Object recognition method and device, equipment, medium and product

Abstract

The invention provides an object identification method, which can be applied to the fields of artificial intelligence and big data, and comprises the following steps: converting transaction data into graph data wherein nodes of the graph data are transaction objects contained in the transaction data and edges of the nodes are transaction relationships contained in the transaction data; determining an initial community containing a target node based on the graph data; determining a core node set in the initial community and an attribute initial weight of the core node set; the quality index of the initial community is determined, the quality index comprises community density and community attribute entropy, the community density is used for representing the density degree of node pairs of the initial community, and the community attribute entropy is used for representing the confusion degree of nodes of the initial community; expanding a core node set of the initial community based on the attribute initial weight, the community density and the community attribute entropy to obtain a local community; and identifying the transaction object corresponding to the core node contained in the local community as a target object. The invention further provides an object recognition device and equipment, a storage medium and a program product.

Description

Technical Field

[0001] The present disclosure relates to the fields of artificial intelligence and big data technology, specifically data mining and financial technology, and more specifically to an object recognition method, device, equipment, medium and product. Background Art

[0002] The long-tail customer group has the characteristics of large size, high potential, and difficulty in maintenance. How to activate or improve the quality of this group has always been the focus of attention of various financial institutions. With the increasing diversification of consumer demand, traditional marketing models often cannot meet the fragmented long-tail customer group's demand for products. Financial institutions promote through general market segmentation methods. For groups with different characteristics hidden in the long tail, it is often difficult to find effective entry points. They can only adopt the marketing strategy of "casting a wide net to catch more fish" and then "selecting the best and following them". This strategy inevitably brings about the problem of high investment costs and the mismatch between benefits and costs.

[0003] At present, for the behavioral characteristics of the long-tail customer group, a variety of precision marketing methods have emerged in the field of data mining, among which graph data has become a highly watched data mining and identification technology. In graph data, a community is a collection of nodes that are tightly linked within the community and sparsely linked to nodes outside the community. Community discovery is divided into full-graph mining and local community discovery. Among them, unlike the general model, the local community discovery method mines groups of different sizes in the market by analyzing a relatively dense set of connected nodes and the unique attributes of the nodes. Through further analysis of the groups, financial institutions can better understand the potential needs of the long-tail customer group. The advantage of this method is that it can capture fragmented and highly personalized consumer behavior and provide more accurate market insights.

[0004] However, the traditional local community discovery method has certain defects. That is, when considering node attributes, it is often measured only from the perspective of similarity, ignoring the performance significance of attribute values ​​in the probability space, resulting in large fluctuations in the quality of local community discovery. In other words, it is difficult to stably and accurately identify long-tail customer groups using traditional local community discovery technology. Summary of the invention

[0005] In view of the above problems, the present disclosure provides an object recognition method, device, equipment, medium and product for improving the accuracy of target object recognition in big data.

[0006] According to a first aspect of the present disclosure, an object identification method is provided, the method comprising: converting transaction data into graph data, wherein the nodes of the graph data are transaction objects contained in the transaction data, and the edges of the nodes are transaction relationships contained in the transaction data; determining an initial community containing a target node based on the graph data; determining a core node set in the initial community and initial attribute weights of the core node set; determining quality indicators of the initial community, the quality indicators comprising community density and community attribute entropy, the community density being used to characterize the density of node pairs in the initial community, and the community attribute entropy being used to characterize the degree of disorder of nodes in the initial community; expanding the core node set of the initial community based on the initial attribute weights, the community density, and the community attribute entropy to obtain a local community; and identifying the transaction objects corresponding to the core nodes contained in the local community as target objects.

[0007] According to an embodiment of the present disclosure, a core node set in an initial community is determined, including: determining multiple maximal cliques in the initial community that include a target node; merging the two maximal cliques into a new node set when there are N-1 overlapping nodes between two maximal cliques of size N, and the attribute similarity of the two overlapping nodes is greater than the average attribute similarity of the two maximal cliques at the same time; repeating the above operations until all nodes in the initial community do not meet the merging conditions, and determining the node set with the largest average attribute similarity as the core node set.

