A risk monitoring method and device

Abstract

The application provides a risk monitoring method and device, relates to the technical field of artificial intelligence, and can be used in the financial field or other technical fields.The method comprises the following steps: inputting fusion data into a preset risk monitoring model, and taking an output result of the preset risk monitoring model as a risk monitoring result.The device executes the above method.The risk monitoring method and device provided in the embodiments of the application can not only reduce labor cost and improve efficiency, but also overcome the shortcoming of limited data dimension and comprehensively perform risk early warning according to business characteristics.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and particularly relates to a risk monitoring method and device. BACKGROUND

[0002] With the development of data technology, a large amount of data is generated, and at the same time, data risks are also accompanied. At present, the technology of automatically monitoring risk indicators is more and more widely used, and there are related risk early warning technologies, but there are deficiencies including: the data dimensions concerned are limited, and the risk early warning cannot be comprehensively performed according to the business characteristics; in addition, the risk early warning needs to rely on manpower, resulting in high labor cost and low efficiency. SUMMARY

[0003] In view of the problems in the prior art, the embodiments of the present application provide a risk monitoring method and device, which can at least partially solve the problems in the prior art.

[0004] In one aspect, the present application provides a risk monitoring method, comprising:

[0005] inputting fusion data into a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data fuses customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective information graph data and risk monitoring graph data corresponding to each information graph data;

[0006] outputting the output result of the preset risk monitoring model as a risk monitoring result.

[0007] The neural network model is obtained by combining BRNN and LSTM.

[0008] The risk monitoring result corresponds to the customer dimension, the product dimension and the enterprise dimension respectively; accordingly, after the step of outputting the output result of the preset risk monitoring model as a risk monitoring result, the risk monitoring method further comprises:

[0009] if at least one of the customer dimension risk monitoring result, the product dimension risk monitoring result and the enterprise dimension risk monitoring result exists risk, generating a risk early warning message corresponding to the at least one risk monitoring result existing risk.

[0010] The fusion data is obtained by:

[0011] constructing customer information graph data, product information graph data and enterprise information graph data respectively;

[0012] constructing customer risk monitoring graph data, product risk monitoring graph data and enterprise risk monitoring graph data according to the risk monitoring threshold of the respective data item corresponding to the customer information graph data, the product information graph data and the enterprise information graph data and the respective data item;

[0013] fusing the customer information graph data and the customer risk monitoring graph data, the product information graph data and the product risk monitoring graph data, and the enterprise information graph data and the enterprise risk monitoring graph data respectively to obtain the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data.

[0014] Wherein, after the step of obtaining the fusion data, the risk monitoring method further comprises:

[0015] calculating the node weight in the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data by using the degree centrality calculation method.

[0016] Wherein, after the step of calculating the node weight in the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data by using the degree centrality calculation method, the risk monitoring method further comprises:

[0017] assigning the node weight to the entity in the dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data, and performing vectorization representation on the dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data after assigning the node weight.

[0018] performing operation processing on the vectorized representation of the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data by using MainfoldE algorithm.

[0019] Wherein, the risk monitoring method further comprises:

[0020] inputting the vectorized representation of the fusion data after operation processing into the preset risk monitoring model, and continuing to execute the subsequent steps.

[0021] Wherein, obtaining the enterprise risk monitoring graph data comprises:

[0022] obtaining enterprise information and identifying the text content of the enterprise information, matching the text content with the words in the preset positive public opinion database and the preset negative public opinion database respectively;

[0023] determine the type of the text content according to the matching result; the type of the text content includes positive content or negative content;

[0024] traverse all the text content, and obtain the number of text content corresponding to the positive content or the negative content according to the type of the text content of each text content;

[0025] if the number of text content corresponding to the positive content is less than the number of text content corresponding to the negative content, it is determined that the enterprise information is negative enterprise information;

[0026] obtain enterprise interaction information corresponding to the negative enterprise information, and if the statistical value of at least one interaction index data in the enterprise interaction information is greater than a preset statistical data threshold, the negative enterprise information is taken as the enterprise risk monitoring graph data.

