetermined criterion. When a similar item, which is an item similar to the one item, is extracted as a target item from the plurality of items, and the node reaches the node to which the special edge is connected at the time of searching the graph information, the node and the node are used. A search process for extracting other nodes connected by the special edge as the target item, and
An information processing method characterized by including. A plurality of nodes corresponding to each of the plurality of items are graph information connected by edges according to the similarity of the plurality of items, and among the plurality of nodes, a predetermined viewpoint different from the similarity is determined. Graph information in which nodes satisfying the above conditions are connected by a special edge different from the edge, an acquisition procedure for acquiring item information related to one item, and an acquisition procedure.
By searching the graph information from the plurality of nodes of the graph information acquired by the acquisition procedure, starting from the starting node that is the starting point for searching the graph information determined based on a predetermined criterion. When a similar item, which is an item similar to the one item, is extracted as a target item from the plurality of items, and the node reaches the node to which the special edge is connected at the time of searching the graph information, the node and the node are used. A search procedure for extracting other nodes connected by the special edge as the target item, and
An information processing program characterized by having a computer execute.

Images