[0008] According to an embodiment of the present disclosure, the initial attribute weights of the core node set in the initial community are determined, including: constructing an objective function according to the number of nodes in the core node set, the size of the attribute set, and the values ​​of each core node on each attribute, the objective function representing the sum of the weighted distances based on the attributes between the core nodes in the initial community; and determining the initial attribute weights by minimizing the sum of the weighted attribute distances of the core node set.

[0009] According to an embodiment of the present disclosure, determining a quality index of an initial community includes: determining a community density according to node pairs, neighbor sets of nodes, and the number of neighbors in the initial community.

[0010] According to an embodiment of the present disclosure, determining the quality index of the initial community further includes: determining the community attribute entropy according to the attribute probability of the nodes in the initial community on the attributes.

[0011] According to an embodiment of the present disclosure, the core node set of the initial community is expanded based on the initial attribute weights, community density and community attribute entropy to obtain a local community, including: constructing a community quality function based on the initial attribute weights, community density and community attribute entropy; based on the community quality function, adding neighbor nodes of the core nodes that have a gain in the community quality of the initial community to the initial community to obtain a local community.

[0012] According to an embodiment of the present disclosure, the method further includes: storing the local community in a graph database, and displaying the real-time local community.

[0013] According to an embodiment of the present disclosure, the method further includes: in response to a transaction request, obtaining original transaction data generated by executing the transaction request; performing feature screening on the original transaction data to obtain transaction data; and storing the transaction data in a transaction database.

[0014] According to a second aspect of the present disclosure, there is provided an object recognition device, which includes: a conversion module, which is used to convert transaction data into graph data, wherein the nodes of the graph data are transaction objects contained in the transaction data, and the edges of the nodes are transaction relationships contained in the transaction data; a first determination module, which is used to determine an initial community containing a target node based on the graph data; a second determination module, which is used to determine a core node set in the initial community and an initial attribute weight of the core node set; a third determination module, which is used to determine a quality indicator of the initial community, wherein the quality indicator includes a community density and a community attribute entropy, wherein the community density is used to characterize the density of node pairs in the initial community, and the community attribute entropy is used to characterize the degree of disorder of nodes in the initial community; an expansion module, which is used to expand the core node set of the initial community based on the initial attribute weight, the community density and the community attribute entropy to obtain a local community; and an identification module, which is used to identify the transaction objects corresponding to the core nodes contained in the local community as target objects.

[0015] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0016] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.

[0017] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0019] Figure 1 The application scenario diagram of the object recognition method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown;

[0020] Figure 2 A flowchart of long-tail customer group mining according to the object recognition method of an embodiment of the present disclosure is schematically shown;

[0021] Figure 3 A flowchart of an object recognition method according to an embodiment of the present disclosure is schematically shown;

[0022] Figure 4 A flowchart of determining a core node set in an initial community according to an embodiment of the present disclosure is schematically shown;

[0023] Figure 5 A flowchart for determining initial attribute weights of a core node set in an initial community according to an embodiment of the present disclosure is schematically shown;

[0024] Figure 6 A flowchart for determining a local community according to an embodiment of the present disclosure is schematically shown;

[0025] Figure 7 A detailed flow chart of local community discovery based on attribute entropy weight and topology according to an embodiment of the present disclosure is schematically shown;

[0026] Figure 8 A schematic diagram of a structure of an object recognition device according to an embodiment of the present disclosure is shown; and

[0027] Fig. 9 A block diagram of an electronic device suitable for implementing an object recognition method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0028] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0029] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0030] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0031] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0032] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0033] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating a person's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.

[0034] An embodiment of the present disclosure provides an object identification method, which includes: converting transaction data into graph data, wherein the nodes of the graph data are transaction objects contained in the transaction data, and the edges of the nodes are transaction relationships contained in the transaction data; determining an initial community containing a target node based on the graph data; determining a core node set in the initial community and initial attribute weights of the core node set; determining quality indicators of the initial community, the quality indicators including community density and community attribute entropy, the community density is used to characterize the density of node pairs in the initial community, and the community attribute entropy is used to characterize the degree of disorder of nodes in the initial community; expanding the core node set of the initial community based on the initial attribute weights, the community density, and the community attribute entropy to obtain a local community; and identifying the transaction objects corresponding to the core nodes contained in the local community as target objects.