[0027] In one aspect, the present application provides a risk monitoring device, comprising:

[0028] an input unit configured to input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective corresponding information graph data and risk monitoring graph data corresponding to each information graph data;

[0029] a monitoring unit configured to take the output result of the preset risk monitoring model as a risk monitoring result.

[0030] In another aspect, the present application provides an electronic device, comprising a processor, a memory and a bus, wherein,

[0031] the processor and the memory complete mutual communication through the bus;

[0032] the memory stores program instructions executable by the processor, and the processor calling the program instructions can execute the following method:

[0033] input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data fuses customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective information graph data and risk monitoring graph data corresponding to each information graph data;

[0034] output results of the preset risk monitoring model as risk monitoring results.

[0035] Embodiments of the present application provide a kind of non-transient computer readable storage medium, comprising:

[0036] The non-transient computer readable storage medium stores computer instructions, and the computer instructions make the computer execute the following method:

[0037] input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data fuses customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective information graph data and risk monitoring graph data corresponding to each information graph data;

[0038] output results of the preset risk monitoring model as risk monitoring results.

[0039] The risk monitoring method and device provided by embodiments of the present application input fusion data to a preset risk monitoring model, and output results of the preset risk monitoring model as risk monitoring results, which not only can reduce labor cost and improve efficiency, but also can overcome the shortcoming of limited data dimension and comprehensively perform risk early warning according to business characteristics. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor. In the drawings:

[0041] Figure 1 is a flowchart of the risk monitoring method provided by an embodiment of the present application.

[0042] Figure 2 is a structural schematic diagram of a risk monitoring device provided by an embodiment of the present application.

[0043] Figure 3 is an entity structure schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0044] To make the objectives, technical solutions and advantages of embodiments of the present application clearer, further detailed descriptions will be made to the embodiments of the present application with reference to the drawings. Here, the illustrative embodiments of the present application and their descriptions are used to explain the present application, but are not used as limitations to the present application. It should be noted that, in the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other at will.

[0045] Figure 1 is a flowchart of a risk monitoring method provided by an embodiment of the present application, as shown in Figure 1 the risk monitoring method provided by the embodiment of the present application comprises:

[0046] Step S1: input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective corresponding information graph data and risk monitoring graph data corresponding to each information graph data.

[0047] Step S2: take the output result of the preset risk monitoring model as a risk monitoring result.

[0048] In the above step S1, the device inputs fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective corresponding information graph data and risk monitoring graph data corresponding to each information graph data. The device can be a computer device for executing the method, and it should be noted that the customer, product and enterprise related data involved in the embodiment of the present application are all authorized by users.

[0049] The customer information graph data is knowledge graph data constructed according to customer information, and the customer information can include:

[0050] Basic information of the client, including: the client's financial asset size at the end of last month, the client's identity network verification proof, and the client's credit rating, etc.

[0051] Basic information of the client's family members, including: the family members' financial asset size at the end of last month, the network verification proof, and the credit rating, etc.

[0052] Information of the client's holding enterprises, including: the number of holding enterprises, the industry to which the holding enterprises belong, the holding proportion, and the holding amount, etc.

[0053] The product information graph data is knowledge graph data constructed according to product information, and the product information can include:

[0054] Investment information of the underlying investment of the trust product, including: the proportion of underlying asset allocation, the name of the investment product, the product code, the product type, and the holding proportion, etc.

[0055] The running index data of the trust product during the term, including: the unit net value, the historical net value, and the interval yield, etc.

[0056] The disclosure report information returned by the trust company, including: the text content of the disclosure report, and the return date of the disclosure report, etc.

[0057] The enterprise information graph data is knowledge graph data constructed according to enterprise information, and the enterprise information can include enterprise news, etc.

[0058] The risk monitoring graph data corresponding to each information graph data can be understood as graph data for monitoring the risk of each information graph data.

[0059] The client risk monitoring graph data corresponding to the client information graph data is used for monitoring the risk of the client information graph data.

[0060] The product risk monitoring graph data corresponding to the product information graph data is used for monitoring the risk of the product information graph data.