[0035] Figure 1 The application scenario diagram of the object recognition method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.

[0036] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, a server 105, and a database 106. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, the server 105, and the database 106. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0037] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 and the database 106 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, financial consumption applications, etc. (only for example).

[0038] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0039] The server 105 may be a server that provides various services, such as a backend management server that provides support for transaction requests sent by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103, and processes transaction data to mine target objects (for example only). The backend management server may analyze and process the received data such as user requests, store the processing results in the database 106, and feed back to the terminal device (for example, web pages, information, or data obtained or generated according to user requests, etc.).

[0040] It should be noted that the object recognition method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the object recognition device provided in the embodiment of the present disclosure can generally be set in the server 105. The object recognition method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the object recognition device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0041] It should be understood that Figure 1 The number of terminal devices, networks, servers and databases in the embodiment is only for illustration. Any number of terminal devices, networks, servers and databases may be provided as required.

[0042] The object recognition method provided by the disclosed embodiment can be applied to the field of financial technology. For example, financial institutions use the general market segmentation method for promotion. For groups with different characteristics hidden in the long tail, it is often difficult to find an effective entry point. They can only adopt the marketing strategy of "casting a wide net to catch more fish" and then "selecting the best ones". This strategy inevitably brings about the problem of high investment cost and the problem that the benefits do not match the costs.

[0043] Figure 2 The flowchart of long-tail customer group mining according to the object recognition method of an embodiment of the present disclosure is schematically shown.

[0044] like Figure 2 As shown, the user initiates a transaction request through the first terminal device 101, the second terminal device 102, and the third terminal device 103. The transaction processing center of the server 105 processes the transaction request, marks the generated historical transaction data and stores it in the transaction database, processes the historical transaction data and converts it into graph data, trains based on the graph data to determine the local community model, sends the local community to the detection center, and identifies the current transaction data based on the local community model to find out whether the user corresponding to the current transaction data is the target object.

[0045] By adopting the object recognition method of the embodiment of the present disclosure, it is possible to accurately identify the target object from the transaction data based on the local community determined by a large amount of transaction data, so as to complete precise marketing for the target object.

[0046] It should be noted that the object identification method and device disclosed herein can be used in the mining of long-tail customers in the field of financial technology, and can also be used in any field other than the field of financial technology, for example, e-commerce. The application field of the object identification method and device disclosed herein is not limited.

[0047] The following will be based on Figure 1 The scene described by Figure 2~Figure 6 The object recognition method of the disclosed embodiment is described in detail.

[0048] Figure 3 The flowchart of the object recognition method according to the embodiment of the present disclosure is schematically shown.

[0049] like Figure 3 As shown, the object recognition of this embodiment includes operations S310 to S360, and the object recognition method can be executed by the server 105.

[0050] In operation S310, the transaction data is converted into graph data. The nodes of the graph data are transaction objects included in the transaction data, and the edges of the nodes are transaction relationships included in the transaction data.

[0051] In operation S320 , an initial community including a target node is determined based on the graph data.

[0052] In operation S330 , a core node set in the initial community and initial attribute weights of the core node set are determined.

[0053] In operation S340 , a quality index of the initial community is determined. The quality index includes community density and community attribute entropy. The community density is used to characterize the density of node pairs in the initial community. The community attribute entropy is used to characterize the disorder of nodes in the initial community.

[0054] In operation S350 , the core node set of the initial community is expanded based on the initial attribute weight, the community density, and the community attribute entropy to obtain a local community.

[0055] In operation S360 , a transaction object corresponding to a core node included in the local community is identified as a target object.

[0056] For example, transaction data may include date, location, transaction type, transaction parties (buyer and seller), amount, supplier, transaction device identification, customer behavior patterns, product ID, product category, etc.

[0057] The transaction relationship can be regarded as the edge of the node, and the two parties of the transaction can be regarded as the nodes on both sides of the edge, and the transaction data can be processed into a graph G. ,in is the node set of graph G, with a total of n nodes; is the edge set of graph G, with a total of m edges; is the attribute set of graph G, with a total of d attributes. Is a node The attribute vector of Representation Node In Properties The value on .