[0061] The enterprise risk monitoring graph data corresponding to the enterprise information graph data is used for monitoring the risk of the enterprise information graph data.

[0062] The fusion sample data can be understood as the following data that can be used as the training of the neural network model and is pre-selected:

[0063] The client dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data, and the enterprise dimension comprehensive knowledge graph data.

[0064] Neural Networks (NN) is a complex network system formed by a large number of simple processing units (called neurons) widely interconnected, which reflects many basic characteristics of brain function, and is a highly complex nonlinear dynamic learning system. Neural networks have large-scale parallelism, distributed storage and processing, self-organization, self-adaptation and self-learning ability, and are particularly suitable for processing information processing problems that need to consider many factors and conditions, and are imprecise and fuzzy.

[0065] The training sample data adopts the customer dimension comprehensive knowledge graph data C1, the product dimension comprehensive knowledge graph data C2 and the enterprise dimension comprehensive knowledge graph data C3 constructed by stock data, and the training sample data includes risk data, risk edge data and no risk data. The training method is a conventional training method in the art, which will not be described here.

[0066] In the above step S2, the device takes the output result of the preset risk monitoring model as the risk monitoring result.

[0067] The risk monitoring method provided by the embodiment of the application inputs the fusion data into the preset risk monitoring model, and takes the output result of the preset risk monitoring model as the risk monitoring result, which not only can reduce the labor cost and improve the efficiency, but also can overcome the defect of limited data dimension, and comprehensively perform risk early warning according to the business characteristics.

[0068] Further, the neural network model is obtained by combining BRNN and LSTM. BRNN is a bidirectional recurrent neural network (Bidirectional RNN), which mainly solves the problem that the elements of the previous sequence cannot perceive the output of the later sequence.

[0069] Long short-term memory (Long short-term memory, LSTM) is a special RNN, which is mainly used to solve the problems of gradient disappearance and gradient explosion in long sequence training process.

[0070] Considering the association relationship between data and the timeliness of data, the model adopts a bidirectional recurrent neural network (BRNN), and considering that the storage of context information by the recurrent neural network is very limited and part of the information of the early time sequence will be lost, a long short-term memory network layer LSTM is added in the model.

[0071] Considering the information context association, timeliness and information storage, the neural network model has five layers, including an input layer Input Layer, a BRNN-Forward Layer, a BRNN-Backward Layer, an LSTM and an output layer Output Layer.

[0072] The risk monitoring method provided by the embodiment of the present application can overcome the shortcomings of a single model, thereby improving the operation efficiency of the model and the accuracy of the output result of the model.

[0073] Further, the risk monitoring result corresponds to the customer dimension, the product dimension and the enterprise dimension respectively; accordingly, after the step of taking the output result of the preset risk monitoring model as the risk monitoring result, the risk monitoring method further comprises:

[0074] If at least one of the customer dimension risk monitoring result, the product dimension risk monitoring result and the enterprise dimension risk monitoring result exists risk, a risk early warning message corresponding to the at least one risk monitoring result existing risk is generated. For example, if the product dimension risk monitoring result exists risk, a risk early warning message for the product dimension is generated.

[0075] The risk monitoring method provided by the embodiment of the present application can more targetedly and more timely obtain the risk monitoring result existing risk.

[0076] Further, the fusion data is obtained, comprising:

[0077] The customer information graph data, the product information graph data and the enterprise information graph data are respectively constructed; the customer information graph data is denoted as A1, and the construction is specifically explained as follows:

[0078] The basic information of the customer, the basic information data of the family members associated with the customer and the information of the enterprises held by the customer are taken as structured and semi-structured data, and through knowledge graph extraction technology, entities and attributes are extracted as nodes of the knowledge graph, the nodes include but are not limited to (customer name, customer age, customer credit rating, customer asset size, customer holding enterprise quantity and customer holding enterprise proportion, etc.), the relationship is taken as the edge of the knowledge graph, and the customer information graph data A1 is constructed. For example, <customer A, holding enterprise, holding enterprise quantity>, <customer A, sister, family member B>, <family member B, credit, credit rating> can constitute a knowledge graph triple.

[0079] The construction steps of the product information graph data A2 and the enterprise information graph data A3 can refer to the description of the construction steps of the customer information graph data A1, and will not be repeated here.

[0080] According to the risk monitoring threshold corresponding to the data items of the customer information graph data, the product information graph data and the enterprise information graph data respectively and the respective corresponding data items, the customer risk monitoring graph data, the product risk monitoring graph data and the enterprise risk monitoring graph data are respectively constructed; the customer risk monitoring graph data is denoted as B1, and the construction is specifically explained as follows:

[0081] Firstly, the data corresponding to the customer information graph data is obtained, which can be understood as data for risk monitoring, and can include:

[0082] The corresponding risk monitoring threshold values of asset fluctuation data and share fluctuation data are 30% and 10%, respectively.

[0083] The data corresponding to the customer information graph data can also be data unrelated to the risk monitoring threshold value, such as the customer identity recognition state result. For example, the networking verification result of the customer himself or the associated customer is in an abnormal state, or the customer credit result has adverse records, and then such results are included in the attention area.

[0084] If the customer's financial asset size decreases by more than 30% at the end of the month, the asset fluctuation data of this data item is included in the attention area.

[0085] If the holding ratio of the customer's holding enterprise decreases by more than 10%, the share fluctuation data of this data item is included in the attention area.

[0086] The customer risk monitoring graph data B1 is constructed, for example, including:

[0087] The <asset fluctuation, threshold value, 30%>, <threshold value, decrease by 30%, attention area>, <customer credit, adverse records, attention area> constitute a knowledge graph triple.

[0088] Further, the enterprise risk monitoring graph data is obtained, including:

[0089] The enterprise information is obtained, and the text content of the enterprise information is identified. The text content is matched with the words in the preset positive public opinion library and the preset negative public opinion library, respectively. The text content identification technology is a mature technology in the art and will not be repeated. The preset positive public opinion library refers to a pre-set positive word corpus, and the preset negative public opinion library refers to a pre-set negative word corpus.

[0090] The type to which the text content belongs is determined according to the matching result. The type to which the text content belongs includes positive content or negative content. If the matching result is a positive word corpus in the preset positive public opinion library, it means that the text content is positive content. If the matching result is a negative word corpus in the preset negative public opinion library, it means that the text content is negative content.

[0091] All text contents are traversed, and the number of text contents corresponding to the positive content or the negative content is obtained according to the type of the text content of each text content. For example:

[0092] There are a to e of 5 text contents, where a to b are positive content and c to e are negative content; then the number of text contents corresponding to positive content is 2 and the number of text contents corresponding to negative content is 3.

[0093] If the number of text contents corresponding to the positive content is less than the number of text contents corresponding to the negative content, then the enterprise information is determined to be negative enterprise information; referring to the above example, the enterprise information is negative enterprise information.

[0094] Obtain enterprise interaction information corresponding to the negative enterprise information. If the statistical value of at least one interaction indicator in the enterprise interaction information is greater than a preset statistical data threshold, then the negative enterprise information is used as the enterprise risk monitoring map data. Enterprise interaction information may include interaction indicator data such as the number of likes, comments, and reposts related to the enterprise information.

[0095] Taking the number of likes as an example, if the corresponding preset statistical data threshold is set to 5,000, then if the number of likes is greater than 5,000, the negative corporate information will be used as the corporate risk monitoring map data.

[0096] Specifically, negative corporate information can be placed in the watchlist and used as data content in the construction of the corporate risk monitoring map.

[0097] Alternatively, the following methods can be used:

[0098] Positive content is divided into four levels, A to D, according to the degree of positive impact from high to low. Negative content is divided into four levels, E to H, according to the degree of negative impact from low to high. Negative corporate information that meets the preset statistical data threshold is classified into level G or level H. Negative corporate information of level G or H is placed in the attention area, and then the attention area is used as the data content in the corporate risk monitoring map data.

[0099] The risk monitoring method provided in this invention can further improve the accuracy of risk monitoring results by reasonably acquiring enterprise risk monitoring map data.