[0058] A subgraph C (i.e., initial community) containing the target node can be selected on graph G. The initial community node set is , a total of c nodes; is the edge set of the graph.

[0059] After determining the initial weights of the attributes, the core node set can be expanded to obtain the final local community. Since the contributions of different attributes in discovering communities are not the same and not constant, the weights of the attributes can be determined based on a two-stage attribute weighting strategy. Considering that the core node set of the initial community can represent the final local community to a certain extent, the initial attribute weights can be learned based on the core node set of the community. In the process of community expansion, the community density and community attribute entropy are combined to expand the core node set to obtain the final local community.

[0060] According to the object recognition method of the embodiment of the present disclosure, the abstract transaction data is converted into graph data. Based on the graph data technology, first, the core node set of the target node is mined in combination with "node attributes" and "topological structure". Secondly, the initial attribute weights of the community are learned according to the core node set, and then the core node set is expanded in combination with the community weighted attribute entropy to determine the local community. The community weighted attribute entropy can measure the attribute homogeneity of the community, so as to better and accurately discover the local community. The two-stage attribute weighting strategy based on the initial attribute weight and the community weighted attribute entropy is used to distinguish the weights of different attributes to meet the dynamic changes of the community expansion process, and can cope with the changeable bank network structure, so as to stably and quickly identify the target object. In addition, by improving the local community discovery method, the speed and accuracy of object recognition are accelerated, so as to improve the computing efficiency and execution effect of the server hardware. And the improved local community can cope with the changeable bank network structure, avoid the server from repeatedly executing the local community discovery method, so as to further improve the hardware operation efficiency of the server and improve the hardware processing speed.

[0061] Figure 4 The flowchart of determining the core node set in the initial community according to an embodiment of the present disclosure is schematically shown.

[0062] like Figure 4 As shown, based on the above embodiment, determining the core node set in the initial community may include operations S410 to S430.

[0063] In operation S410 , a plurality of maximal cliques including a target node in an initial community are determined.

[0064] In operation S420, when there are N-1 overlapping nodes between two maximal clusters of size N, and the attribute similarity of the two overlapping nodes is greater than the average attribute similarity of the two maximal clusters, the two maximal clusters are merged into a new node set.

[0065] In operation S430, the above operations S410 and S420 are repeated until all nodes in the initial community do not meet the merging condition, and the node set with the largest average attribute similarity is determined as the core node set.

[0066] A maximal clique refers to a clique in an undirected graph that is not contained in any other clique, that is, it is not a proper subset of any other clique. In this case, the clique is called the maximal clique of the graph G.

[0067] Since the maximal clusters containing the target node may overlap with each other, screening is required. If there are N-1 overlapping nodes between two maximal clusters of size N, and the attribute similarity of the two non-overlapping nodes is greater than the average attribute similarity of the two maximal clusters, the two maximal clusters are merged into a new node set. When all node sets cannot be merged, the node set with the largest average attribute similarity is selected as the core node set. Discrete attribute similarity is calculated using cosine similarity, and continuous attribute similarity is calculated using Euclidean similarity.

[0068] According to the object recognition method of the embodiment of the present disclosure, the "maximum clique" method is used to mine a core node set around the "seed node" instead of a single seed node, thereby avoiding the "seed dependence" problem, and then being able to accurately discover local communities, thereby improving the accuracy of object recognition and enhancing the execution effect of the server hardware.

[0069] Figure 5 The flowchart of determining the initial attribute weights of the core node set in the initial community according to an embodiment of the present disclosure is schematically shown.

[0070] like Figure 5 As shown, based on the above embodiment, determining the initial attribute weights of the core node set in the initial community may include operations S510 to S520.

[0071] In operation S510, an objective function is constructed according to the number of nodes in the core node set, the size of the attribute set, and the value of each core node on each attribute. The objective function represents the sum of the weighted distances based on the attributes between the core nodes in the initial community.

[0072] In operation S520 , the initial attribute weight is determined by minimizing the sum of weighted attribute distances of the core node set.