[0100] Further, the product risk monitoring map data is obtained, including:

[0101] Investment products are categorized into equity, fixed-income, commodities and financial derivatives, and mixed categories. The proportions of each type of investment product and their volatility are obtained, and the corresponding risk monitoring thresholds can be set to 30% and 20%, respectively. The formula for calculating product volatility is a mature technology in this field and will not be elaborated further.

[0102] If the proportion of investment products exceeds 30%, then this data item will be included in the watchlist.

[0103] If the product operation fluctuation rate exceeds 10%, the data item is included in the attention area.

[0104] The data item corresponding to the product information graph data can also be data unrelated to the risk monitoring threshold, for example:

[0105] By pre-setting the enterprise website address, the relevant disclosure reports or financial statements are regularly crawled. Each financial indicator in the report or statement is identified.

[0106] A multivariate model is constructed, and Z-score method is used to predict the possibility of enterprise financial failure. The solvency, profitability and operation ability indicators of the enterprise in each period can be reflected by five variables, and the possibility of enterprise financial failure or bankruptcy can be comprehensively analyzed and predicted.

[0107] Among them, the solvency indicators X1=(operating funds / total assets)×100, X4=(total market value of common stock and preferred stock / total liability book value)×100.

[0108] The profitability of the enterprise X2=(retained earnings / total assets)×100, X3=(EBIT / total assets)×100.

[0109] The operation ability of the enterprise X5=sales revenue / total assets.

[0110] Z is the discriminant function value, Z=W1×X1+W2×X2+W3×X3+W4×X4+W5×X5, when the Z value is lower than the discriminant threshold, Z is put into the attention area. The discriminant threshold can be set according to the actual situation, which can be selected as 1.81.

[0111] W1-W5 are the weights corresponding to X1-X5 respectively, which can be set according to the actual situation, and can be selected as 0.012, 0.014, 0.033, 0.006 and 0.999 respectively.

[0112] By reasonably obtaining product risk monitoring graph data, the accuracy of the risk monitoring result can be further improved.

[0113] The construction steps of product risk monitoring graph data B2 and enterprise risk monitoring graph data B3 can refer to the above description of the construction of customer risk monitoring graph data B1, and will not be repeated here.

[0114] The customer information graph data and the customer risk monitoring graph data, the product information graph data and the product risk monitoring graph data, and the enterprise information graph data and the enterprise risk monitoring graph data are fused respectively to obtain customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data. The customer dimension comprehensive knowledge graph data is denoted as C1, and the construction is specifically as follows:

[0115] The entities of the customer information graph data A1 and the customer risk monitoring graph data B1 are matched, for example, the customer assets in the customer information graph data A1 and the asset fluctuation in the customer risk monitoring graph data B1 are directed to the same entity, and knowledge merging is performed by using an entity linking technology, so that the customer information graph data A1 and the customer risk monitoring graph data B1 are fused to obtain the customer dimension comprehensive knowledge graph data C1. The construction steps of the product dimension comprehensive knowledge graph data C2 and the enterprise dimension comprehensive knowledge graph data C3 can refer to the construction description of the customer dimension comprehensive knowledge graph data C1, and will not be repeated here.

[0116] The risk monitoring method provided by the embodiment of the application is helpful to accurately and comprehensively monitor the risk by comprehensively fusing the data into the customer dimension comprehensive knowledge graph data C1, the product dimension comprehensive knowledge graph data C2 and the enterprise dimension comprehensive knowledge graph data C3.

[0117] Further, after the step of obtaining the fused data, the risk monitoring method further includes:

[0118] The node weight in the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data is calculated by using a degree centrality calculation method.

[0119] Since each entity in the knowledge graph and the connection mode between the entities are different, including one-to-many, many-to-many and many-to-one modes, the importance of each node is also different. The degree of the node refers to the associated edge of the node, and the more the associated edges, the greater the weight of the node.

[0120] For example, the holding enterprise entity node, the customer entity in the customer information graph data A1, and the enterprise share entity and the enterprise share threshold entity in the customer risk monitoring graph data B1 are associated with the customer entity, and compared with the customer age, the holding enterprise entity node should be given a greater weight.

[0121] The degree centrality calculation method is a mature method in the art, and will not be repeated here.

[0122] The risk monitoring method provided by the embodiment of the application can accurately calculate node weights, so that the close degree of the association between entities is reflected, thereby helping to improve the accuracy of the output result of the preset risk monitoring model.

[0123] Further, after the step of calculating the node weights in the dimension comprehensive knowledge graph data of the customer, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data by using the degree centrality calculation method, the risk monitoring method further comprises:

[0124] The node weights are assigned to the entities in the dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data, and the dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data to which the node weights are assigned are vectorized; that is, the knowledge graph is vectorized, and the entities and the relationships in the knowledge graph need to be mapped into spatial vectors through different mapping matrices for representation. Taking the triple <customer A, holding enterprise, number of holding enterprises> as an example, the head entity "customer A" is trained into a vector h through a word2vec model, the relationship / attribute "holding enterprise" is a vector r, and the tail entity "number of holding enterprises" is a vector t, and the entity and the relationship vectors can be mapped into spatial vectors h T =M rh ×h and t T =M rh ×t respectively through mapping functions M.

[0125] The vectorized customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data are processed by using a MainfoldE algorithm. The MainfoldE algorithm is a mature algorithm in the field, and maps the relationship r into a hyper-sphere on a flow body, that is, h+r is a hyper-sphere with r as the center and the radius, instead of an accurate point close to h+r, which is more conducive to processing a knowledge graph with complex relationships.

[0126] The risk monitoring method provided by the embodiment of the application is more conducive to processing a knowledge graph with complex relationships.

[0127] Further, the risk monitoring method further comprises:

[0128] The vectorized fusion data after processing is input into the preset risk monitoring model, and the subsequent steps are continued. That is, the fusion data in the foregoing steps is replaced by the vectorized fusion data after processing in this step, and other descriptions are omitted.

[0129] It can be understood that the fusion data in the vectorized representation after the operation processing is taken as the input of the model, and the operation efficiency of the preset risk monitoring model can be improved.

[0130] The risk monitoring method provided by the embodiment of the present application can further improve the operation efficiency of the preset risk monitoring model.

[0131] It should be noted that the risk monitoring method provided by the embodiment of the present application can be used in the financial field, and can also be used in any technical field other than the financial field, and the application field of the risk monitoring method is not limited by the embodiment of the present application.

[0132] Figure 2 is a structural schematic diagram of a risk monitoring device provided by an embodiment of the present application, as Figure 2 indicated, the risk monitoring device provided by the embodiment of the present application comprises an input unit 201 and a monitoring unit 202, wherein:

[0133] The input unit 201 is configured to input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data comprise respective information graph data and risk monitoring graph data corresponding to each information graph data; and the monitoring unit 202 is configured to take the output result of the preset risk monitoring model as a risk monitoring result.

[0134] Specifically, the input unit 201 in the device is configured to input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data comprise respective information graph data and risk monitoring graph data corresponding to each information graph data; and the monitoring unit 202 is configured to take the output result of the preset risk monitoring model as a risk monitoring result.

[0135] The risk monitoring device provided by the embodiment of the present application inputs fusion data to a preset risk monitoring model, and takes the output result of the preset risk monitoring model as a risk monitoring result, which not only can reduce the labor cost and improve the efficiency, but also can overcome the defect of limited data dimension, and comprehensively perform risk early warning according to the business characteristics.

[0136] The embodiment of the risk monitoring device can be specifically used for executing the processing procedure of the method embodiments, and details are not described herein again. Please refer to the detailed description of the method embodiments.

[0137] Figure 3 An electronic device entity structure schematic diagram provided by the embodiment of the present application is shown in FIG. 1. Figure 3 As shown in FIG. 1, the electronic device includes a processor 301, a memory 302 and a bus 303.

[0138] The processor 301, the memory 302 and the bus 303 complete mutual communication.

[0139] The processor 301 is configured to invoke program instructions in the memory 302 to execute the method provided by the method embodiments, for example, including:

[0140] inputting fusion data into a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective information graph data and risk monitoring graph data corresponding to each information graph data.