[0073] For example, given a core node set Z, the initial weight vector of each attribute can be solved by minimizing the sum of the weighted attribute distances of the core node set. The objective function is defined as follows:

[0074]

[0075] Where c is the number of nodes in the core node set, d is the size of the attribute set, Representative attributes The weight of is the weight Parameters, Is a node In Properties The value on , Is a node and nodes In Properties The difference of the upper values ​​is the norm. This objective function describes the sum of the attribute-based weighted distances between nodes in the initial community, and the attribute weights follow the constraints: ,in For the solution of the constrained function extreme value, the Lagrange multiplier method can be used to solve it, and the initial weight finally solved is :

[0076]

[0077] in:

[0078] .

[0079] Through the object recognition method of the embodiment of the present disclosure, since the core node set of the initial community can represent the final local community to a certain extent, the initial attribute weights can be learned according to the core node set of the initial community, and then the local community can be accurately discovered, thereby improving the accuracy of object recognition and improving the execution effect of the server hardware.

[0080] Based on the above embodiment, determining the quality index of the initial community may include: determining the community density according to the node pairs, the neighbor sets of the nodes, and the number of neighbors of the initial community.

[0081] In an embodiment of the present disclosure, the community density of the initial community can measure the structural cohesion of the community. It can be defined as the average density of node pairs in the community. Can be:

[0082]

[0083] in, is a pair of nodes in community C; is the neighbor set of node a; is the number of neighbors of node a; is a neighbor of node a; is a neighbor of node b; As an indicative function, it can be as follows:

[0084]

[0085] Furthermore, the local community density can be calculated as follows:

[0086]

[0087] When using community density When used as an evaluation criterion for community detection, the higher the community density, the closer the node pairs within the community are. This trend is very consistent with the community structure required for community discovery. Therefore, community density can be used to judge the quality of community discovery results.

[0088] According to the object recognition method of the embodiment of the present disclosure, since community density can be used to judge the quality of community discovery results, community density is added in the process of discovering local communities to further accurately discover local communities, thereby improving the accuracy of object recognition and enhancing the execution effect of the server hardware.

[0089] Based on the above embodiment, determining the quality index of the initial community may further include: determining the community attribute entropy according to the attribute probability of the nodes in the initial community.

[0090] In the embodiments of the present disclosure, community attribute entropy is proposed based on the idea that information entropy can measure the degree of disorder of a system.

[0091] Assuming that the attributes in the community are independent of each other, we can first calculate the information entropy value of each attribute in the community, and then add up the weighted entropy of each attribute to get the weighted attribute entropy of the community. The calculation method is as follows:

[0092] Assume any node in the community In Properties The attribute probability on Node In Properties The value can be obtained Obviously, Naturally normalized, community C has the following attributes: The attribute entropy on is defined as:

[0093] .

[0094] When the attributes are independent of each other, the weighted attribute entropy calculation formula of the community is:

[0095] .

[0096] Since the value range of community density in the previous article is [0,1], and the value range of attribute entropy is [0, ], and the two cannot be linearly combined when the value ranges are different. Therefore, the embodiment of the present disclosure uses the inverse tangent normalization method to map the community attribute entropy to [0,1], and finally forms a local community quality index with the community density. The calculation formula of the inverse tangent normalization method is:

[0097] .

[0098] Although community weighted attribute entropy provides a new perspective for considering node attributes based on the probability of attribute values, it also ignores the information of attribute values ​​themselves. Therefore, in the process of community expansion, attribute similarity is used as a basic indicator to make up for the deficiency of community attribute entropy, and the attribute weight adjustment in the second stage is completed accordingly.

[0099] According to the object recognition method of the embodiment of the present invention, the weighted attribute entropy is determined from the probability of node attribute values, and the attribute's own probability value is taken into consideration, which can better measure the homogeneity of community attributes, and then accurately determine the local community, thereby improving the accuracy of object recognition and improving the execution effect of the server hardware.

[0100] Figure 6 The flowchart of determining a local community according to an embodiment of the present disclosure is schematically shown.

[0101] like Figure 6 As shown, based on the above embodiment, determining a local community may include operations S610 to S620.

[0102] In operation S610, a community quality function is constructed according to the initial attribute weights, community density, and community attribute entropy.