[0141] The output result of the preset risk monitoring model is taken as a risk monitoring result.

[0142] The embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the method provided by the method embodiments, for example, including:

[0143] inputting fusion data into a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective information graph data and risk monitoring graph data corresponding to each information graph data.

[0144] The output result of the preset risk monitoring model is taken as a risk monitoring result.

[0145] The embodiment provides a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program causes the computer to execute the method provided in each method embodiment and the method comprises the following steps of:

[0146] input fusion data to a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fusion sample data; the fusion data is fused with customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data comprise respective information graph data and risk monitoring graph data corresponding to each information graph data;

[0147] output results of the preset risk monitoring model are taken as risk monitoring results.

[0148] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a computer program product in the form of being implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0149] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system) and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be realized by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device for realizing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus for realizing the functions specified in one or more flows and / or blocks.

[0150] These computer program instructions can also be stored in a computer readable storage medium which can guide the computer or other programmable data processing devices to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction apparatus, which realizes the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus for realizing the functions specified in one or more flows and / or blocks.

[0151] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 Figure 1

[0152] In the description of the present specification, the description of the terms "one embodiment", "one specific embodiment", "some embodiments", "for example", "exemplary", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0153] The specific embodiments described above are further intended to describe the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only for specific embodiments of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.​​

Claims

1. A risk monitoring method, characterized by, The method comprises the following steps: Obtaining fusion data comprises: Respectively constructing customer information graph data, product information graph data and enterprise information graph data; According to the risk monitoring threshold and the respective corresponding data items of the customer information graph data, the product information graph data and the enterprise information graph data, respectively, customer risk monitoring graph data, product risk monitoring graph data and enterprise risk monitoring graph data are constructed; Constructing product risk monitoring graph data comprises: Dividing investment products into equity, fixed income, commodities and financial derivatives, and mixed types according to types, obtaining the proportion of each type of investment product and the product running volatility, and respectively configuring the risk monitoring threshold corresponding to the proportion of investment products and the product running volatility; The data of the product information graph data corresponding to the data items exceeding the risk monitoring threshold is included in the attention area; Through the pre-set enterprise official website address, the related disclosure report or financial statements are regularly crawled, a multivariate model is constructed, and the Z score method is used to predict the possibility of enterprise financial failure; According to the attention area, the risk monitoring threshold and the possibility of enterprise financial failure, a knowledge graph triple is constructed to obtain product wind direction monitoring graph data; Constructing enterprise risk monitoring graph data comprises: Obtaining enterprise information and identifying the text content of the enterprise information, matching the text content with the words in the pre-set positive public opinion database and the pre-set negative public opinion database respectively; Determine the type of the text content according to the matching result; The type of the text content includes positive content or negative content; Traverse all text contents, and obtain the number of text contents corresponding to the positive content or the negative content according to the type of the text content; If the number of text contents corresponding to the positive content is less than the number of text contents corresponding to the negative content, the enterprise information is determined as negative enterprise information; Obtain the enterprise interaction information corresponding to the negative enterprise information, if at least one interaction index data in the enterprise interaction information is greater than the pre-set statistical data threshold, the negative enterprise information is taken as the enterprise risk monitoring graph data; Fusing the customer information graph data and the customer risk monitoring graph data, the product information graph data and the product risk monitoring graph data, and the enterprise information graph data and the enterprise risk monitoring graph data respectively to obtain fusion data; Input the fusion data into a pre-set risk monitoring model; The pre-set risk monitoring model is obtained by training a neural network model according to fusion sample data; The fusion data fuses customer dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; The customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective corresponding information graph data and risk monitoring graph data corresponding to each information graph data; The output result of the pre-set risk monitoring model is taken as the risk monitoring result; The neural network model is obtained by combining a BRNN and an LSTM, and the neural network model comprises an input layer Input Layer, a BRNN-Forward Layer, a BRNN-Backward Layer, an LSTM, and an output layer Output Laye.

2. The risk monitoring method according to claim 1, characterized in that, The risk monitoring results correspond to the customer dimension, the product dimension and the enterprise dimension respectively; accordingly, after the step of taking the output result of the preset risk monitoring model as the risk monitoring result, the risk monitoring method further comprises: If at least one of the customer dimension risk monitoring result, the product dimension risk monitoring result and the enterprise dimension risk monitoring result exists risk, a risk early warning message corresponding to the at least one risk monitoring result existing risk is generated.

3. The risk monitoring method according to claim 1, characterized in that, After the step of obtaining the fusion data, the risk monitoring method further comprises: The node weights in the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data are calculated by using a degree centrality calculation method.

4. The risk monitoring method according to claim 3, characterized in that, After the step of calculating the node weights in the customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data by using the degree centrality calculation method, the risk monitoring method further comprises: The node weights are assigned to the entities in the dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data, and the dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data to which the node weights are assigned are represented in vectors; The vectorized customer dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data are processed by using a MainfoldE algorithm.

5. The risk monitoring method according to claim 4, characterized in that, The risk monitoring method further comprises: The vectorized fusion data processed is input to the preset risk monitoring model, and subsequent steps are continued to be executed.

6. A risk monitoring apparatus, characterized by, Comprises: The input unit is configured to obtain fusion data, comprising: Customer information graph data, product information graph data and enterprise information graph data are constructed respectively; Customer risk monitoring graph data, product risk monitoring graph data and enterprise risk monitoring graph data are constructed respectively according to the risk monitoring thresholds of the data items corresponding to the customer information graph data, the product information graph data and the enterprise information graph data and the respective corresponding data items; The product risk monitoring graph data is constructed, comprising: The investment products are divided into equity, fixed income, commodities and financial derivatives, and mixed types according to types, the proportions of various types of investment products and product running volatility rates are obtained, and the risk monitoring thresholds corresponding to the proportions of investment products and product running volatility rates are configured respectively; the product information graph data corresponding data items data exceeding the risk monitoring threshold are included in the attention area; By pre-setting the enterprise official website address, regularly crawling the relevant disclosure reports or financial statements, building a multivariate model, and using Z-score method to predict the possibility of enterprise financial failure; according to the concerned area, risk monitoring threshold and enterprise financial failure possibility, a knowledge graph triple is constructed to obtain product wind direction monitoring graph data; The enterprise risk monitoring graph data is constructed, including: Obtaining enterprise information and identifying the text content of the enterprise information, matching the text content with the words in the preset positive public opinion database and the preset negative public opinion database respectively; determining the type of the text content according to the matching result; the type of the text content includes positive content or negative content; traversing all text contents, obtaining the number of text contents corresponding to the positive content or the negative content according to the type of the text content; if the number of text contents corresponding to the positive content is less than the number of text contents corresponding to the negative content, the enterprise information is determined as negative enterprise information; obtaining enterprise interaction information corresponding to the negative enterprise information, if the statistical value of at least one interaction index data in the enterprise interaction information is greater than the preset statistical data threshold, the negative enterprise information is taken as the enterprise risk monitoring graph data; The client information graph data and the client risk monitoring graph data, the product information graph data and the product risk monitoring graph data, and the enterprise information graph data and the enterprise risk monitoring graph data are fused respectively to obtain fused data; Input the fused data into a preset risk monitoring model; the preset risk monitoring model is obtained by training a neural network model according to fused sample data; the fused data is fused with client dimension comprehensive knowledge graph data, product dimension comprehensive knowledge graph data and enterprise dimension comprehensive knowledge graph data; the client dimension comprehensive knowledge graph data, the product dimension comprehensive knowledge graph data and the enterprise dimension comprehensive knowledge graph data include respective corresponding information graph data and risk monitoring graph data corresponding to each information graph data; A monitoring unit is configured to take the output result of the preset risk monitoring model as a risk monitoring result. The neural network model is obtained by combining BRNN and LSTM, and the neural network model includes an input layer Input Layer, a BRNN-Forward Layer, a BRNN-Backward Layer, an LSTM, and an output layer Output Laye.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 5.

9. A computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions that, when executed by a computer, enable the computer to implement the steps of the method of any one of claims 1 to 5.

Images