[0103] In operation S620, based on the community quality function, neighbor nodes of the core node that have a gain in community quality of the initial community are added to the initial community to obtain a local community.

[0104] In the embodiment of the present disclosure, if most nodes in the community have the attribute If the values ​​are similar, then the attribute It is conducive to community discovery; on the contrary, if the value distribution is very random, Not a very good community attribute. The calculation formula for community attribute weight is as follows:

[0105]

[0106] in, Is a node About Attributes The similarity.

[0107] Figure 7 The detailed flowchart of local community discovery based on attribute entropy weight and topology according to an embodiment of the present disclosure is schematically shown.

[0108] like Figure 7As shown, the community quality gain that can be brought by neighbor nodes joining the community is calculated. The community local quality function calculation function measures the community quality from two perspectives: community density and community weighted attribute entropy. The function formula can be:

[0109]

[0110] in, Parameters for balancing structural information and attribute information; for community density; is the normalized community weighted attribute entropy. Since the method proposed in the present invention includes two hyperparameters α and β that need to be set according to the data set, it is necessary to calculate the influence of different values ​​of α and β on the accuracy and local modularity of the algorithm results, so as to determine the optimal hyperparameter value.

[0111] Traverse the neighbor nodes of the current community and calculate the community quality gain that different neighbor nodes can bring to the community. If adding the neighbor node can bring the community an average quality gain that exceeds the historical joining nodes, then add the node to the community and update the community attribute weight; otherwise, skip the node. The loop is terminated until there are no neighbor nodes to join, and then the local community is determined.

[0112] According to the object recognition method of the embodiment of the present disclosure, based on constructing a community quality function according to the initial weight of the attribute, the community density, and the community attribute entropy, the community quality gain that can be brought about by the neighbor nodes joining the community is calculated, and the local community can be accurately determined, thereby improving the accuracy of object recognition.

[0113] Based on the above embodiment, the method further includes: storing the local community in a graph database, and displaying the real-time local community.

[0114] According to the object recognition method of the embodiment of the present disclosure, after determining the local community, the determined local community model can be stored in the graph database, and the local community model can be directly called to perform the object recognition model later, avoiding repeated determination of the local community model, and can reduce the repeated determination of the local community by the server hardware, thereby reducing the amount of calculation and saving computing resources.

[0115] Based on the above embodiment, the method also includes: in response to a transaction request, obtaining original transaction data generated by executing the transaction request; performing feature screening on the original transaction data to obtain transaction data; and storing the transaction data in a transaction database.

[0116] After the customer initiates a transaction, the transaction data will be transferred to the bank's transaction center for processing. All raw transaction data can be saved in batches in the transaction database at a frequency of T+1. The amount of data actually contained in the raw transaction data is relatively large. Before local community discovery, the raw transaction data can be processed. For example, according to a series of data feature selection rules defined in advance, all meaningful features (such as date, location, transaction type, amount, supplier, transaction device identification, customer behavior pattern, etc.) are retained. Then, graph data is constructed based on these meaningful features.

[0117] According to the object recognition method of the embodiment of the present disclosure, by pre-processing the transaction data, invalid features are removed and saved in the transaction database, and then directly called when determining the local community later, the amount of calculation is reduced and computing resources are saved.

[0118] Based on the above object recognition method, the present disclosure also provides an object recognition device. Figure 8 The device is described in detail.

[0119] Figure 8 The structure block diagram of the object recognition device according to the embodiment of the present disclosure is schematically shown.

[0120] like Figure 8 As shown, the object recognition device 800 of this embodiment includes a conversion module 810 , a first determination module 820 , a second determination module 830 , a third determination module 840 , an expansion module 850 and a recognition module 860 .

[0121] The conversion module 810 is used to convert the transaction data into graph data, wherein the nodes of the graph data are transaction objects contained in the transaction data, and the edges of the nodes are transaction relationships contained in the transaction data. In one embodiment, the conversion module 810 can be used to perform the operation S310 described above, which will not be described in detail here.

[0122] The first determination module 820 is used to determine the initial community including the target node based on the graph data. In one embodiment, the first determination module 820 can be used to perform the operation S320 described above, which will not be described in detail here.

[0123] The second determination module 830 is used to determine the core node set in the initial community and the initial attribute weights of the core node set. In one embodiment, the second determination module 830 can be used to perform the operation S330 described above, which will not be described in detail here.

[0124] The third determination module 840 is used to determine the quality index of the initial community, the quality index includes community density and community attribute entropy, the community density is used to characterize the density of the node pairs of the initial community, and the community attribute entropy is used to characterize the degree of disorder of the nodes of the initial community. In one embodiment, the third determination module 840 can be used to perform the operation S340 described above, which will not be repeated here.

[0125] The expansion module 850 is used to expand the core node set of the initial community based on the initial attribute weight, community density and community attribute entropy to obtain a local community. In one embodiment, the expansion module 850 can be used to perform the operation S350 described above, which will not be repeated here.

[0126] The identification module 860 is used to identify the transaction object corresponding to the core node included in the local community as the target object. In one embodiment, the identification module 860 can be used to perform the operation S360 described above, which will not be repeated here.

[0127] According to an embodiment of the present disclosure, a core node set in an initial community is determined, including: determining multiple maximal cliques in the initial community that include a target node; merging the two maximal cliques into a new node set when there are N-1 overlapping nodes between two maximal cliques of size N, and the attribute similarity of the two overlapping nodes is greater than the average attribute similarity of the two maximal cliques at the same time; repeating the above operations until all nodes in the initial community do not meet the merging conditions, and determining the node set with the largest average attribute similarity as the core node set.

[0128] According to an embodiment of the present disclosure, determining the initial attribute weights of the core node set in the initial community may include: constructing an objective function based on the number of nodes in the core node set, the size of the attribute set, and the values ​​of each core node on each attribute, the objective function representing the sum of the attribute-based weighted distances between the core nodes in the initial community; and determining the initial attribute weights by minimizing the sum of the weighted attribute distances of the core node set.

[0129] According to an embodiment of the present disclosure, determining the quality index of the initial community may include: determining the community density according to the node pairs, the neighbor sets of the nodes, and the number of neighbors of the initial community.

[0130] According to an embodiment of the present disclosure, determining the quality index of the initial community may further include: determining the community attribute entropy according to the attribute probability of the nodes in the initial community on the attributes.

[0131] According to an embodiment of the present disclosure, expanding the core node set of the initial community based on the initial attribute weights, community density and community attribute entropy to obtain a local community may include: constructing a community quality function based on the initial attribute weights, community density and community attribute entropy; based on the community quality function, adding neighbor nodes of the core nodes that have a gain in the community quality of the initial community to the initial community to obtain a local community.

[0132] According to an embodiment of the present disclosure, the device further includes: a storage module, used to store the local community in a graph database, and a display module, used to display the real-time local community.

[0133] According to an embodiment of the present disclosure, the device also includes: a feature screening module, which is used to respond to a transaction request and obtain the original transaction data generated by executing the transaction request; perform feature screening on the original transaction data to obtain transaction data; and store the transaction data in a transaction database.

[0134] According to an embodiment of the present disclosure, any multiple modules of the conversion module 810, the first determination module 820, the second determination module 830, the third determination module 840, the expansion module 850 and the identification module 860 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the conversion module 810, the first determination module 820, the second determination module 830, the third determination module 840, the expansion module 850 and the identification module 860 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in a proper combination of any of them. Alternatively, at least one of the conversion module 810, the first determination module 820, the second determination module 830, the third determination module 840, the expansion module 850 and the identification module 860 can be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function can be performed.

[0135] Fig. 9 A block diagram of an electronic device suitable for implementing the object method according to an embodiment of the present disclosure is schematically shown.

[0136] like Fig. 9As shown, the electronic device 900 according to an embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage part 908 to a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include an onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0137] In RAM 903, various programs and data required for the operation of electronic device 900 are stored. Processor 901, ROM 902 and RAM 903 are connected to each other via bus 904. Processor 901 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 902 and / or RAM 903. It should be noted that the program can also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0138] According to an embodiment of the present disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to the bus 904. The electronic device 900 may further include one or more of the following components connected to the input / output (I / O) interface 905: an input portion 906 including a keyboard, a mouse, etc.; an output portion 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 908 including a hard disk, etc.; and a communication portion 909 including a network interface card such as a LAN card, a modem, etc. The communication portion 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage portion 908 as needed.

[0139] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0140] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 902 and / or RAM 903 described above and / or one or more memories other than ROM 902 and RAM 903.

[0141] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the object recognition method provided by the embodiment of the present disclosure.

[0142] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 901. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0143] In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and downloaded and installed through the communication part 909, and / or installed from a removable medium 911. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0144] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0145] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0146] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0147] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0148] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. An object recognition method, characterized in that: The method comprises: Converting the transaction data into graph data, wherein the nodes of the graph data are transaction objects contained in the transaction data, and the edges of the nodes are transaction relationships contained in the transaction data; Based on the graph data, determining an initial community including a target node; Determining a core node set in the initial community and an initial attribute weight of the core node set; Determining a quality indicator of the initial community, the quality indicator includes a community density and a community attribute entropy, the community density is used to characterize the density of node pairs in the initial community, and the community attribute entropy is used to characterize the disorder of nodes in the initial community; Expanding the core node set of the initial community based on the initial attribute weight, community density and community attribute entropy to obtain a local community; The transaction object corresponding to the core node included in the local community is identified as the target object.

2. The method according to claim 1, characterized in that The determining of the core node set in the initial community includes: Determine a plurality of maximal cliques including the target node in the initial community; When there are N-1 overlapping nodes between two maximal clusters of size N, and the attribute similarity of the two overlapping nodes is greater than the average attribute similarity of the two maximal clusters, the two maximal clusters are merged into a new node set; The above operation is repeated until all nodes in the initial community do not meet the merging condition, and the node set with the largest average attribute similarity is determined as the core node set.

3. The method according to claim 1, characterized in that The determining of the initial attribute weights of the core node set in the initial community includes: Constructing an objective function according to the number of nodes in the core node set, the size of the attribute set, and the value of each core node on each attribute, wherein the objective function represents the sum of the weighted distances based on the attributes between the core nodes in the initial community; The initial attribute weight is determined by minimizing the sum of weighted attribute distances of the core node set.

4. The method according to claim 1, characterized in that: The step of determining the quality index of the initial community includes: The community density is determined according to the node pairs, the neighbor sets of the nodes, and the number of neighbors in the initial community.

5. The method according to claim 1, characterized in that The determining of the quality index of the initial community further includes: The community attribute entropy is determined according to the attribute probabilities of the nodes in the initial community.

6. The method according to claim 1, 4 or 5, characterized in that: The method of expanding the core node set of the initial community based on the initial attribute weight, community density and community attribute entropy to obtain a local community includes: Constructing a community quality function according to the initial weight of the attribute, community density, and community attribute entropy; Based on the community quality function, neighbor nodes of the core node that have a gain in community quality of the initial community are added to the initial community to obtain the local community.

7. The method according to claim 1, characterized in that The method further comprises: The local communities are stored in a graph database, and real-time local communities are displayed.

8. The method according to claim 1, characterized in that The method further comprises: In response to a transaction request, obtaining raw transaction data generated by executing the transaction request; Performing feature screening on the original transaction data to obtain the transaction data; The transaction data is stored in the transaction database.

9. An object recognition device, characterized in that: The device comprises: A conversion module, used to convert the transaction data into graph data, wherein the nodes of the graph data are transaction objects contained in the transaction data, and the edges of the nodes are transaction relationships contained in the transaction data; A first determination module, configured to determine an initial community including a target node based on the graph data; A second determination module is used to determine a core node set in the initial community and an initial attribute weight of the core node set; A third determination module is used to determine the quality index of the initial community, wherein the quality index includes community density and community attribute entropy, wherein the community density is used to characterize the density of node pairs in the initial community, and the community attribute entropy is used to characterize the disorder degree of nodes in the initial community; An expansion module, used to expand the core node set of the initial community based on the initial attribute weight, community density and community attribute entropy to obtain a local community; The identification module is used to identify the transaction object corresponding to the core node included in the local community as the target object.

10. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

12. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images